ETTC 2015

Le programme préliminaire de la conférence ETTC 2015 est disponible sur le site ettc2015.org
L’Association Aéronautique et Astronautique de France (3AF) et la Société de l’Electricité, de l’Electronique et des Technologies de l’information et de la Communication (SEE) vous invitent à soumettre une communication ou à exposer à la

A propos

ETTC2015 wiIl provide the opportunity for scientists and engineers to report and discuss the latest developments in testing methods, especially in aeronautic and space domain.
ETTC 2015 fournira aux scientifiques et aux ingénieurs l’occasion de présenter et de discuter les derniers développements des méthodes d’essai, en particulier dans le monde aéronautique et spatial.

L’Association Aéronautique et Astronautique de France (3AF) et la Société de l’Electricité, de l’Electronique et des Technologies de l’information et de la Communication (SEE) organisent la nouvelle édition de la Conférence Européenne des Essais et Télémesures, ETTC 2015. Une attention particulière sera portée cette année sur  comment les technologies utilisées pour le « Big data » peuvent aider la communauté des essais. Une session spéciale sera organisée par l’ICTS (International Consortium for Telemetry Spectrum) et une par l’ETSC (European Telemetry Standardization Committee).

The 3AF and SEE societies organize the next ETTC2015 edition. This year, specific attention will be paid to: how “Big data” technology can help the tests community. A special session will be organised by ICTS (the International Consortium for Telemetry Spectrum) and one by ETSC (European Telemetry Standardization Committee).

Exposition / Exhibition

Les développements technologiques en cours et les matériels existants seront mis en valeur par l’exposition qui accompagne la conférence.
On-going technological developments and recent test equipments will be on display at the exhibition associated with the conference.

Secrétariat ETTC 2015 / ETTC 2015 Office

3AF – 10, avenue Edouard Belin – 31400 TOULOUSE – France
Tel: +33(0)5 62 17 52 80 – Fax: +33(0)5 62 17 52 81

Depuis 1985, ETTC est organisé conjointement par la 3AF et la SEE, en liaison avec l’Arbeitskreis Telemetrie EV en Allemagne et l’International Foundation for Telemetering aux Etats-Unis. ETTC est organisé, en France les années impaires en alternance avec le colloque ETC, en Allemagne, les années paires.
Since 1985, ETTC is jointly organised by the 3AF and SEE with Arbeitskreis Telemetrie EV in Germany and the International Foundation for Telemetering in the United State. ETTC is held in France each odd year, alternately with the German conference ETC the even year.
 

Sponsors et organisateurs

Documents

XLS
Creative Commons Aucune (Tous droits réservés) Aucune (Tous droits réservés)
OAI = oai:www.see.asso.fr:ETTC 2015:13616

Authors =
Keywords =
Abstract
ETTC 2015 Programme
 
 



Programme ETTC 2015

ETTC 2015 PROGRAMME ISSUE 1 PLENARY SESSION - N° 1 – A350 Flight test campaign – Patrick DU CHE - AIRBUS – France - N° 2 – nEUROn Flight test campaign – Sylvain COURTOIS - DASSAULT AVIATION - France - N° 3 – SNCF Railway Rolling Stock Test Centre and Test in railway domain –Franck BOURGETEAU – Daniel CHAVANCE - SNCF – France - N° 4 – Last news from Rosetta – Philippe GAUDON – CNES – France - N° 5 – NASA’s Optical Communications Program for 2015 and Beyond – Donald CORNWELL – NASA – USA TECHNICAL PROGRAMME ORAL PRESENTATIONS SESSION N° 1 – Transducers, measurement devices – AIM2 research European Programme. Session chairman: Fritz Boden – DLR - Germany - N°1 - Advanced In-flight Measurement Techniques – Fritz Boden – DLR – Germany. - N° 2 - Recalibration of a Stereoscopic Camera System for in-flight Wing Deformation Measurements - Tania Kirmse - DLR Göttingen – Germany. - N° 3 - In-flight wing deformation measurements by image correlation technique on A350 - Benjamin Mouchet and Vincent Colman - AIRBUS Operation SAS – France - N° 4 - Rotating Camera System for Propeller and Rotor Blade Deformation Measurements - Fritz Boden, Boleslaw Stasicki and Marek Szypula – DLR – Germany - N° 5 - Development of Fibre Optic Strain and Pressure Instruments for Flight Test on an Aerobatic Light Aircraft - Nicholas Lawson, Ricardo Goncalves Correia, Ralph Tatam, Stephen James and Jim Gautrey - Cranfield University – United Kingdom - N° 6 - Recent achievements in Doppler lidars for aircraft certification - Claudine Besson, Beatrice Augere, Agnès Dolfi-Bouteyre, William Renard and Guillaume Canat – ONERA – France - N° 7 - NURMSYS - New Upstream Rotating Measurement System for gas turbine exhaust gases analysis - Betrand Carré, Sylvain Loumé and David Lalanne - AKIRA Technologies – France.. - SESSION N°2 – Test data acquisition and recording Chairman Part 1: Christian Herbepin – AIRBUS HELICOPTERS - France - N° 1 - FALCON 5X1 Flight Test Instrumentation – Jean-Pierre Rouby – DASSAULT AVIATION – France - N° 2 - High speed development of a temperature remote acquisition system to reduce instrumentation heat sink in an aircraft engine - Jean-Christophe Combier - AIRBUS Operation SAS – France. - N° 3 – Combined position, attitude measurement with precise time distribution for observation payload - Emmanuel Sicsik-Pare, Gilles Boime and John Fischer – Spectracom – France – USA - N° 4 - Optimizing Bandwidth in an Ethernet Telemetry Stream using a UHF Uplink - Moises Gonzalez-Martin and Pedro Rubio-Alvarez - AIRBUS DEFENCE AND SPACE – Spain - N° 5 - Flexible Switching for Flight Test Networks – Diamuird Collins - Curtiss-Wright Defense Solutions, Avionics & Electronics – Ireland - N° 6 - Evolving embedded electronics testing in HIL simulation and largescale test cells through sub-ns synchronization systems via Time Sensitive Networks in Ethernet- Kurt Veggeberg and Olivier Daurelles -National Instruments - United States, France 1    ETTC 2015 PROGRAMME ISSUE 1 Chairman Part 2: Diarmuid CORRY – Curtiss-Wright Controls Avionic & Electronic - Ireland - N° 7 - PTPv1 vs PTPv2: Characteristics, differences and time synchronization performances - Guillermo Martinez - Airbus Military – Spain - N° 8 - User Programmable FPGA I/O for Real-Time Systems – Combining User Friendliness, Performance, and Flexibility – Yannick Hildenbrand, Andreas Himmler and Jürgen Klahold - dSPACE GmbH – Germany - N° 9 – Guaranteed end-to-end latency through Ethernet - Øyvind Holmeide and Markus Schmitz - OnTime Networks - Norway, United States - N° 10 - Lessons for Onboard Data Storage from the worlds of Electronic Data Processing and Airborne Video Exploitation - Malcolm Weir - Ampex Data Systems Corporation – United States. - N° 11 - Cabin Comfort Flight Test Installation - Joel Galibert, Aymeric Plo and Stephane Garay - AIRBUS Operation SAS – France - N° 12 - The research on wireless sensor network for the aerocraft measurement system - Juan Lu, Ying Wang and Bingtai Liu - Beijing Institute of Aerospace Systems Engineering,- China SESSION N° 3: Big data and test data processing and analysis Chairman: Guy Destarac – 3AF- France - N° 1 - How to Harness the Value of Iot / Fast Data / Big Data and Data Analytics Technologies for the Tests Community - Frédéric Linder and Stéphane Biguet – Oracle – France - N° 2 - BigData applications for Telemetry -Greg Adamski and Gilles Kbidy – L-3-Communications Telemetry-West – United States - N° 3 - Case Study: Proposal of Architecture for Big Data Adoption - Luiz E. G. Vasconcelos, André Y. Kusumoto, Nelson P. O. Leite and Cristina M. A. Lopes- IPEV/ITA, ITA – Brazil - N° 4 - How Big Data technology brings added-value and agility during a flight campaign? - Laurent Peltiers and Jean-Marc Prangère - AIRBUS operations – France - N° 5 - Big Analog Data - Extracting Business Value from Test & Telemetry Data - Otmar Foehner, Robert Lee and Olivier Daurelles - National Instruments - United States, United Kingdom, France - N° 6 - Improving Test Cell Efficiency by Monitoring Measurements - Gouby Aurélie - Snecma – France - N° 7 – Processing Ethernet Flight Test Data with Open Source Tools – Paul Ferrill – Avionics Test and Analysis Corporation – United States SESSION n° 4 – ICTS (International Consortium for Telemetry Spectrum) Chairman: Jean-Claude GHNASSIA – 3AF – France - N°1 - Welcome and Introduction by ICTS Chair J.-C. Ghnassia - N°2 - Regional Reports: RI : J-C. Ghnassia RII: G.Mayer RIII: M.Ryan by J-C.Ghnassia - N°3 - “World Radiocommunication Conference 2015 (WRC-15) – Agenda Items Relative to Telemetry" G. Mayer - N°4 - “C-band for Airbus telemetry : status and improvement” L. Falga  - N°5 - “Eurocopter’s Conversion to C-band” tbc  - N°6 – Threats to aeronautical telemetry in USA : update #10. S. Hoshar  - N°7 - Conclusion and Closure J.-C. Ghnassia 2    ETTC 2015 PROGRAMME ISSUE 1 SESSION n° 5 – Telemetry frequency (spectrum management), modulation, telemetry systems. Chairman: Gilles FREAUD – Airbus – France - N° 1 : A new design to ground TM/TC communications for spacecraft launch campaign at Guiana Space Centre - Nicolas Hugues and Michel Thomas - CNES , ZDS – France - N° 2 : The entry into service of C-band Telemetry at Airbus Test Centre: first result and way of improvement - Luc Falga - AIRBUS Operations – France - N° 3 – Combining a Reed-Salomon block code with a blind equalizer: synchronization and bit error rate performance - Alexandre Skrzypczak, Gregory Blanc and Tangi Le Bournault - Zodiac Data Systems – France - N° 4 – Limitation of the 2 antennas problem for aircraft telemetry by using a blind equalizer. - Alexandre Skrzypczak, Gregory Blanc and Tangi Le Bournault - Zodiac Data Systems – France - N° 5 - A Gaussianization-based performance enhancement approach for coded digital PCM/FM - Guojiang Xia, Xinglai Wang and Kun Lan - Beijing Institute of Astronautical Systems Engineering – China - N° 6 - Real time C Band Link Budget Model Calculation - Francisco-M-Fernandez – Airbus Space and Defence - Spain SESSION n° 6 – Space Telemetry Chairman: Jean-Luc ISSLER - CNES – France - N° 1 - Rosetta-Philae RF link, from separation to hibernation - Clément Dudal, Céline Loisel, Emmanuel Robert, Miguel Angel Fernandez, Yves Richard and Gwénaël Guillois - CNES, Syrlinks – France - N° 2 - JASON3, a story of TT&C interference handling - Céline Loisel and Gérard Zaouche – CNES – France - N° 3 - Wavelet and source coding on Ariane 5 telemetry data - Didier Schott - Airbus Defence & Space – France - N° 4 – Cubesat communication CCSDS hardware in S and X band - Issler Jean-Luc and Lafabrie Philippe – CNES – France - N° 5 - Implementation of a high throughput LDPC decoder in space-based TT&C – Wen Kuang, Nan Xie and Xianglu Li - Inst. Of electronic engineering, china academy engineering physics – China - N° 6 - The Implement of IP-Based Telemetry System of Launch Vehicle - Feng Tieshan, Lan Kun and Zhao Weijun - Beijing Institute of Astronautical Systems Engineering - China SESSION n° 7 – MDL (Meta data language group) Chairman: Lee H. ECCLES – Boeing - USA Programme tbc . - Target of the group. - Detail of the actual situation - Discussion 3    ETTC 2015 PROGRAMME ISSUE 1 SESSION n°8 – ETSC (European Telemetry Standardization Committee) Chairman: Gerhard MAYER – GMV Consulting – Germany Welcome & Introduction (G. Mayer) 1. Committee Reports • ETSC Subcommitees SC-1:Spectrum & Frequency Management ( J.-C. Ghnassia, S. de Penna) SC-2: Data Acquisition & Processing (W. Lange, Ch. Herpepin) SC-3:Data Recording & Storage (B. Bagó, P. Morel, S.W. Lyons ) SC-4: Networked Telemetry (E. Schulze, Ch. Eder) • Telemetering Standards Coordination Committee(TSCC) (D. Corry) • Consultative Committee for Space Data Systems (CCSDS) (R. Ritter) 2. Short Presentation & Discussion Briefing on the “Working Draft IRIG 106, Chapter 7” (B. Bagó ) 3. New Business Membership standings, coming elections, preparation for the ETC 2016 Conclusions & Adjourn POSTER PRESENTATIONS SESSION n° P1 – TEST METHODS - N° 1 - Computational results for flight test points distribution in the flight envelope and dynamic relocation - Lina Mallozzi, Alessandro D’argenio, Pierluigi De Paolis and Giuseppe Schiano - Dipartimento di Ingegneria Aerospaziale - Università degli Studi di Napoli “Federico II” – Italy - N° 2 - High-resolution electro-acoustic transducer for dielectric characterization of outer space materials - Lucie Galloy-Gimenez, Laurent Berquez, Fulbert Baudoin and Denis Payan - LAPLACE, CNES – France - N° 3 - Non-contacting Methods with Lidar for Spacecraft Separation Ranging - Shengzhe Chen, Hui Feng and Yuzhi Feng - Beijing Institute of Aerospace Systems Engineering – China - N° 4 - How do you go about achieving your video recorder? (Chapter 2) - Pierrick Lamour and Loic Mauhourat – TDM – France - N° 5- SpaceWireless: Time-synchronized & reliable wireless sensor networks for Spacecraft - Damon Parsy – Beanair – Germany SESSION n° P2 – TEST TOOLS AND SIMULATION - N° 1 - LTM scalable to the tests - Sylvain Derlieu - AIRBUS OPERATIONS SAS – France - N° 2 - Optimized Automatic Calibration Tool. Application for Flight Test Programs – Enrique Torello, Jose Manuel Baena, Lorenzo Miranda and Pilar Vicaria - AIRBUS DEFENCE AND SPACE – Spain - N° 3 - Means driven by tests - Jerome Sartolou and Bruno Chaduteau – Nexeya – France - N° 4 - Design and implementation of LAN-based real-time simulation system of high frequency communication -Rui Song, Daquan Li, Guangming Zhou and Guojiang Xia - Beijing Institute of Astronautical Systems Engineering - China 4    ETTC 2015 PROGRAMME ISSUE 1 5    SESSION n° P3 – PROPAGATION, JAMMING AND ASSOCIATED MITIGATION - N° 1 - Characterization of the unavailability due to rain of an X band radar used for range safety at Kourou Space Center - Frédéric Lacoste, Jérémie Trilles and Clément Baron – CNES – France - N° 2 - Channel capacity estimation of stacked circularly polarized antennas suitable for drone applications - Ioannis Petropoulos, Jacques Sombrin, Nicolas Delhote and Cyrille Menudier - SigmaLim Labex, University of Limoges – France - N° 3 - Pattern-reconfigurable antenna design for telemetry and wireless communication systems - Gaojian Kang, Daquan Li and Xinglai Wang - Beijing institute of astronautical systems engineering - China

Creative Commons Aucune (Tous droits réservés) Aucune (Tous droits réservés)
contenu protégé  Document accessible sous conditions - vous devez vous connecter ou vous enregistrer pour accéder à ou acquérir ce document.
- Accès libre pour les ayants-droit

OAI = oai:www.see.asso.fr:ETTC 2015:13668

Authors =
Keywords =
Abstract
This zip file contains all ETTC 2015 communications and the final programme



no preview

Archive: Proceedings ETTC 2015.zip Length Date Time Name --------- ---------- ----- ---- 80547 2015-05-28 19:29 0 Programme iss 1.pdf 0 2015-06-07 10:27 __MACOSX/ 177 2015-05-28 19:29 __MACOSX/._0 Programme iss 1.pdf 311474 2015-05-26 07:57 1-1.pdf 450542 2015-05-07 12:21 1-2.pdf 177 2015-05-07 12:21 __MACOSX/._1-2.pdf 26311 2015-05-28 17:31 1-3.pdf 539065 2015-05-26 07:56 1-4.pdf 731466 2015-05-09 14:15 1-5.pdf 907111 2015-05-11 21:57 1-6.pdf 1489593 2015-05-01 16:46 1-7.pdf 557202 2015-05-25 09:02 2-1.pdf 316958 2015-05-05 18:59 2-10.pdf 213089 2015-05-28 17:48 2-11.pdf 345205 2015-05-28 17:47 2-12.pdf 12352 2015-05-28 17:38 2-2.pdf 604815 2015-05-07 12:23 2-3.pdf 143661 2015-05-18 08:54 2-4.pdf 588149 2015-04-30 18:23 2-5.pdf 253267 2015-05-28 17:39 2-6.pdf 263010 2015-05-01 16:37 2-7.pdf 917660 2015-05-25 10:17 2-8.pdf 12296 2015-05-28 18:09 2-9.pdf 12653 2015-05-28 18:14 3-1.pdf 1062817 2015-05-01 16:54 3-2.pdf 506649 2015-06-01 14:23 3-3mod.pdf 459056 2015-05-11 21:53 3-4.pdf 12399 2015-05-28 18:18 3-5.pdf 595049 2015-05-09 10:03 3-6.pdf 12172 2015-05-28 18:20 3-7.pdf 1183723 2015-04-30 09:32 5-1.pdf 12895 2015-05-28 18:25 5-2.pdf 354529 2015-05-28 17:24 5-3.pdf 329996 2015-05-28 18:27 5-4.pdf 390969 2015-05-09 14:34 5-5.pdf 12479 2015-05-28 18:29 5-6.pdf 492982 2015-05-01 17:48 6-1.pdf 580340 2015-04-30 19:31 6-2.pdf 276464 2015-05-28 18:41 6-3.pdf 13566 2015-05-28 18:31 6-4.pdf 12183 2015-05-28 18:44 6-5.pdf 294044 2015-04-30 10:04 6-6.pdf 714526 2015-05-01 17:45 P1-1.pdf 13263 2015-05-28 19:10 P1-2.pdf 177 2015-05-28 19:10 __MACOSX/._P1-2.pdf 12912 2015-05-28 19:11 P1-3.pdf 33043 2015-05-01 17:34 P1-4.pdf 392029 2015-06-01 14:23 P1-5mod.pdf 177 2015-06-01 14:23 __MACOSX/._P1-5mod.pdf 284436 2015-05-08 09:10 P2-1.pdf 132126 2015-05-09 14:44 P2-2.pdf 132124 2015-04-30 09:36 P2-3.pdf 12391 2015-05-28 19:14 P2-4.pdf 13175 2015-05-28 19:17 P3-1.pdf 486231 2015-04-30 19:33 P3-2.pdf 390969 2015-04-30 09:43 P3-3.pdf --------- ------- 17996671 56 files

Creative Commons Aucune (Tous droits réservés) Aucune (Tous droits réservés)
contenu protégé  Document accessible sous conditions - vous devez vous connecter ou vous enregistrer pour accéder à ou acquérir ce document.
- Accès libre pour les ayants-droit

Advanced In-flight Measurement Techniques F. Boden DLR, Bunsenstraße 10, 37073 Göttingen, Germany, fritz.boden@dlr.de Abstract: Advanced optical measurement techniques are beneficial compared to classical sensor measurements in terms of non-intrusiveness, measurement time and areal measurement. Although those techniques (e.g. the Particle Image Velocimetry (PIV) or the Image Pattern Correlation Technique (IPCT)) are well established in wind-tunnel and laboratory applications, it is a very challenging task to make them easily applicable for flight testing. Therefore, about more or less nine years ago the first EC project “Advanced In-flight Measurement Techniques” (AIM) was launched within the EU FP6. Researchers and specialists proofed in general the feasibility of applying their non-intrusive measurement techniques to industrial flight testing. Later the follow up project AIM² (EU FP7 contract no. 266107) has been started in 2010 intended to improve the AIM techniques towards routinely application in flight testing. In the paper, the two AIM projects will be presented briefly and an overview of the applied measurement techniques will be given. Keywords: Flight Test, Instrumentation, Optical Measurement Techniques, AIM, PIV, IPCT, IRT, LIDAR, PSP, BOS, FBG 1. Introduction Making new advanced optical measurement techniques that have their roots in wind tunnel or laboratory research applicable to time and cost efficient flight testing seems to be a very long journey. Even if some of these non- intrusive measurement techniques are still standard methods in wind-tunnel testing it has to be demonstrated again, that they are able to deliver reliable measurement results under flight testing conditions. Although optical measurement techniques are beneficial compared to classical sensor measurements in terms of non- intrusiveness, measurement time and areal measurement, the latter techniques are often applied in flight testing for reasons of simplicity. Therefore, in 2006 the European project “Advanced In- flight Measurement Techniques” (AIM) was launched. Researchers and specialists proofed in general the feasibility of applying their non-intrusive measurement techniques to industrial flight testing. After the AIM project the follow up project AIM² has been started in 2010. It was running for 48 months and was intent to improve the AIM techniques towards routinely application in flight testing. Therefore, lots of challenges identified within AIM had to be addressed and basic application rules as well as tool boxes had to be developed. This paper will give you a brief overview on both AIM projects and shortly describe the optical measurement techniques such as BOS, FBG, IPCT, IRT, LIDAR, PIV and PSP. 2. The AIM - project By the 1st of November 2006 the European Specific Targeted Research Project (STReP) “AIM - Advanced In- Flight Measurement Techniques” was launched within the 6th European research framework programme FP6. Its duration was 42 months. The goal of the project was to show the applicability of highly sophisticated optical measurement techniques to industrial flight tests. To achieve this target, eleven partner organisations from aircraft industries, airport services and research organisations closely worked together within AIM. The project was split into seven work packages (WP) and further subdivided into several tasks: • WP0 – coordination, • WP1 – wing deformation studies, • WP2 – propeller deformation studies, • WP3 – helicopter studies, • WP4 – surface flow measurements, • WP5 – high lift flow structures, • WP6 – industrial flight testing. Figure 1: Structure of the AIM project ETTC 2015– European Test & Telemetry Conference In what follows, the main content of all these work packages is briefly presented. 2.1 WP0 – Coordination WP0 was the main work package for all project management aspects like the co-ordination of the work package activities, the management of any contractual, financial and administrative issues, as well as the communication with the EC. Furthermore it comprised the exploitation of results, the creation of a communication platform for the partners and an official website [1]. 2.2 WP1 – Wing Deformation Studies WP1 mainly contained the in-flight measurement of wing deformation by means of the Image Pattern Correlation Technique (IPCT). Digital image correlation methods have been further developed to apply them for wing deformation measurements. Flight tests have been conducted on a Fairchild Metro II [2] and on a Piaggio P.180 [3] [4]. 2.3 WP2 – Propeller Deformation Studies In WP2 propeller blade deformation measurements by means of IPCT have been realized on a Piaggio P.180 [5]. In addition, an assessment of the IPCT for propeller deformation measurements was performed by one aircraft manufacturer [6]. 2.4 WP3 – Helicopter Studies WP3 was the work package for all the helicopter measurements. IPCT has been applied to main rotor blade deformations on a Eurocopter EC 135 helicopter [7]. The blade tip vortices have been investigated by means of LIDAR [8], BOS and PIV [9] on an MBB Bo105. 2.5 WP4 – Surface Flow Measurements In WP4 PSP was applied to measure the surface pressure distribution on the pylon of the VFW 614 ATTAS [10]. In addition IRT was improved for in-flight temperature measurements [11]. 2.6 WP5 – High Lift Flow Structures The WP5 was the work package with the most challenging tasks, as it intended to use the wind tunnel measurement techniques PIV and BOS for the ground based measurement of wake vortices of a landing aircraft [12] and PIV for non-intrusive in-flight flow field measurements [13]. Within AIM, it was the first time a PIV flight test installation was flown. 2.7 WP6 – Industrial Flight Testing WP6 was intended to apply the promising techniques IPCT and IRT under real industrial boundary conditions. On the one hand, the IPCT has been applied to wing deformation measurements on an Airbus A380 [14]. Due to the enormous dimensions of the aircraft it was a quite challenging task. The IRT on the other hand was applied to engine exhaust temperature measurements on a Eurocopter EC225 Superpuma [11]. 3. The AIM² project By the 1st of October 2010 the EC funded project AIM² (Advanced In-flight Measurement Techniques 2) was launched as a continuation of the preceding project AIM. Whereas the first AIM project proved the principle feasibility of using modern optical wind tunnel measurement techniques for in-flight measurements, AIM² focused on developing reliable and easy to use dedicated measurement systems and on defining design and application rules for these new in-flight measurement techniques. The project was running for 48 months structured in progressive steps starting with basic studies on challenges discovered in the preceding project, leading to optimised measurement systems to be tested under research conditions and finally to be proven in an industrial environment. AIM² comprised four partners from aerospace industries, one SME, three research organisations and three universities with expertise in optical measurement techniques, flight testing and training. The project AIM² was structured into 6 main work packages (WP), which were: • WP1 – Management, • WP2 – Deformation Measurements on Wings and Control Surfaces, • WP3 – Deformation Measurements on Propeller Blades, • WP4 – Surface Flow Measurements, • WP5 – Flow Field Measurements, • WP6 – Tools and Demonstration. Figure 2: Structure of the AIM² project ETTC 2015– European Test & Telemetry Conference 3.1 WP1 – Management WP1 was the overall management work package to cope with all the topics relevant for co-ordinating the project, organising the general meetings and disseminate the gathered knowledge by creating a communication platform and a website [15]. Furthermore it was designated to take care on gender issues and financial statements. 3.2. WP2 – Deformation Measurements on Wings and Control Surfaces WP2 was the work package for the improvement of the IPCT and marker based optical deformation techniques for measuring wing and control surface deformations in flight. Flight tests have been performed on a Fairchild Metro II and an EVEKTOR VUT 100 Cobra. Furthermore data of ground vibration testing by means of a marker technique on an Airbus A340 have been evaluated. 3.3. WP3 – Deformation Measurements on Propeller Blades WP3 was intended for the improvement of the IPCT and marker based techniques for propeller deformation measurements. A rotating stereo camera was designed, built and successfully flight tested on the EVEKTOR VUT100 Cobra [16]. 3.4. WP4 – Surface Flow Measurements FBG and unsteady IRT were further developed within WP4. The FBG sensors were flight tested on a Scottish Aviation Bulldog [17] including several wind-tunnel and laboratory tests. IRT flight tests were performed on a PW- 6 glider [18]. 3.5. WP5 – Flow Field Measurements WP5 comprised further development to enhance the optical measurement techniques BOS and PIV for in- flight flow field measurements. A second in-flight PIV application was performed on a Dornier Do-228 aircraft [19]. Furthermore LIDAR was flight tested on a Piaggio P.180 in order to calibrate standard FTI for airspeed measurements [20]. 3.6. WP6 – Tools and Demonstration WP6 comprised the development of useful tools to perform optical measurements and the creation of an application matrix for “non-experienced” users to set up their own advanced in-flight measurements. The tools have been finally applied for the landing gear measurements on a Piaggio P.180 and wing deformation measurements on a PW-6 glider. To spread the gained knowledge within the flight testing community a dedicated AIM² Advanced Flight Testing Workshop took place in Rzeszów (PL) from 9th to 14th of September 2013. Furthermore a small handbook on the AIM² techniques has been published [21] providing basic information about the methods and their possible applications. 3. The AIM - techniques 3.1 BOS – Background Oriented Schlieren Method The Background Oriented Schlieren (BOS) method technique is an image based density measurement technique. It uses the deviation of light due to refractive index changes in density gradients [22] (e.g. in compressible flow regimes, non-uniform temperature fields, gas mixtures). A randomly patterned background is once recorded without density gradient (reference image) and later with the density gradient in the line of sight (measurement image). Cross correlation algorithms identify pattern shifts between both images and thus enable the visualisation of the density gradients, e.g. vortex core location, jet location (see Figure 3) or shock position. Figure 3: Example of BOS processing - cross correlation of reference image (a) and measurement image (b) with jet exhaust jet between camera and background results in the visualisation (c) of the exhaust jet 3.2 FBG – Fiber Bragg Gratings FBG comprises a periodic modulation of the refractive index of the core of an optical fibre by the application of an interference pattern. Light of a known spectrum is send through the fibre and partly reflected by the grating (Figure 4). If the shape of the grating changes, e.g. due to strain or temperature change, the reflected spectrum changes proportionally. Thus the strain on the fibre can be deduced from the change of the reflected spectrum. Within AIM² strain and pressure sensors based on FBG have been developed and flight tested [17], providing the advantages of an much easier and lighter installation of sensors compared to strain gauges or pressure taps. Figure 4: Basic principle of FBG methods 3.3 IPCT – Image Pattern Correlation Technique The IPCT is an optical shape and deformation measurement technique based on the correlation of images of the investigated object painted with an irregular dot pattern [23]. If two cameras in a stereoscopic arrangement are applied, direct 3D measurements of the objects shape, its movements and deformations can be ETTC 2015– European Test & Telemetry Conference performed (see Figure 5). Within AIM and AIM² the IPCT mainly was applied to wing deformation [2] [3] [4] and rotor [7] as well as propeller deformation measurements [5] [16]. Figure 5: Principle of IPCT 3.4 IRT – Infrared Thermography The Infrared Thermography (IRT) is based on the measurement of the infrared radiation from surfaces and allows a global determination and visualisation of the surface temperature distribution with high accuracy. In aerodynamic research (in wind tunnel and flight tests) the thermography is used for the investigations of the boundary layer. Due to the jump in the wall stress coefficient and therefore in the heat transfer coefficient at the laminar-turbulent transition it allows the detection and visualisation of the transition from laminar to turbulent flow as well as laminar separations [24] and in some cases also vortices. Figure 6: Example of IRT measurements on a glider (left - setup, right measurement result) 3.5 LIDAR – Light Detection and Ranging LIDAR is based on Doppler shift determination of a light wave obtained from a single frequency laser that is reflected on natural atmospheric aerosols (Figure 7). The frequency shift is proportional to the air velocity and is detected via an interferometer measuring the beat between the backscattered wave from aerosols and a reference wave from a local oscillator. The coherent mixing enables the recovery of the backscattered wave phase, containing the radial velocity information along the laser line of sight. If required, the true air speed in three axes can be derived from multi axis sensing. This can be performed using 3 beams or more or a scanning device. LIDAR is able to give the velocity with no in-flight calibration. It is primary information without bias and thus it can directly be used to calibrate e.g. FTI [20]. By scanning an area or volume also complex flow fields (e.g. vortices) can be analysed. Figure 7: Principle of LIDAR 3.6 PIV – Particle Image Velocimetry The particle image velocimetry (PIV) is an image based measurement technique for instantaneous flow velocity fields. Tracer particles in the measured flow are illuminated by two co-planar pulsed laser light sheets. The backscattered light from the particles is imaged by one or more cameras. The cross correlation of both particle images deliver a displacement vector field directly depicting the flow field topology. With the known time delay between the laser light pulses and the magnification of the recording system the velocity vector field and thus the velocity components can be measured. Figure 8 shows a sketch of the measurement setup for the AIM and AIM² in-flight PIV campaigns and an example result of the measurements. Figure 8: Sketch of the AIM and AIM² PIV setup (left) and an example result (right) 3.7 PSP – Pressure Sensitive Paint PSP is an optical pressure measurement technique based on the "oxygen quenching" called photochemical reaction. In the presence of oxygen molecules, luminescence intensity from excited dye molecules which are implemented in the pressure sensitive paint is influenced by energy transfer. As a result, the luminescence intensity and lifetime changes with oxygen concentration or air ETTC 2015– European Test & Telemetry Conference pressure. The change can be observed by using digital cameras. Within AIM the PSP was applied for in-flight pressure measurements on the pylon of an aircraft. Figure 9: Example in-flight PSP measurements - raw image of the PSP (right) and extracted pressure values (left) 5. Conclusion This paper has provided a brief overview on the contents of the past AIM and AIM² projects. It delivered a small preview to the achievements of the projects and should encourage the reader to take a deeper look into the presented developments of advanced in-flight measurement techniques, e.g. by reading the publications in the references. At the beginning of the first AIM project optical measurement techniques and their abilities have been more or less unknown in the flight test community. Within the AIM consortium the growing close cooperation between research organisations and aircraft industry led to the first demonstrations of the feasibility of the applicability of the optical methods PSP, PIV, IRT, LIDAR, BOS and IPCT for industrial flight testing. In the follow up project AIM² the techniques have been further developed in order to make them applicable to flight test much easier. Useful tools and an application guide have been created during the project, and in addition a new method – the FBG – was introduced. Furthermore the focus of the AIM² project was laying in the dissemination of the knowledge to the flight test community. Now, after the finalisation of the two projects, this development has to be continued. Although the optical methods become more and more established for in-flight measurements several further steps have to be done in order to make them routinely be applicable and reliable. Some new features of the used measurement techniques will still be valuable for a wide variety of experimental applications even long after the finalisation of the AIM² project. The results of the improvement of the measurement techniques have mainly been assessed by the industrial partners. The exploitation therefore was mainly done in an interaction between the industrial partners and the developing research organisations. To keep this fruitful collaboration alive, to gain more knowledge and fields of application the wonderful AIM and AIM² consortium has to be kept together e.g. in a kind of an “AIM community” growing in the future by more and more partners. 6. Acknowledgement The author acknowledge the valuable work performed from all partners within the fantastic AIM and AIM² consortia as well all supporters of the new optical measurement techniques. 6. References [1] http://aim.dlr.de [2] H. P. J. Veerman, H. Kannemans, H. W. Jentink: "Highly Accurate Aircraft In-Flight Wing Deformation Measurements Based on Image Correlation", In: Research Topics in Aerospace Advanced In-Flight Measurement Techniques, Springer Verlag, Heidelberg New York Dordrecht London, 2013. [3] T. Wolf, C. Lanari, A. Torres, F. Boden: "IPCT Ground Vibration Measurements on a Small Aircraft", In: Research Topics in Aerospace Advanced In-Flight Measurement Techniques, Springer Verlag, Heidelberg New York Dordrecht London, 2013. [4] C. Lanari, A. Torres, T. Weikert, F. Boden: "In-flight IPCT Wing Deformation Measurements on a SMALL Aircraft", In: Research Topics in Aerospace Advanced In-Flight Measurement Techniques, Springer Verlag, Heidelberg New York Dordrecht London, 2013. [5] C. Lanari, B. Stasicki, F. Boden, A. Torres: "Image Based Propeller Deformation Measurements on the Piaggio P.180", In: Research Topics in Aerospace Advanced In-Flight Measurement Techniques, Springer Verlag, Heidelberg New York Dordrecht London, 2013. [6] P. Ruzicka, J. Rydel, M. Josefik, F. Boden, C. Lanari: "Assessment of Propeller Deformation Measurement Techniques for Industrial Application", In: Research Topics in Aerospace Advanced In-Flight Measurement Techniques, Springer Verlag, Heidelberg New York Dordrecht London, 2013. [7] F. Boden, C. Maucher: “Blade Deformation Measurements with IPCT on an EC 135 Helicopter Rotor.” In: Research Topics in Aerospace Advanced In- Flight Measurement Techniques, Springer Verlag, Heidelberg New York Dordrecht London, 2013. [8] B. Augere, C. Besson, A. Dolfi, D. Fleury, D. Goular, M. Valla: “1.5µm LIDAR for Helicopter Blade Tip Vortex Detection.” In: Research Topics in Aerospace Advanced In-Flight Measurement Techniques, Springer Verlag, Heidelberg New York Dordrecht London, 2013. [9] K. Kindler, K. Mulleners, M. Raffel: “Towards In-flight Measurements of Helicopter Blade Tip Vortices.” In: Research Topics in Aerospace Advanced In-Flight Measurement Techniques, Springer Verlag, Heidelberg New York Dordrecht London, 2013. [10] Y. Egami, C. Klein, U. Henne, K. de Groot, J. B. Meyer, C.-P. Krückeberg, F. Boden: “In-flight Application of Pressure Sensitive Paint.” In: Research Topics in Aerospace Advanced In-Flight Measurement Techniques, Springer Verlag, Heidelberg New York Dordrecht London, 2013. [11] L. Girard: “Application of Infrared Technology to Helicopter Flight Testing.” In: Research Topics in Aerospace Advanced In-Flight Measurement ETTC 2015– European Test & Telemetry Conference Techniques, Springer Verlag, Heidelberg New York Dordrecht London, 2013. [12] C. Politz, R. Geisler, S. Ranasinghe: “Ground Based Large Scale Wake Vortex Investigations by Means of Particle Image Velocimetry: A Feasibility Study.” In: Research Topics in Aerospace Advanced In-Flight Measurement Techniques, Springer Verlag, Heidelberg New York Dordrecht London, 2013. [13] C. Politz, N. J. Lawson, R. Konrath, J. Agocs, A. Schröder: “Development of Particle Image Velocimetry for In-flight Flow Measurement.” In: Research Topics in Aerospace Advanced In-Flight Measurement Techniques, Springer Verlag, Heidelberg New York Dordrecht London, 2013. [14] F. Boden, H. Jentink, C. Petit: “IPCT Wing Deformation Measurements on a Large Transport Aircraft.” In: Research Topics in Aerospace Advanced In-Flight Measurement Techniques, Springer Verlag, Heidelberg New York Dordrecht London, 2013. [15] http://aim2.dlr.de [16] F. Boden, B. Stasicki, M. Szpula: “Rotating Camera System for Propeller and Rotor Blade Deformation Measurements.” European Test and Telemetry Conference 2015, Toulouse, France, 9th – 11th of June 2015. [17] N. Lawson, R. Goncalves Correia, R. Tatam, S. James, J. Gautrey: “Development of Fibre Optic Strain and Pressure Instruments for Flight Test on an Aerobatic Light Aircraft.” European Test and Telemetry Conference 2015, Toulouse, France, 9th – 11th of June 2015. [18] P. Rzucidło, G. Kopecki , A. Kucaba-Piętal, R. Smusz, M. Szewczyk, M. Szumski, K. de Groot: “Flight parameters measurement system for PW6 in flight boundary layer mapping”, 9th AIRTEC 2014 International congress, Frankfurt / Main, 28th – 30th of October 2014. [19] C. Politz, C. Roloff, F. Philipp, H. Ehlers, A. Schröder, R. Geisler: “Free flight boundary layer investigations by means of Particle Image Velocimetry.” 17th International Symposium on Application of Laser Techniques to Fluid Mechanics, Lisbon, Portugal, 07th -10th of July 2014. [20] C. Besson, B. Augere, A. Dolfi-Bouteyre, W. Renard, G. Canat: “Recent achievements in Doppler lidars for aircraft certification” European Test and Telemetry Conference 2015, Toulouse, France, 9th – 11th of June 2015. [21] Boden, F. (ed.): “AIM² Advanced Flight Testing Workshop - HANDBOOK of ADVANCED IN-FLIGHT MEASUREMENT TECHNIQUES”, BoD Books on Demand, Norderstedt, 2013. [22] T. Kirmse: “Background Oriented Schlieren (BOS).” In: AIM² Advanced Flight Testing Workshop - HANDBOOK of ADVANCED IN-FLIGHT MEASUREMENT TECHNIQUES BoD – Books on Demand, Norderstedt. [23] F. Boden, T. Kirmse, H. Jentink: “Image Pattern Correlation Technique (IPCT).” In: AIM² Advanced Flight Testing Workshop - HANDBOOK of ADVANCED IN-FLIGHT MEASUREMENT TECHNIQUES BoD – Books on Demand, Norderstedt. [24] K. de Groot: “Infrared Thermography (IRT).” In: AIM² Advanced Flight Testing Workshop - HANDBOOK of ADVANCED IN-FLIGHT MEASUREMENT TECHNIQUES BoD – Books on Demand, Norderstedt. 7. Glossary 3D : three dimensional AIM : Advance In-flight Measurement Techniques (EU project) AIM²: follow up project of AIM (see above) BOS : Background Oriented Schlieren EC : European Commission FBG : Fibre Bragg Grating FP6, FP7: 6th and 7th European Research Framework Programmes FTI : Flight Test Instrumentation IPCT : Image Pattern Correlation Technique IRT : Infrared Thermography LIDAR : Light Detection and Ranging PIV : Particle Image Velocimetry PSP : Pressure Sensitive Paint SME : Small and Medium Enterprise STReP : Specific Targeted Research Project WP : Work Package ETTC 2015– European Test & Telemetry Conference

Creative Commons Aucune (Tous droits réservés) Aucune (Tous droits réservés)
contenu protégé  Document accessible sous conditions - vous devez vous connecter ou vous enregistrer pour accéder à ou acquérir ce document.
- Accès libre pour les ayants-droit

Recalibration of a Stereoscopic Camera System for In-Flight Wing Deformation Measurements T. Kirmse DLR Göttingen, Institute of Aerodynamic and Flow Technology, Bunsenstr. 10, 37073 Göttingen Abstract: A decalibration of a stereoscopic camera system by slight changes in the camera position and alignment affect the accuracy of the results directly. The paper describes criteria to assess the decalibration level of a stereo camera system. An approach for a recalibration and its limits are demonstrated for a dynamic wing deformation measurement performed on an Evektor Cobra VUT100 aeroplane by means of the Image Pattern Correlation Technique (IPCT). Keywords: stereo photogrammetry, digital image correlation, wing deformation measurement 1. Introduction The applicability of wing deformation measurements techniques based on stereoscopic photogrammetry like the Image Pattern Correlation Technique (IPCT) has already been successfully demonstrated for in-flight applications [1,2]. IPCT combines the principles of stereoscopic photogrammetry with image correlation methods to determine corresponding areas in a stereo image pair and delivers the 3D wing surface as result. Especially at dynamic flight manoeuvres varying multidirectional loads and the vibration level can affect the camera installation and induce slight changes in the camera position and alignment. Often an appropriate support and camera fixation cannot avoid the decalibration completely due to other constraints of the flight test integration like available space and maximal weight. Especially for applications in large transport aircrafts the distance between the stereo cameras can amount 2 m and more to ensure a sufficient accuracy. With increasing camera base distance the effort for the installation of the stereo cameras on a common stiff support increases significantly. The camera movements induce a decalibration of the measurement system and affect the accuracy of the results directly. Thus criteria must be found to assess the decalibration level of a stereo camera system to decide whether a correction is necessary or not to obtain reliable results. The triangulation error as a measure for the quality of the results can be caused by different sources. Because IPCT is delivering field data of a complete surface, the distribution of the triangulation error can be analysed to infer possible error sources. The paper describes the influence of different error sources to the triangulation error distribution. An approach for a recalibration to correct camera movements and its limitations are demonstrated for a dynamic wing deformation measurement performed on an Evektor Cobra VUT100 airplane by means of IPCT. 2. Image Pattern Correlation Technique The stereo IPCT combines the principles of stereo photogrammetry with digital image correlation methods. The image correlation is used to determine point correspondences in the stereo views of the cameras. Therefore structures of randomly distributed dots with a diameter of about 2-3 pixel in the image sensor are applied to the surface to be measured. The calibration of the cameras and the triangulation of the point correspondences to obtain the 3D coordinates of the surface finally are conforming to the methods of the standard stereo photogrammetry. The determination of the point correspondences based on the images only and is independent of the calibration parameters. The measurement principle is also called Digital Image Correlation (DIC) in the literature [3], but for applications in the field of aeronautics it is called IPCT mostly. 2.1 Camera calibration The used camera model is based on a pinhole camera extended by parameters to include radial distortions. The calibration procedure based on the work of Tsai [4] and Zhang [5]. The internal camera parameters describe the optics of the camera. They include the camera constant f which is nearly the focal length of the camera lenses, the pixel width, the pixel aspect ratio, the shear factor s, the radial distortion parameters of first and second order κ1 and κ2 and finally the pixel coordinates of the principal point [u0,v0]. The external camera parameter describe the position and the alignment of the cameras as relation between the camera coordinates [xc yc zc]T and the world coordinates [xw yw zw]T by a rotation matrix R and a translation vector t  by the equation: t z y x R z y x w w w c c c  +           ⋅=           [1] ETTC 2015– European Test & Telemetry Conference The projection centre of the camera defines the origin of the camera coordinate system whose z-axis is perpendicular to the image plane. 2.2 Triangulation and triangulation error Applying the calibration for every pixel position of a camera image the corresponding line of sight can be determined by defining a directional vector starting at the common point of the projection centre of the specific camera. The intersection of the lines of sight of both cameras for corresponding points delivers the 3D position of the point to be measured. In reality the line of sights are skew lines with no intersection. The point location is then estimated at the position of the shortest distance of the skew lines. The triangulation error can be express as the real distance of the skewed lines of sight or as disparity between the detected pixel coordinate of a point and the position of the re-projection of its measured 3D-coordinate to the camera sensors. The triangulation error is a measure for the quality of the measurement. It can be caused by: 1) Calibration errors (limits of the camera model, detection of the calibration grid, accuracy of calibration target) 2) Errors of the point correspondences (accuracy of the cross correlation or marker detection algorithm) 3) Decalibration caused by camera movements 4) Refraction index changes along the line of sight The portion of the different error sources to the absolute value of the triangulation error cannot be assigned clearly. Nevertheless the distribution of the triangulation error over the complete evaluated surface can be used to assess the dominating error source. For a good calibration the triangulation errors should be below 0.5 pixels, which can be checked by an evaluation of a reference surface recorded during the calibration process. Errors of the point correspondences cause local peaks in the triangulation error field. Due to the image correlation the outliers are often located at the edge of the evaluation area or at positions where the correlation pattern is disturbed, e.g. by markers. A decalibration caused by a camera movement will introduce a continuous offset of the triangulation error of some pixels with a small variance. Its variation is mainly a function of the distance of the 3D position to the cameras and the distance of the pixel position to the centre of the camera sensors. Depending on the specific application and their source the influence of refractive index changes can cause local errors, e.g. when some line of sights pass a shock, or global errors, caused among others. if the deformed shape of a window is changing the optics of the imaging system and hence the internal calibration parameters of the cameras. 2.3 Recalibration of the external camera parameters If the decalibration of the stereoscopic camera system is caused just by a camera movement and the internal parameters are not affected, it can be recalibrated by a correction of the external calibration parameters. A rotation matrix Rrecal and translation vector trecal is searched for which the sum of the triangulation error of all corresponding points becomes a minimum. Therefor the correction terms cR and ct  are introduced with corigrecal RRR *= [2] and corigrecal ttt  += [3] Because the rotation matrix R is an orthogonal matrix defined by three parameters e.g. the Euler angles, 6 parameters has to be optimized for a single camera. For laboratory tests or tests on the ground in general there is the possibility to use reference markers in the background of a defined position in the world coordinate system to determine the absolute correction of the position and alignment of both cameras. For in-flight application it cannot be distinguished clearly if a movement of the pattern seen by a camera is caused by a movement of the camera itself, by a deformation of the observed object or a combination of both. Thus the external parameters of one camera can only be corrected with respect to the second camera. Therefore it is assumed that only one camera must be corrected leading to a reduction of the overall number of parameters to be optimized. Using only a number of well distributed corresponding points but without the knowledge of their 3D-position is not sufficient to solve the problem explicitly, because there is no information about the scale of the measured area included. Thus additional constraints must be set to get realistic results and to improve the convergence of the minimisation problem. At least 2 dedicated markers are used to fix the scale of the measurement object. Therefore the change of their distance with respect to each other is limited strictly, where the reference distance is taken from a measurement of valid calibration, ideally obtained from recordings taken during the calibration procedure. Further constraints implemented are the limitation of the change of the camera basis (distance of the projection centres) and a maximum angular change of the optical axes of the camera to be recalibrated. The computation time of the recalibration depends strongly on the number of corresponding points used as input. Because IPCT delivers often thousands of data points for a measured surface it is necessary to reduce the number. The used nodes are selected randomly, whereas a minimum distance between the single nodes ensures a uniform distribution. Additionally the triangulation error of a selected point must be within the limits of the mean triangulation error ± its standard deviation over the complete field of view. This requirement ensures that no outliers are selected which could be caused e.g. by an locally incorrect correlation result 3. IPCT measurements on Cobra VUT100 The IPCT was applied to measure the wing deformation of a VUT100 Cobra, a four-seated single engine motor aircraft of 10.2 m span manufactured by Evektor. The two ETTC 2015– European Test & Telemetry Conference high speed cameras of type AOS S-EM with a maximum frame rate of 500 fps at full resolution of 1280 x 1024 pixels were installed on a customized camera support behind the front seats of the aircraft (figure 1). C-mount adapters were used to fix the objective lenses of a nominal focal length of 35 mm to the cameras. The random dot pattern was designed by means of a digital mock-up (DMU) to obtain an optimal dot diameter of 2-3 pixels in the camera images for the complete field of view. Accounting the decrease of the viewing angle toward the wing tip the dots had to be elongated with increasing spanwise position. Additional checkerboard- like markers were applied as reference points and initial points for the image correlation. Figure 2 shows a sample stereo image pair recorded on ground. 3.1 Accuracy of the setup Based on the specific calibration parameter set the local accuracy can be calculated for a defined uncertainty of the point correspondences. The accuracy estimation assumes a correct calibration. The left plot of the figure 3 depicts the estimated local accuracy ey in vertical direction of the Cobra measurement based on a surface determined from a ground recording taken during the calibration procedure. An uncertainty of 0.2 pixel for the point correspondences was used for the accuracy estimation. Due to the small stereo angle the accuracy ez in spanwise direction has higher values and increases with increasing distance to the cameras from 0.8 mm to 4 mm at the wing tip. Figure 2: Stereo image pair recorded during the ground test The local triangulation error of the pixel coordinate is shown in the right plot of figure 3. For the most part of the measurement area its level is below 0.5 pixels, which is a typical range for a good calibration. The mean triangulation error amounts 0.22 pixels over the complete surface with a standard deviation of 0.3 pixels. The structures seen in the contour plot could be caused by a deviation between the camera model and the properties of the real cameras. Figure 3: Accuracy estimation ey (left) and local triangulation error of the wing surface recorded during the calibration 3.2 Ground test results A deformation measurement was performed on ground to prove the accuracy of the IPCT results by a comparison with simple ruler measurements for some dedicated points. Figure 4 shows a picture of the setup. Three rulers were fixed on the bottom side of the wing along the main spar and an fourth ruler was fixed at the rear auxiliary spar at the wing tip. Different loads were attained by different tank levels (empty, half filled, full) and the application of an additional weight of 40 kg near the wing tip for the maximal load case. Figure 4: Setup of the wing deformation measurement on ground The load case of an empty tank was used as reference for zero deformation. The IPCT delivers the surface of the wing as direct result. For a comparison with the ruler measurements by Evektor the surface coordinates of the IPCT results were extracted along the main spar for every Figure 1: IPCT cameras installed in the cabin of the Cobra VUT100 ETTC 2015– European Test & Telemetry Conference load. The y-coordinates of the reference were subtracted from the values of the further load cases to obtain the deformation in y-direction. Figure 5 depicts the deformation versus the spanwise position. The IPCT results agree very well with the ruler measurements. Figure 5: Wing deformation with respect to empty tank load case along main spar Additionally the deformation extracted from the IPCT data of the ground recording taken during the calibration procedure was evaluated in the same way. Its results are plotted as black dash-dotted line in figure 5 and differ from the other measurement significantly. A positive deformation is detected meaning an upward wing movement. During the calibration the tank level was between half-filled and full. Even a strong gust could not explain an upward bend of this level. But there are several hours between the recordings of the ground test and the calibration images and meanwhile the aircraft was moved. The comparison of the triangulation error extracted at the main spar position in figure 6 indicates a decalibration of the camera system. The shape of the triangulation error along the main spar for the ground test cases agrees basically with the shape of surface recorded during the calibration but there is an offset of about 1 pixel. This is also confirmed by the mean triangulation errors and its standard deviation of the complete measured surface listed in table 1. There is a large offset of the mean triangulation errors whereas the change of its standard deviation compared to the calibration surface is marginal. Table 1: Triangulation errors of the ground test Triangulation error calib empty half full full full + 40 kg Mean value 0.22 1.22 1.41 1.27 1.40 Standard deviation 0.3 0.34 0.34 0.32 0.37 The deformation values are a relative measure with respect to a specific reference state. There was no further remarkable decalibration between the recording of the empty tank reference and the other wing load cases of the ground test. But using a reference of a different decalibration level leads to errors of the relative measure as well, because a change of the camera positions and alignment will cause a change of the absolute frame of reference. This must be taken into account for the evaluation of flight test points. Here the time lag between ground reference recording and in-flight measurement is high with a lot of alternations of the load the camera installation has to withstand. Figure 6: Triangulation error of measured surface coordinates along main spar 3.3 Flight test results The IPCT measurements were conducted at several static and dynamic flight conditions. For each flight test point 720 image pairs were recorded at a frame rate of 120 Hz. Here only the results of a parabolic flight are presented with a variation of the load factor between -0.5g and 2g over the measurement sequence. The triangulation error of the IPCT results is very high and varies strongly from 16 pixels to 6 pixels. Figure 7a shows the development of the load factor Nz. The time series of the triangulation error and the y-coordinate of a selected point near the wing tip triangulated with the original calibration parameters of the setup is shown in figure 7b. There is a clear link between the triangulation error and the y-position observable within the first 250 frames. Additionally the time series in figure 7c shows the mean value and the standard deviation of the triangulation error for the complete surfaces evaluated by IPCT. The curve of the mean triangulation error corresponds to the curve of the single point in figure 7b apart from an outlier at frame 121 for the single point which is levelled off by the averaging. The peaks of the standard deviation curve at some lower frame numbers indicate a higher number of outliers in these surface results. Nevertheless the standard deviation stays below 0.5 pixels for most frames suggesting that the triangulation error is reasoned mainly by a decalibration of the cameras. With a triangulation error of 6 pixels and more the decalibration cannot be neglected anymore. 4. Recalibration of the flight test data The recalibration procedure described in chapter 2.3 was applied to the measurement sequence of the parabolic flight. The external camera parameters of the left camera were corrected by a minimisation of the triangulation errors. In this case the distance of the skew lines of sight of the corresponding points was used as measure for the ETTC 2015– European Test & Telemetry Conference triangulation error. The change of the distance between the cameras ΔB was limited to 18 mm and the change of the view angle was limited to 0.5°. Furthermore the allowable change of the marker distance between four markers close to the wing root was restricted to 0.35 mm. Their positions measured during the calibration delivered the reference distances of correct calibration parameters. To reduce the computation time, 130 well distributed nodes were selected from the thousands of point correspondences. The parallelised recalibration process took 1.32 CPU hours on Intel Core i7-3770 processor for the complete sequence of 720 frames. Figure 7: Time series of parabolic flight Figure 8 shows the results of the recalibration by means of the time series of several measures. The plot of the residuum of the cost function in figure 8c demonstrates that the minimisation was successful for the most frames. This is approved by the mean triangulation error and its standard deviation depicted in figure 8d. For the most frames the mean triangulation error is now below 0.2 pixels and even the standard deviation could be decreased compared to the original results seen in figure 7c. The recalibration failed only for 7 frames, which can be clearly identified by the peaks in figure 8c. This is less than 1% of the data points. In the upper time series of figure 8a the absolute y- coordinate of a selected surface point is shown for the original calibration (red) and the recalibration (black). The difference between the absolute values amounts up to 7 mm. The recalibration cannot reproduce the real absolute coordinate system. Hence a reliable absolute deformation with respect to a ground reference of a valid original calibration cannot be derived. But relative measures can be determined by defining the deformation of a part of the wing in the field of view to be zero. Therefore a rigid body transformation (RBT) was applied to the surface results of the flight test which maps the 6 markers close to the wing root to their position at the ground reference. For the area of the six markers the deformation is defined to be zero. Figure 8b compares the y-coordinate after the application of the RBT. Now the difference is decreased, but nevertheless it amounts up to 2 mm. In figure 8a the maximal differences occur at the beginning of the time series, which is the part of the maximal triangulation error for the original calibration (compare figure 7c), whereas the differences for the mapped coordinate system are vanished in this area. Looking for the change of the camera position and alignment indicates a different kind of the camera movement. In figure 8e the change of the camera base distance ΔB and the angular change Δαcams of the optical axis and the x-axis of the camera sensor is shown. In the first part of the time series the calibration was corrected by a rotation of the camera around the optical axis expressed by the higher values of Δαcams for the sensor x-axis. For the frames of failed recalibration the values of ΔB reached the constraint of the minimisation process of 18 mm. Figure 8: Time series of recalibration results of the parabolic flight 5. Conclusion The triangulation error is a measure for the quality of the results of a stereoscopic photogrammetry measurement. The distribution of the triangulation error over the complete field of view can be used to assess, which kind of error sources contributing to the triangulation error mainly. The IPCT was applied to measure the wing deformation at a VUT100 Cobra aeroplane. The principle applicability of the method was proven by a ground test and the IPCT results agreed very well to the results of a standard measurement method. A slight decalibration of the cameras was indicated in the IPCT evaluation on the ground but caused no significant error in the results. Here the unloaded reference recordings were affected by the decalibration in the same order of magnitude as the loaded measurement recordings. This was not the case for the in-flight measurements, were the mean triangulation error varied even within a measurement sequence very strongly. The camera system can be recalibrated by correcting the external camera calibration parameters by a minimisation ETTC 2015– European Test & Telemetry Conference of the triangulation error. But if no fixed reference points are captured in the field of view, because of an overall deformation and moving background, alignment of the cameras can only be corrected with respect to each other. The absolute frame of reference cannot be restored, which must be taken into account for an analysis of the results. A recalibration of the external camera parameters of one camera was demonstrated on a wing deformation measurement at the Cobra VUT100 successfully for a parabolic flight manoeuvre. The mean triangulation error could be decreased even below the level of the original valid calibration. The application of the recalibrated camera parameters to the corresponding points changed the position of the surface points remarkably. The displacements were above the accuracy estimation assuming a correct calibration. The recalibration can only correct the position and alignment of the cameras with respect to each other but not their absolute position thus only relative deformations can be determined finally. Mapping the coordinates by defining a zone of zero deformation in the field of view reduces the differences between the results using the original calibration and the recalibration, but for the presented case it is not sufficient to reach the estimated accuracy, thus the recalibration is necessary to obtain reliable results. Since the determination of the point correspondences by image correlation is independent of the camera calibration the recalibration can be implemented easily in the evaluation process. 6. Acknowledgement The measurement was part of the project ‘Advanced In- Flight Measurement Techniques 2’ (AIM²) funded by the EC within the 7th Framework Programme for Research of the EU (Contract number 266107). The author wants to thank the team of EVEKTOR for the preparation and conduction of the flight test and provision of the aircraft data and the NLR for providing and operating the camera system and the delivery of the image raw data. 6. References [1] F. Boden, N. Lawson, H. Jentink, J, Kompenhans (ed.): "Advanced In-Flight Measurement Techniques ", Springer, 2013. [2] Meyer R., Kirmse T., Boden F.: “Optical In-Flight Wing Deformation Measurements with the Image Pattern Correlation Technique” In: New Results in Numerical and Experimental Fluid Mechanics IX Notes on Numerical Fluid Mechanics and Multidisciplinary Design, 124. pp 545-553,Springer, 2014 [3] Sutton M., Orteu J.J., Schreier H.:“Image Correlation for Shape, Motion and Deformation Measurements“, Springer 2009 [4] Tsai, R.Y.: "A Versatile Camera Calibration technique for High-Accuracy 3D Machine Vision Metrology Using Off-the-Shelf TV Cameras and Lenses", IEEE Journal of Robotics and Automation 3 (4), 1987. [5] Zhang Z.: "Flexible Camera Calibration By Viewing a Plane From Unknown Orientations", International Conference on Computer Vision (ICCV’99), 666-673, September 1999. ETTC 2015– European Test & Telemetry Conference

Creative Commons Aucune (Tous droits réservés) Aucune (Tous droits réservés)
contenu protégé  Document accessible sous conditions - vous devez vous connecter ou vous enregistrer pour accéder à ou acquérir ce document.
- Accès libre pour les ayants-droit

N° 3 - In-flight wing deformation measurements by image correlation technique on A350 -Benjamin Mouchet and Vincent Colman - AIRBUS Operation SAS – France The perfect knowledge of the shape of an aircraft wing is a key element for an aircraft manufacturer to validate its models. The classical technique used in flight tests is the photogrammetry based on cameras, flashes and reflective targets, directly set on the wing skin. Despite the good results it provides, the targets are aerodynamically intrusive and prevent from flexibility in the flight test campaign. An innovative in-flight wing deformation measurement technique, called Image Pattern Correlation Technique (IPCT) is a valuable alternative. The technology, developed by the DLR, is based on a stereoscopic method with stickers making specific patterns on the wing and a complex data post-processing. Partially applied on the A380, this technique was successfully applied to a larger scale on the A350- 900 during its certification campaign. The installation, method and results are presented 

Creative Commons Aucune (Tous droits réservés) Aucune (Tous droits réservés)
contenu protégé  Document accessible sous conditions - vous devez vous connecter ou vous enregistrer pour accéder à ou acquérir ce document.
- Accès libre pour les ayants-droit

Rotating Camera System for Propeller and Rotor Blade Deformation Measurements F: Boden, B. Stasicki, M. Szypuła DLR, Bunsenstraße 10, 37073 Göttingen, Germany, fritz.boden@dlr.de Abstract: Within the EU project AIM² a rotating stereoscopic camera system was designed, build and successfully flight tested in order to apply the non- intrusive Image Pattern Correlation Technique (IPCT) to 360° propeller deformation measurements. The complete system was affixed to the axis of the aircraft engine rotating together with the propeller at its full rotational speed. It enabled the direct measurement of the propeller blades shape as well as its local pitch angle under real operating conditions. Although the system was exposed to extreme vibration and centrifugal forces it delivered good images and demonstrated the applicability of the method to flight testing. In the paper, this highly sophisticated rotary camera system and the measurement technique IPCT is described. Some results of the first flight tests are given and the next steps for the future applications are introduced. Keywords: Advanced In-flight Measurement Techniques, IPCT, rotating camera, blade deformation 1. Introduction The blades deformation directly affects the efficiency of an aircraft propeller. Therefore, the measurement of this parameter is of high interest. A standard approach for measuring the bending and torsion of a propeller blade is the application of strain gauges at several single locations on the blades surface. Due to the size of the sensors as well as the required cabling, the number of measurement locations on the rotating propeller is very limited. In addition the application of the sensors on the blade can affect negatively its mechanical and aerodynamic parameters. Furthermore a direct measurement of the shape and location of the blade is not possible. The optical measurement method IPCT (Image Pattern Correlation Technique) [1] enables non-intrusive shape and deformation measurements of such structures. Some years ago, this method has been applied to propeller deformation [2] in flight and rotor blade deformation [3] on ground. Both measurements were performed with a non-rotating camera system in the fix frame observing the blade passing the cameras field of view. They demonstrated the feasibility of the application of IPCT for in-flight propeller and rotor deformation measurements. Other optical measurement techniques e.g. like 3D DIC [4] and Moiré techniques [5] have also successfully been applied to rotor measurements. But all these activities have been performed with cameras outside the rotating frame. Since the knowledge of the blades deformation and movement is important for the hole revolution it is important to observe the complete rotor disc. For small scales or on test benches [6] this can be done by a stationary camera system with sufficient resolution. In case of full-scale propeller deformation measurements in flight it is nearly impossible to install such a camera system in a proper way. Therefore within the EC funded project AIM² project the development of a rotating camera system was launched. In what follows the IPCT as well as the rotating camera system itself are presented briefly. The successful in-flight application of the system and some results of the image post-processing can be found in the latter part of this paper. 2. Measurement Technique and Required Installations 2.1 IPCT The stereoscopic Image Pattern Correlation Technique (IPCT) is an optical method based on digital image correlation (DIC) and 3D reconstruction by means of triangulation. At least two cameras are observing the patterned measurement object from slightly different viewing angles (see Figure 1). The IPCT processing software dewarps the camera images and identifies corresponding pattern regions in both images. To obtain the 3D surface and its orientation, the resulting camera coordinates are finally triangulated using a 3D camera calibration based on the recordings of a well-known calibration target. A comparison of the measured surfaces resulting from different load cases (e.g. non-rotating, ground idle and full thrust) finally delivers the deformation of the investigated propeller or rotor blade. Figure 1: Sketch of the IPCT processing ETTC 2015– European Test & Telemetry Conference In order to separate the deformation from the solid body movements of the measurement area occurring due to the deformation of the not observed propeller part and the change of the blade pitch angle, additional markers (e.g.checker board markers) are applied to the IPCT pattern. In principal obtained 3D surfaces are “stitched together” by using markers close to the hub. The remaining differences between the surfaces yield the local deformation. In addition, the markers in processing are also used for getting a first initial image displacement before the dot pattern is correlated. Furthermore those markers can be used to recalibrate the cameras if vibrations of the camera support lead to small misalignments. Usually, the measurement inaccuracy of IPCT is in the order of 0.2 pixels on the camera sensor. Figure 2 shows the estimation of the inaccuracy of the rotating camera according to [7] and for three different focal lengths (f = 8 mm, 12.5 mm and 16 mm). The obtained inaccuracy perpendicular to the blade surface is dz. The inaccuracy increases towards the blade tip and for smaller focal lengths. This is due to the decrease of the resolution of mm per pixel with increasing distance and decreasing focal length. The estimated inaccuracy of the rotating camera applied to the test (f = 8 mm) thus is in the range of 0.12 to 0.68 mm perpendicular to the propeller blade. Figure 2: Estimation of the inaccuracy dz perpendicular to the blade surface of the presented measurement setup for different focal lengths f Indeed, the given local inaccuracy is higher than that for conventional methods (e.g. strain gauges down to 0.01‰ to IPCT down to 1‰ of the measurement area), but the big advantage of IPCT is its non-intrusiveness and its spatial information (field measurement). Furthermore the shape of the investigated surface and its 3D position are measured directly and do not have to be derived from other parameters (e.g. such as from the strain and the location of the strain gauges by means of a structural model). 2.2 Rotating Camera System For the application of the 3D IPCT processing images of the propeller blade taken by at least two cameras in a stereoscopic arrangement are required. Therefore within AIM², DLR and Polish HARDsoft company developed a novel rotating camera system enabling the observation of one propeller blade for its whole revolution during operation. A sketch of this device is shown in Figure 3. It is mounted on the propeller hub (8) and consists of several coaxial stages (1 to 7) co-rotating with the propeller at its full speed. The camera stage contains two CMOS camera sensors (1, 2) watching the investigated blade in a stereoscopic arrangement. Each of the sensors has a resolution of 1,280 x 1,024 pixels. The next stage (3) contains the image acquisition board, a GPS module and a phase shifter circuit. The reflection sensor (11) on the hub (8), delivering one pulse per revolution, is connected to this phase shifter to obtain the propeller rpm. With this information, triggering of the camera exposure at any dedicated propeller phase angle is enabled. The phase shift can either be set to a constant value for recording the propeller at the same phase, i.e. the blade position for each revolution, independent of the propeller speed, or with a phase shift change for each revolution by a given increment, hence providing scans of a phase angle range (also for a complete 360° revolution). In the further stage (4) an embedded computer is implemented to control the complete system and store the images on the removable drive of the SSD type which is located in the stage (5). The self-sustained system is powered by four rechargeable LiFePO4 batteries (6) with a total voltage of 14.8 V and a capacity of 3,500 mAh, enabling an operation of approximately one hour with image recording at 45 image pairs per second. During the flight test operation, the system is exposed to severe vibrations (in our application up to 20 g in a range of 20 to 150 Hz) and significant centrifugal forces at 2,700 rpm. To avoid damage to the electronics due to these high loads, all printed circuit boards were firmly fixed in a rigid metal frame preventing their stretching. To control the system whilst the propeller is running, a WLAN module is included enabling a “remote desktop” connection with the cabin for setting the recording and camera parameters and taking a quick look on the acquired images. Figure 3: Sketch of the rotating camera system ETTC 2015– European Test & Telemetry Conference Figure 4: IPCT pattern with checker board markers (a - pattern design with progressive dot size, b - simulated view in digital mock-up, c - camera image of the painted blade) 3. Measurements Since the propeller deformation measurement task within the AIM² project has been performed in collaboration with the Czech aircraft manufacturer EVEKTOR and the Czech propeller manufacturer AVIA PROPELLER, the first measurement object for the rotating camera was the propeller of the Evektor VUT100 Cobra airplane. To apply the IPCT on the propeller, the investigated blade was painted with a dedicated IPCT pattern. For the reason of an optimal imaging on the camera sensors, the pattern has been designed with a progressing dot and marker size towards the blade tip (see Figure 4a). After the verification of the pattern design in a digital mock-up (see Figure 4b) a paint mask has been manufactured in order to spray the white dots and markers on the black primed propeller blade. Figure 4c shows an example image of the painted blade taken by one of the camera sensors. All dots and markers nearly have the same size providing an optimal exploitation of the sensor resolution. The final painted blade and the rotating camera system in the next step have been mounted on the propeller and the complete installation has been balanced before it was mounted on the airplane. After the installation on the airplane, the camera system had to be calibrated. This was achieved by placing a checker board plate as calibration target in the cameras field of view (see Figure 5). Figure 5: Calibration of the rotating camera system Figure 6: EVKETOR VUT100 Cobra during flight test with rotating camera From the images recorded during calibration, the IPCT software directly calculates the extrinsic (e.g. position, orientation) and intrinsic (e.g. focal length, distortions) parameters of the cameras. Furthermore the position and orientation of the calibration target in the first image pair defines the measurement coordinate system of the rotating camera. Once the system had been calibrated and was operating properly the first flight tests have been performed. In total three measurement flights and one ground run have been conducted. During the tests the camera system was operating autonomously with one hour continuous recording with a maximum frame rate of 45 image pairs per second (equivalent to one image per revolution). In Figure 6 the Cobra flying with the mounted rotating camera is shown. For the first tests, the camera had no aerodynamic cover. In order to optimize the aerodynamics a suitable spinner could be used in the future. Figure 7: Example image pairs recorded at different phase angles during revolution ETTC 2015– European Test & Telemetry Conference Figure 7 shows some example recordings of the rotating camera for different propeller phase angles. The blurred background gives an impression of the high rotational speed of around 2,700 rpm. The images provide enough contrast for the IPCT processing even with the massive change of background illumination. The pattern itself is depicted with sufficient sharpness. 4. Results Next to the recording the IPCT processing of the image data was performed by means of in-house developed software containing three major steps: calibration, marker detection and 3D surface calculation. In Figure 8 an example of the 3D surfaces obtained from the stereoscopic image pairs of the rotating camera is depicted. The X - co-ordinate corresponds to the blades span direction, whilst the Y - co-ordinate incidences with the chord of the blade at near the root section (in vicinity to the extracted profile Figure 8b). The overall span-wise length of the surface is 681 mm being in agreement with the dimensions of the observed area on the blade. The overall shape of the blade surface is well reconstructed. Only at the leading edge close to the root, where the pattern is too coarse for the strong curvature, the reconstruction is not working well. The markers are also clearly visible on the surface (e.g. Figure 8a) because at those positions no pattern is applied and thus the algorithm produces a local discontinuity. Figure 8b, c and d show blade profiles extracted from the surface in chord-wise direction (Y - direction) and for three different span-wise locations. They clearly show the change of the curvature and local pitch angle - strong curvature and zero pitch angle at the root, less curvature and lower pitch angle at the tip. Doing this chord-wise extraction for the same span-wise locations for different flight conditions and propeller settings, the change of the propeller pitch as well as the change of the blades twist can be directly measured. Figure 8: Example resulting surface (a - detail of the surface; b, c, d – profiles at different span-wise sections) Figure 9: Examples of chord-wise slices at X = 400 mm ( = Figure 8c ) extracted from surfaces of different measurement points and normalized to the reference shape at this location, Δθ is the change in pitch angle compared to the reference state Figure 9 shows some example curves extracted at X = 400 mm ( = position of Figure 8c) for different flight measurement points. For a better comparison all slices had been normalized to the reference state at stand- still. The change Δθ of the local pitch angle with respect to the reference state can be directly read out of the diagram by determining the slope of the curves. To obtain a bending line of the blade the extraction of data can be performed in span-wise direction for different load cases. Figure 10 shows such an extraction for a line at constant Y = -40 mm and normalized to the reference state at stand-still. The given value of ΔZ is the difference of the measured Z - coordinate during flight compared to the reference surface at stand-still. The maximum deflection occurred during the test is about 12 mm at the blade tip. As expected, the comparison of both – the pitch angle and the bending lines for the same flight conditions – shows the increase of the bending for higher pitch angle values. Surprisingly the highest bending value in Figure 10 does not occur for the highest pitch angle in Figure 9. Figure 10: Example span-wise slices at Y = -40 mm extracted from surfaces of different measurement points and normalized to the reference shape at this location ETTC 2015– European Test & Telemetry Conference The reason could be a flow separation at the blade tip for the highest pitch value like indicated by the decrease of the bending curves slope towards the blade tip too. 5. Conclusion The presented rotating camera system in combination with the measurement technique IPCT for the first time enabled the observation and direct measurement of a propeller blade’s behaviour in flight. Up to now such a detailed measurement was not possible and the real behaviour of the blade during flight had to be simulated or theoretically calculated. Strain gauges and accelerometers have been the only means to obtain the blades deformation but merely at a limited number of single locations and with a modification of the blades surface affecting the aerodynamics and the mechanical properties. For the IPCT measurements with the rotating camera the only modification on the airplane was the painting of one propeller blade with a dedicated IPCT pattern. The camera itself has been directly mounted on the propeller hub of the airplane. After the measurement campaign the camera system and the paint of the propeller blade were removed and the airplane was able to go back to normal service directly. The preparation of the test, especially the creation of the required dot pattern, was performed by using an in-house digital mock-up, thus avoiding costly pre-tests on the airplane. The Evektor VUT100 COBRA was only required for the final flight test and the effort for the flight test installation was very low. A classical installation of strain gauges for a similar surface measurement would have been nearly impossible due to the required number of sensors and the fitting on the propeller would have taken much longer. The presented rotating camera system enables the self- sustained recording of images of the blade at any phase angle and with a frame rate of maximum 45 double frames per second equivalent to one image pair per revolution of the Cobra propeller. Further development to increase this frame rate is presently performed by the authors. The IPCT processing of the recorded images delivers continuous 3D surfaces of the investigated blade. These surfaces can directly be used for the comparison of the real shape with shapes estimated from numerical design studies e.g. to validate such methods or to validate the performance of the blade design. In addition the obtained in-flight shapes can be used for numerical calculations of the flow with real geometries. Extractions of points or lines from the IPCT surfaces can be done in order to virtually obtain “local sensor data” to be processed in the standard approach like for strain gauges or accelerometers. By using another flange, the rotating camera can also be applied to measurements on other airplanes and with some modification also on large aircraft propellers. With the lessons learned on the small airplane, where the worst vibrations and centrifugal forces occur, similar devices can now be designed to carry out measurements on rotors of helicopters or wind turbines. 6. Acknowledgement The authors would like to thank U. Füllekrug (DLR) and T. Korbiel (AGH - University of Science and Technology, Cracow) for their assistance in carrying out the vibration test of the rotating camera and R. Ollech (DLR) and L. Dorn (DLR) for their cooperation in its mechanical construction and rotation test. Furthermore they would like to thank P. Růžička (EVEKTOR), Z. Tvrdik (AVIA Propeller) and K. Ludwikowski (HARDsoft) for their valuable work on the camera system and the flight tests. 6. References [1] Boden, F. (ed.): “AIM² Advanced Flight Testing Workshop - HANDBOOK of ADVANCED IN-FLIGHT MEASUREMENT TECHNIQUES”, BoD Books on Demand, Norderstedt, 2013. [2] Lanari, C., Stasicki, B., Boden, F., Torres, A.: “Image Based Propeller Deformation Measurements on the Piaggio P180.” In: Research Topics in Aerospace Advanced In-Flight Measurement Techniques, Springer Verlag, Heidelberg New York Dordrecht London, 2013. [3] Boden, F., Maucher, C.: “Blade Deformation Measurements with IPCT on an EC 135 Helicopter Rotor.” In: Research Topics in Aerospace Advanced In- Flight Measurement Techniques, Springer Verlag, Heidelberg New York Dordrecht London, 2013. [4] Sicard, J., Sirohi, J.: “Measurement of the deformation of an extremely flexible rotor blade using digital image correlation.” Measurement Science and Technology 24(6):065203, 2013. [5] Fleming, G. A., Gorton S. A.: “Measurement of Rotorcraft Blade Deformation Using Projection Moiré Interferometry.” Shock and Vibration, vol. 7, no. 3, pp. 149-165, 2000. doi:10.1155/2000/342875. [6] Sirohi J., Lawson M. S.: “Measurement of helicopter rotor blade deformation using digital image correlation.” Opt. Eng. 0001;51(4):043603-1-043603-8. doi:10.1117/1.OE.51.4.043603. [7] Krauss, K.: “Photogrammetry: Geometry from Images and Laser Scans”, Walter DeGruyter, 2007 7. Glossary 3D: three dimensional AIM : Advanced In-flight Measurement Techniques (EU project) AIM²: follow up project of AIM (see above) CMOS: Complementary Metal–Oxide–Semiconductor DC: Direct Current DIC: Digital Image Correlation f: focal length IPCT: Image Pattern Correlation Technique LiFePO4: Lithium (Li) Iron (Fe) Phosphate (PO4) rpm: Revolutions Per Minute SSD: Solid State Drive WLAN: Wireless Local Area Network X, Y, Z: Cartesian Coordinates Δ, d: Difference of the Parameter θ: Blade Pitch angle φ: Phase angle of the propeller ETTC 2015– European Test & Telemetry Conference

Creative Commons Aucune (Tous droits réservés) Aucune (Tous droits réservés)
contenu protégé  Document accessible sous conditions - vous devez vous connecter ou vous enregistrer pour accéder à ou acquérir ce document.
- Accès libre pour les ayants-droit

ETTC 2015– European Test & Telemetry Conference Development of fibre optic strain and pressure instruments for flight test on an aerobatic light aircraft N.J. Lawson1 , R. Correia 2 , S.W. James2 , , J.E. Gautrey1 and R.P. Tatam2 1: National Flying Laboratory Centre, Cranfield University, Cranfield, Beds. MK43 0AL, U.K. 2: Engineering Photonics, Cranfield University, Cranfield, Beds. MK43 0AL, U.K. Abstract: Fibre optic based measurement systems offer advantages for flight testing of aircraft. Through the EU FP7 program ‘Advanced In-Flight Measurement 2 (AIM2 )’, Cranfield University developed and flight tested Fibre Bragg grating strain and Fabry-Perot pressure fibre optic sensors on a certified Bulldog aerobatic light aircraft. Laboratory development demonstrated that the strain sensor had a resolution better than 0.2m/m and the pressure sensor to have a resolution better than 0.2 Pa over 400 Pa. Flight tests have proven the sensors performance at several kHz data rates with steady state and dynamic manoeuvres (-1g to +4g) including a spin. Keywords: Fibre Bragg grating, fibre optic Fabry-Perot sensor, flight test instrumentation 1. Introduction A recent research program entitled Advanced In-Flight Measurement 2 (AIM2 ), supported by the European Framework 7 (FP7) funding, has allowed a group of 10 partners to develop advanced in-flight measurement techniques based primarily on optical methods. The foundations of AIM2 were laid in the FP6 research program Advanced In-Flight Measurement (AIM) with significant outcomes including the application of pressure sensitive paint (PSP), LIDAR, background orientated Schlieren (BOS), particle image velocimetry (PIV) and image pattern correlation technique (IPCT) to flight test using a series of fixed wing and rotary wing flight test platforms [1]. The development of a range of different flight test measurement techniques has been ongoing since the 1950’s and has been well documented in publications such as AGARD and RTO Flight Test Instrumentation Series AGARDograph 160 (AG 160) [2,3]. In a given flight test programme, it is not uncommon for a comprehensive range of simultaneous parameters to be measured from a flying aircraft including, static and total pressure, temperature and wing shape, including twist. On a large transport aircraft, measurement points can be positioned significant distances (tens of metres) from the sensor power supplies or acquisition boxes and this can present electromagnetic compatibility (EMC) limitations and limit data quality and accuracy. More recent measurement techniques, in particular optical or fibre optic methods can overcome these limitations and provide improvements in temporal resolution and accuracy [4-8]. More specifically, fibre Bragg grating (FBG) sensors can be used to measure surface strain [4-7] and Extrinsic Fibre Fabry-Perot Interformeters (EFFPI) sensors can be used to measure static pressure [8]. The following paper describes the application of both the FBG and EFFPI fibre optic sensor systems to flight tests by using a modified Bulldog aerobatic light aircraft. Over a series of seven flight tests, the sensors were successfully demonstrated through a set of steady and dynamic flight test conditions (-1g to +4g), although for the pressure measurements, an offset was found between the optical method and the traditional Kulite sensor. At the time of writing, these discrepancies are still under investigation but are thought to be linked to a calibration-temperature effect or sensor contamination. 2. Development of Bulldog Flight Test Platform The objective of the Cranfield AIM2 research was to develop and flight test a fibre optic based surface strain and static pressure sensor which would be applicable to flight test instrumentation on both large scale and light aircraft. As Cranfield University already has extensive experience of development, design and fabrication of FBG sensors, this method was chosen for the strain measurement system. For the static pressure system, Cranfield University had also completed previous work with FBG’s for static pressure measurement [9]. However, this previous work did not demonstrate sufficient resolution for the project in this paper. Therefore it was decided a Fabry-Perot based sensor would be developed for the pressure sensing. As far as the authors are aware, this application of a Fabry-Perot pressure to flight test is seminal work. 2.1 Fibre Bragg Grating Surface Strain Sensors The principle of the FBG sensor is to measure the grating period of a fibre optic core which has been modified with a periodic refractive index grating. As the Bragg wavelength, generated by specific reflections inside the fibre core, is dependent on the period of the grating, as the fibre environment changes such as fibre strain and temperature, the Bragg wavelength changes. This wavelength can be monitored using an FBG interrogator and if using a suitable calibration, direct measurements of surface strain can be made. The major advantage of the FBG system is that a series of unique grating frequencies ETTC 2015– European Test & Telemetry Conference can be etched onto a single fibre and simultaneously monitored using a single FBG interrogator. FBG interrogators are available as commercial off-the-shelf items with sample rates of up to 20kHz. This multi-sensor FBG system can be extended to measuring a distribution of strain points on a surface such as a wing and by using a suitable model, simultaneous surface or wing shape is possible in-flight [10]. In this project, Cranfield University, using SMF-28 fibre, fabricated five FBGs at positions specified for the flight test (see section 2.3). The fibre was hydrogen loaded to increase photosensitivity and the FBGs were etched into the fibre core using a series of five different phase masks and a frequency-doubled Nd:YAG pulsed laser operating at 266nm to generate FBGs with five different Bragg wavelengths. Sections of the polyacrylate fibre buffer jacket were removed before fabrication. All five FBGs had a length of around 4mm and were not recoated. Following the fibre fabrication a laboratory strain calibration was completed by mounting a section of the fibre using Cynoacrylate superglue onto a test sample of equivalent material to the aircraft wing skin. This stage of development was required to check the package and mounting method for the fibre before mounting it for the flight test. A conventional 2mm length resistive foil strain gauge (RFSG) was also mounted adjacent to the FBG under test and the test sample was loaded up to 600. This calibration gave an FBG repeatability better than 0.29% (1.7) of full scale with high linearitywhich compared to 0.41% (2.4) for the conventional strain gauge system. A second identical fibre FBG was then fabricated and mounted onto the aircraft port wing and a further calibration completed before flight also using a set of five RFSGs mounted adjacent to each FBG point. The results from this pre-flight calibration showed similar repeatibility to the laboratory test. To simplify the aircraft modification, the wing mounted RFSGs were then disconnected and left glued onto the wing next to the FBGs. 2.2 Optical Fibre Fabry-Perot Pressure Sensors The requirement for the fibre optic pressure sensor was to measure steady and unsteady pressure on the aircraft from a selected point over a pressure range expected when flying the full g-range and speed range of the aircraft. Prior to development of the pressure sensor, in order to simply the certification process, a decision was made to modify an existing test plate behind the cockpit (see section 2.3). As significant variations of pressure can occur at different points on the aircraft under different flight conditions, a computational fluid dynamic (CFD) model and wind tunnel model were developed for the aircraft [11]. This work showed the expected pressure coefficient Cp range at the test point to be -0.05 < Cp < 0.05 which equated to less than 200 Pa of relative pressure change at the sensor point. Therefore a resolution of less than 2 Pa was specified for the sensor. The resolution requirements indicated an interferometric type sensor would be required for the fibre optic pressure system. Therefore an extrinsic fibre Fabry-Perot Interferometer (EFFPI) sensor was developed for the flight test. EFFPIs consist of an optical cavity formed by reflection at the distal fibre end mixing with light reflected from a reflective flexible diaphragm mounted onto the fibre end. As the pressure changes at the end of the fibre, the diaphragm deforms, changing the optical path length, and leading to an interferometric signal[8]. Therefore by monitoring the reflected signal, with a suitable calibration, the pressure at the diaphragm can be dynamically measured with high absolute and temporal resolution. A further advantage of the EFFPI method is with an careful cavity design, an identical interrogation unit, as used for the FBG system, can also be used to measure the EFFPI modulated signal. With reference to Figure 1 below, to fabricate the EFFPI sensor, a 125m fibre was mounted into a 2.43mm diameter ceramic fibre optic connector ferrule and a Mylar microphone sensing membrane mounted onto the polished top of the ferrule. To ensure an optimum cavity separation between the fibre tip and the membrane, before gluing the fibre, the fibre was moved up and down the ferrule sleeve whilst the output spectrum was monitored over a wavelength range of 27.45 nm. This range corresponded to a cavity length of 387 m. The cavity spacing was then set at a point which would produce the broad channeled spectrum intereference fringes required for the FBG interrogator. A 0.5mm diameter venting tube was also mounted onto the side of the fibre inside the ferrule to allow the sensor to measure reference pressure. The complete sensor was then glued into a 3.5mm Zirconia sleeve for mounting in the aircraft. It must be noted, with more specialist fabrication methods, the sensor size could be reduced considerably if the application required it. 2.43mm Mylar microphone sensing membrane fibre optic ferrule fibre core (10m) venting tube (0.5mm)fibre cladding (125m) Figure 1: Schematic of EFFPI ETTC 2015– European Test & Telemetry Conference To calibrate the EFFPI pressure sensor, the fibre was connected to a wavelength tuneable laser source (Santec HSL 2000) and two optical detectors (New Focus 2011- FC). Through an optical coupler, the fibre output signal was then interrogated using an NI PXI 5152 acquisition card and PC, as the pressure on the diaphram was varied though a reference pressure range up to 400 Pa. The resulting calibration showed a resolution better than 0.33% (1.3 Pa) of full scale over a range of 400 Pa. Further dynamic laboratory tests were also completed using a signal generator and loudspeaker arrangement as an input to the sensor. From this test, the dynamic sensor response appeared acceptable for frequencies up to 10 kHz which in this case was the limit of the loudspeaker setup. To further validate the EFFPI sensor during the flight test, a conventional Kulite XCQ-093 (2 psi range) unsteady pressure sensor was also mounted adjacent to the fibre sensor. This Kulite was statically calibrated using a similar input arrangement to the EFFPI and the results showed acceptable linearity and a resolution better than 0.15% (0.4 Pa) over a working scale of 250 Pa. 2.3 Bulldog Light Aircraft Test Bed To test the fibre optic strain and pressure sensors, a Scottish Aviation Bulldog aerobatic light aircraft was modified under Certification Standard 23 (CS-23) as a ‘minor modification’. This allowed the aircraft to be certified without requiring a flight test program following the certification and resulted in no limitation to the aircrafts operating speeds or g-load envelope (-3g to +6g). There was also no change to the aircraft’s centre of gravity and a minor increase in the aircrafts mass of around 9kg with a maximum take-off weight of 1066kg. The overall modification consisted of removing the existing floor plate behind the pilots seat and replacing it with a lightweight honeycomb 0.5 inch thick Teklam plate. A power supply box and a Smartscan Aero fibre optic interrogator box were then mounted onto the plate and the power supply box connected to the 28V aircraft power supply. The Smartscan box was connected to the power supply box and the FBG and EFFPI fibre optics. A further UEI data logging box was also mounted in the power supply box and connected to the Kulite pressure sensor. Further additions to the overall system included an attitude heading and reference system (AHRS), also connected to the UEI data logging cube, and a remote trigger to allow synchronisation of the Smartscan Aero box and the UEI data logger. Finally, additional hand held equipment including a PDA and Druck portable barometer were also carried in the onboard to monitor the cockpit reference pressure throughout the flight. Figure 2 shows the general arrangement of the set-up inside the aircraft where in summary the following equipment consisted of:  Bespoke power supply box (0.36 kW)  UEI data logging cube  Trigger box  SBG Systems SBG Systems IG-500A-G4A2P1-B AHRS (mounted adjacent to CoG)  Smartfibre Aero fibre optic interrogation box  Fibre optic 1 : five wing mounted FBGs  Fibre optic 2 : EFFPI pressure sensor (fuselage test plate)  XCQ-093 Kulite pressure sensor (fuselage test plate)  PDA connected to Druck DPI 740 barometer  on-board cockpit camera and mount Fabry-Perot (Fibre2)Kulite G1 G2 G3 G4 G5 FBGs (Fibre1) Smartscan Aero Box Fibre1 Trigger Button UEI Data Logging Cube Laptop LAN LAN Aircraft Power Supply Interface Fibre2 Kulite 28V DC cockpit reference pressure AHRS Figure 2: Schematic of Bulldog flight test instrumentation The FBG sensors were mounted using cyanoacrylate superglue at five points on the port wing spar centreline at 200mm, 400mm, 1200mm, 2200mm and 2600mm relative to mainplane station 26, which was positioned chordwise 350mm from the fuselage side. The outermost point was mounted inside a hypodermic sheath to provide temperature compensation to the other four points. The complete fibre length was then protected by covering with a length of 3M 425-50 speed tape. As indicated in the previous figure, the EFFPI and Kulite sensor were mounted on top of the fuselage on a 161mm diameter test plate, 35mm apart, with the circular plate positioned between the cockpit rear bulkhead and the front of the tailplane. The wire and fibre outputs of the two pressure sensors were then fed back along the inside of the fuselage on a common loom and connected to the Smartfibre and UEI interrogation units. A single pressure static tube, which terminated in the cockpit, was also connected to both sensors to provide a common cockpit reference pressure. This pressure was monitored by the PDA and Druck portable barometer throughout the flight. Also throughout the flights an external pilots view and an internal view of the main cockpit instruments was monitored using a 50Hz ActionCam mounted in the roof of the cockpit. ETTC 2015– European Test & Telemetry Conference 3. Flight Test Results The flight test program consisted of seven flights completed in June and July 2014. Flights 1 – 6 were used to troubleshoot the measurement systems with issues including data storage loss and Kulite earthing problems. On the initial flight, the tape covering the FBG fibre optic also became detached. A subsequent investigation found the incorrect tape had been fitted. On flight 7, however, the FBG, EFFPI and Kulite sensors all worked correctly and a series of steady state and dynamic manoeuvres were completed over a 50 minute flight which included:  Straight and level profile (67 knots IAS 8400 feet)  Straight and level profile (100 knots IAS 8400 feet)  3 turn left spin (8400 feet – 5500 feet)  Loop (5500 feet - +0.5g to +4g)  Stall turn (5500 feet +0g to 4g)  Slow left roll (5500 feet -1g to 1.5g)  Barrel roll (5500 feet 0.5g to +3g) In all cases a standard altimeter pressure setting of 1013 millibar was used and sea level conditions were ISA +10o C. In the following results, based on the calibrations, uncertainties in the data are FBG strain +/-1.7EFFPI and Kulite pressure +/-1.3 Pa and +/-0.4 Pa respectively. AHRS data was also recorded throughout the flights to allow additional assessment of the dynamic manoeuvres. The AHRS data is presented as inertial axes format (ax, ay, az), i.e. relative to the box axes. Pitch, roll and yaw were also analysed using Quaternion conversions as specified by the manufacturer. For all data, on start-up a common time stamp was achieved by aligning the different system time clocks using a Laptop-LAN connection. This alignment was better than 300ms for all the different systems. 3.1 Straight and Level Results In the first part of flight 7, straight and level results were used to analyse several pressure coefficient Cp characteristics at two different indicated airspeeds of 67 knots and 100 knots. From previous analysis [12], these two different airspeeds correspond to true airspeeds of 40.6m/s and 60.6m/s and angles of attack of around 8o and 4o respectively. Table 1: Summary of straight and level flight test results Speed (knots) Cp Kulite Cp % rms Kulite Cp EFFPI Cp % rms EFFPI 67 0.4131 0.30 0.1315 3.2 100 0.2949 0.29 0.1103 1.7 To ensure stable values of Cp during the tests, the aircraft was flown with reference to the Druck barometer output with samples taken when the Druck output was stable to within 10 Pa. This equates to less than 3 feet deviation in aircraft altitude during a given sampling period. Using these criteria, the best sample sets were analysed and the results are presented in Table 1. The results in Table 1 show a reduction in Cp with increasing airspeed or reducing angle of attack. This result is consistent with the wind tunnel and CFD analysis [11]. However, the magnitude of Cp between the Kulite and EFFPI results is significantly different with the EFFPI values of Cp around 3 times lower than the Kulite values. This difference equates to around 200 Pa between the sensors although the EFFPI values of Cp are much closer to the values reported in the CFD and wind tunnel data. This discrepancy is discussed further in section 4. 3.2 Spin and Aerodynamic Manoeuvres In the second part of flight 7, the dynamic manoeuvres, including the spin, were completed to assess the performance of the fibre optic and Kulite sensors over a wide part of the aircraft flight envelope. During the manoeuvres, AHRS data was also recorded for comparison with the strain and pressure data. Figure 3 shows the dynamic data recorded from all the sensors. An initial study of the data shows that although the general trends of the EFFPI data and Kulite unsteady pressure data are similar for all the manoeuvres, discrepancies still exist between the two sets of data which are greater than the levels of uncertainty and there is no constant offset between the sets of dynamic data as was observed during the steady state measurements. The Kulite data is smoother due to the low pass filters applied to the data during post processing. But the variations between the two sets of dynamic data, however, still varies from between zero to around 400 Pa following the exit from the spin for example. Figure 3: Sample of flight test data recorded during dynamic manoeuvres Figure 4 shows a more detailed output of the FBG strain unsteady data recorded during the spin manoeuvre where the spin rotations, correlating with the AHRS data, can be clearly seen as well as the recovery stage, where the ETTC 2015– European Test & Telemetry Conference aircraft load changes from a positive g-load to a negative g-load as a negative angle of attack is used to impart recovery. Further spectral analysis of this temporal data is shown in Figure 5 where the AHRS data predicts a spin frequency of 0.4Hz which compares to the FBG spin frequency of 0.39Hz. Here the uncertainty in the AHRS spectral data is +/-0.035 Hz and the Fabry-Perot and FBG spectral data +/-0.05Hz. Examination of the on-board camera images and fixed ground features during the spin also confirmed a frequency of 0.4 Hz. Figure 4: Strain and AHRS data recorded during the spin manoeuvre Figure 5: Spectral analysis of FBG and AHRS data during the spin Further AHRS and FBG data taken from the loop and slow roll manoeuvres is shown in Figure 6 and Figure 7 respectively. If we consider the loop, the initial pull up is visible where in this case, increases in negative strain correspond to increases in g-load on the wing. This change in g-load is confirmed in the AHRS data. The top of the loop is also visible where the g-load on the aircraft reduces to around 0.5g with an increase in load as the pull-out of the loop is completed. In terms of the slow roll data, the sequence of ‘entry’, ‘inversion’, ‘maintain’ and ‘recovery’ stages of the roll can also be seen with corresponding changes in the AHRS data. Similar trends were also found in the EFFPI data as can be seen in Figure 3, although the offset of around 200Pa between the values of the Kulite and EFFPI data is also visible. Figure 6: Strain and AHRS data for the loop manoeuvre Figure 7: Strain and AHRS data for the slow roll manoeuvre 4. Discussions The previous section has presented steady state straight and level data and dynamic FGB and EFFPI data at sample rates of up to 2kHz. During this wide range of flight conditions and g-loads, the fibre sensors, data logger and Smartscan fibre interrogation all behaved as expected and data was successfully recorded and post- processed for the entire duration of the flight. Checks of the equipment on the aircraft following this flight test series and frequent use of the aircraft for student flight training has also found the sensors to be stable and robust. There were, however a number of issues with the results which still need to be resolved. The main issue relates to the discrepancy between the EFFPI sensor and the Kulite pressure sensor. A post-flight calibration of the Kulite did not find any significant variation from the original factory calibration. At the time of writing, it is thought the discrepancy may be related to the temperature variation of either the EFFPI sensor or the Kulite sensor or a combination of both as the Kulite was operated around 10o C below its temperature compensated range. It may ETTC 2015– European Test & Telemetry Conference also be related to sensor contamination or an aerial and beacon light protuberance which were positioned adjacent to the pressure sensors. Further studies need to be completed to resolve this difference. 5. Conclusions This paper has presented flight test data from two fibre optic sensors which were fitted to a certified aerobatic category Bulldog light aircraft. The FBG system recorded successfully wing surface strain data during dynamic manoeuvres over a normal g-range of -1g to +4g with sufficient resolution to allow analysis of temporal features of spin. The EFFPI pressure sensor, fitted onto the top of the aircraft fuselage, also allowed analysis of both the dynamic and steady manoeuvres although an offset was found between a conventional Kulite pressure sensor fitted adjacent to the EFFPI sensor. Future work aims to isolate the sources of this error and then complete further flight test campaigns of the sensors. 6. Acknowledgement The authors would like to acknowledge support from European Framework 7 funding, contract number 266107 ‘Advanced In-Flight Measurement 2’ and EPSRC Grant number EP/H02252X/1. (For enquiries relating to access to the research data or other materials referred to in this article, please contact Cranfield University Library and Information Services—library@cranfield.ac.uk). 7. References [1] Boden F., Lawson N., Jentink H.W., Kompenhams J., "Advanced In-Flight Measurement Techniques", Springer-Verlag, Berlin (2013). [2] Kottcamp E., Wilhelm H. and Kohl D. “Strain Gauge Measurements on Aircraft”, AGARD and RTO Flight Test Instrumentation Series AGARDograph 160 (AG 160), Volume 7 (1976) [3] Wuest W., “Pressure and Flow Measurement”, AGARD and RTO Flight Test Instrumentation Series AGARDograph 160 (AG 160), Volume 8 (1980). [4] Measures R.M. “Structural Monitoring with Fiber Optic Technology”, London, Academic Press (2001) [5] D Betz, L Staudigel, M N Trutzel, "Test of a fiber Bragg grating sensor network for commercial aircraft structures," Optical Fiber Sensors Conference Technical Digest, 2002. OFS 2002, 15th , pp. 55- 58 (2002) [6] J-R Lee, C-Y Ryu, B-Y Koo, S-G Kang, C-S Hong, C-G Kim, "In-flight health monitoring of a subscale wing using a fiber Bragg grating sensor system," Smart Materials and Structures 12, pp. 147-155 (2003) [7] J Read and P D Foote, “Sea and flight trials of optical fibre Bragg grating strain sensing systems” Smart Materials and Structures 10 1085–1094 (2001) [8] Rao, Y.J. “Recent progress in fibre-optic extrinsic Fabry-Perot interferometric sensors,“ Optical Fibre Tech., 12(3), pp. 227-237, 2006 [9] Chehura E, James SW, Tatam RP, Lawson N and Garry KP (2009) “Pressure measurements on aircraft wing using phase-shifted fibre Bragg grating sensors,“ 20th International Conference on Optical Fibre Sensors, 2009, Edinburgh, 5th – 9th Oct 2009 [10] J-R Lee, C-Y Ryu, B-Y Koo, S-G Kang, C-S Hong, C-G Kim, “In-flight health monitoring of a subscale wing using a fiber Bragg grating sensor system,“ Smart Materials and Structures 12, pp. 147-155 (2003) [11] Lawson N.J., Gautrey J.E., Salmon N., Garry K.P., Pintiau A., “Modelling of a Scottish Aviation Bulldog using Reverse Engineering, Wind Tunnel and Numerical Methods,“ IMechE Part G, Journal of Aerospace Engineering pp7, DOI: 10.1177/0954410014524740 (2014) [12] Lawson N.J., Salmon N., Gautrey J.E., Bailey R. “Comparison of Flight Test Data with a Computational Fluid Dynamics Model of a Scottish Aviation Bulldog Aircraft” The Aeronautical Journal 117(1198) 1273- 1291 (2013) 8. Glossary FBG: Fibre Bragg grating EFFPI: Extrinsic Fibre Fabry Perot Interferometer RFSG: Resistive Foil Strain Gauge AHRS: Attitude and Heading Reference System CFD: Computational Fluid Dynamics (n – 1): Normal g-load increment Cp: Pressure Coefficient ax: AHRS Inertial longitudinal axes ay: AHRS Inertial lateral axes az: AHRS Inertial directional axes

Creative Commons Aucune (Tous droits réservés) Aucune (Tous droits réservés)
contenu protégé  Document accessible sous conditions - vous devez vous connecter ou vous enregistrer pour accéder à ou acquérir ce document.
- Accès libre pour les ayants-droit

ETTC 2015– European Test & Telemetry Conference Recent achievements in Doppler Lidars for aircraft certification C.Besson, B.Augère, G.Canat, A.Dolfi-Bouteyre, D.Fleury, D.Goular, J.Le Gouët, L.Lombard, C.Planchat, M.Valla, W.Renard ONERA, Chemin de la Hunière, 91321 Palaiseau (France) Abstract: We report on the performance and field tests of recently developed fiber Doppler Lidars. Two types of systems are considered: ground based range resolved Lidar and airborne true air speed sensor. These systems can be used for certification of new air vehicles (fixed wing or rotary wing) and ease their integration in air traffic Keywords: flight test instrument, certification, Lidar, fiber laser, coherent detection, true air speed, TAS, wind, turbulence, EDR 1. Introduction During landing and take-off, a minimal distance separation between aircrafts is necessary in order to avoid the risk of wake vortex encounter from a preceding aircraft. Indeed, wake vortices are two coherent counter- rotating flows created behind the aircraft wings and they induce a potentially dangerous rolling moment to the following aircraft. Atmospheric conditions determine wake vortices lifetime and trajectory. It has been shown that wake vortices dissipation rate varies depending on atmospheric turbulence level. They can also be transported out of the way on oncoming traffic by cross- winds. Other air disturbances such as wind gust or rapid change of the incoming wind direction are also detrimental to airport traffic flow. Thus anticipation of these phenomena in the vicinity of airport is a key information for air traffic optimization and safety. Not only airport safety but also flight tests of new air vehicle, manned or unmanned, could benefit from accurate knowledge of air dynamics disturbances in the vicinity of airports. Wake vortices locations and trajectories, wind turbulence level or wind maps are ancillary data that can be used during flight tests analysis. They can be provided by long range range resolved Doppler Lidar or Radar which measure the wind speed with a high spatial resolution [1]. Such sensors are being evaluated for airport safety and re-categorization purpose in the framework of various projects such as SESAR, CREDOS or FIDELIO [2][3]. Measuring the wind speed at a short distance from the aircraft is also useful for true air speed retrieval during certification procedures. Indeed, calibration of air data sensor of aircraft requires cumbersome procedures including specific equipment and dedicated costly flight tests. Although several techniques have been developed over the years, the most direct and probably the most accurate way to directly obtain the correction factors is to compare the aircraft air data measurements with optically derived onboard measurements obtained non-intrusively from the free stream region in front of the aircraft. Calibrations of the pitot static system and vanes using a laser anemometer have an increased accuracy compared with those obtained with conventional techniques, such as using a towed cone, tower-fly-by, or a pacer aircraft. A short range Lidar allows a precise and remote measurement of air speed just outside the range of the flow disturbance from the aircraft: it is able to give the velocity in real time with no in-flight calibration using autonomous onboard equipment and without a priori assumptions on the atmosphere. In this paper we review recent Lidar achievements at Onera and report on performance and field tests results for two types of Lidar: ground based range resolved Lidar and airborne true air speed sensor. Doppler Lidar (Light detection and ranging) is a well-known sensing technique for the retrieval of air speed. The systems described in this paper are based on Doppler shift determination of a light wave obtained from a pulsed or continuous single frequency laser that is reflected on natural atmospheric aerosols (Mie scattering). The aerosols are the wind field tracers to be analysed. The frequency shift is proportional to the air velocity and is detected via an interferometer measuring the beat frequency between the backscattered wave from aerosols and a reference wave (local oscillator). Coherent mixing enables the recovery of the backscattered wave phase. This phase contains the radial velocity information (along the laser line of sight). It also enhances the detection sensitivity thanks to the optical product of the signal beam with the reference beam which enables small target signal amplification. Lidar based on fiber technology are well adapted to on-the-field or airborne operations thanks to their intrinsic vibration-resistant designs. When emitting around 1.5 µm, they can benefit from telecom industry components with large market at competitive costs and increased reliability. They offer simplified maintenance procedures compared to free-space technology and enable compact system designs. ETTC 2015– European Test & Telemetry Conference 2. Range resolved wind Lidar Long range range resolved coherent scanning wind Lidar can provide radial wind velocity that can be processed in 3D wind maps as well as EDR. For airport safety applications, extended range up to 10 km, as well as fast large area coverage with refresh rate below 10 seconds are necessary. It is now possible thanks to recent progress in high power single-frequency all-fiber lasers, a key component of the system. Indeed, such Lidar require narrow linewidth (few MHz) pulsed laser sources emitting in the µs regime with kW peak power [4][5]. The development at Onera of new high power lasers yielded to such class wind Lidars which have been field tested in 2014 and 2015. 2.1 High power pulsed fiber laser Eyesafe, all-fiber laser sources based on MOPFA (Master Oscillator Power Fiber Amplifier) architecture offer many advantages over bulk sources such as low sensitivity to vibrations and emission versatility. These sources have very good efficiencies and can bear high thermal load, enabling high repetition rate pulsed emission typically from 10 to 100 kHz. However narrow linewidth MOFPA peak power is limited by stimulated Brillouin scattering (SBS) and specific strategies must be deployed to mitigate this effect which usually limits the output peak power to ~100W in single mode fibers. The fiber power handling can be improved without degrading the beam quality by increasing the fundamental mode effective area while maintaining a (quasi-) singlemode propagation. For this purpose, various LMA (large mode area) fiber designs have been proposed at 1µm. For 1.5µm operation, quasi-singlemode propagation in LMA fibers is more challenging. Indeed, the required high index codopants increase the core numerical aperture (NA) and the number of guided modes thus decreasing the resulting beam quality. For example, erbium-ytterbium doped fibers (Er-Yb) require high concentration phosphorous codoping. Ytterbium free, erbium doped fibers require alumina codoping. Commercial LMA fibers can typically emit up to 300W. For this reason, various strategies have been proposed to maintain a good beam quality (see [6] to [10]). Microstructured cores using Erbium-Ytterbium codoped materials have been proposed leading to 940W peak power. However they suffer from manufacturing complexity. We have also developed single-frequency all-fiber amplifiers based on Er-Yb doped LMA fibers with optimal composition. Pulse energy up to 450µJ was achieved with excellent beam quality. Another method to mitigate SBS is to apply a strain gradient along the fiber. This translates into the fiber into an acoustic velocity gradient along the fiber, and thus into an inhomogeneous broadening of the Brillouin gain spectrum. Thanks to this Onera proprietary method, we raised the peak power and energy of the laser source up to 600 W and 500 µJ respectively, which represents a gain of more than 3 dB compared to the same fiber source without strain gradient. 2.2 Long range range resolved wind Lidar tests High power MOFPA lasers can be integrated in a monostatic coherent Lidar architecture such as the one depicted on Figure 1. 50/50 90/10 Preamp. Ampli. de Puissance MI 1545 nm CW laser DET Traitement de signal Rétrodiffusion des aérosols 2 1 CSP Lame λ/4 2.5 3 3.5 4 4.5 Figure 1: MOFPA coherent fiber Lidar set up The master oscillator is a laser diode emitting 1545 nm in continuous regime. Its output is split thanks to a 50/50 fiber coupler. The signal channel (1) goes through an intensity modulator (MI) which shapes the pulse and is then amplified through a preamplification stage and a booster stage. At the output of the booster stage, the beam is polarized, quasi-single mode and its temporal shape is optimized for efficient coherent detection. A passive fiber pigtail is used to connect to the beam splitter. A polarization beam splitter (CSP) enables to circulate the emitted signal channel and the detection channel (2) thanks to a quarter wave plate. Up to the beam splitter all components are fibered. At that point however, the Brillouin effect would occur in fiber and free space optics are used. At the telescope output the signal beam is emitted in the atmosphere boundary layer with a slant angle of a few degrees. The signal backscattered from natural aerosols is coupled in fiber 2 of the beam splitter. It is mixed with the local oscillator thanks to a fiber coupler. The electric signal of the photodetector is analysed and processed in real time. Onera Lidar, LICORNE, is a convenient tool to test different Lidar configurations, signal processings, components and lasers. Wind speeds have been measured with various fiber amplifiers: a 100 µJ commercial amplifier and two homemade amplifiers delivering 400 µJ and 600 µJ. For similar optimal meteorological conditions and using the same Lidar parameters (10 kHz repetition rate, 1024 laser pulse accumulation) the typical maximum ranges obtained are respectively 2,5 km, 10 km and 15 km [9][10]. The range resolved Lidar has a spatial resolution of 150m and displays complete wind profiles in 100 ms. (see Figure 2). range (m) velocity(km/h) Plateforme LICORNE :12 Jan 2015 12:36:25.600 averaging =0.11378 s 0 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 11000 12000 13000 14000 15000 16000 0 20 40 60 80 100 120 0.5 1 1.5 2 Figure 2: Long range wind Lidar LICORNE: air speed vs range - color code is the frequency power spectrum density - spatial resolution: 150m, maximum range: 16 km. The two high power lasers are appropriate for long range large area wind monitoring and these results are to ETTC 2015– European Test & Telemetry Conference the best of our knowledge the longest ranges achieved with a 1.5 µm all-fiber wind Lidar. 2.3 Long range scanning Lidar for airport field tests High power laser sources enable high speed data acquisition and therefore efficient scanning Lidar operation. This is needed for large area monitoring such as airport surveillance. During UFO European project Onera assembled a high power fiber amplifier emitting more than 500W peak power with excellent beam quality [1]. The laser has been integrated in Lesophere Windcube®, an ultra-fast scanning ground-based Lidar. The system shown on Figure 3 has been operating continuously for 2 month at Blagnac airport and enabled the retrieval of relevant quantities for future Weather Dependent Separation (WDS) concepts [11]. Thanks to the new high power laser, wind speed maps were provided at ranges beyond 10 km with a 45 degrees horizontal scan in no more than 8 seconds. A typical radial wind map is shown on Figure 4. Figure 3: UFO high power laser installed in Windcube Lidar Figure 4: radial wind map acquired at Blagnac airport 2.4 EDR retrieval Atmospheric conditions determine wake vortices lifetime and trajectory. It has been shown that wake vortices dissipation rate varies depending on atmospheric turbulence level (characterized by the eddy dissipation rate, EDR). EDR retrieval from Lidar data remains a relative new topic especially for addressing operational purposes as air traffic applications. Doppler Lidars can provide information about wind field spatial statistic and then give an estimation of the turbulence or Eddy Dissipation Rate [12][13][14]. The estimation can be made from - Doppler Spectrum width, - Velocity Variance, or - Velocity Structure function EDR estimation algorithms, although using different processing techniques, all rely on power spectral representations of turbulence. In this approach, the power spectrum density of the velocity fluctuations in the inertial range has a universal shape based on the Kolmogorov theory. For a scanning Lidar, the azimuthal structure function method is preferred. The EDR value is then obtained by fitting the 2/3 slope for the structure function Dv. The output value is often ε1/3 (m2/3 .s-1 ) 3/23/2 )( sCsD vv ε= Where s is spatial unit, ε is the energy dissipation rate and Cv ≈2, the Kolmogorov constant. EDR retrieval has been performed on Lidar data obtained during UFO trials at Toulouse Blagnac airport. An example is given below for a set of PPI scans (PPI with elevations from 2° to 45° and azimuth 47° to 293°), as a function of time, and for different altitudes. Figure 5 shows an example of velocity structure function fit for measurement points width-height between 150 and 200 m and averaged over 10 mn. 0 100 200 300 400 500 600 700 800 0 0.05 0.1 0.15 0.2 0.25 15-Apr-2014 09:14:47 averaged 10 mn m m²/s² EDR 1/3 = 0.0476 m 2/3. s-1 L0 = 505 m Azimuthal structure fonction for 150m https://www.eurocontrol.int/sites/default/files/content/docu ments/sesar/credos-d2-6-wp2-final-report-v11.pdf ETTC 2015– European Test & Telemetry Conference [4] J.-P. Cariou et Al. “Laser source requirements for coherent LIDARs based on fiber technology”, Comptes Rendus Physique, Volume 7, Issue 2, March 2006, Pages 213-223. [5] X. Zhang et Al. “Single-frequency polarized eye-safe all- fiber laser with peak power over kilowatt”, Applied Physics B, pp. 1-5 (2013). [6] G. Canat, et al., “Multifilament-core fibers for high energy pulse amplification at 1.5 µm with excellent beam quality”, Opt. Lett. 33, 2701-2703 (2008) [7] W. Renard et Al. “High peak power single frequency efficient Erbium-Ytterbium doped LMA fiber” Conference on Lasers and Electro-Optics Europe (CLEO Europe 2015).CJ-12.5;25/06/2015 [8] G.Canat et Al. “Eyesafe high peak power pulsed fiber lasers limited by fiber nonlinearity “Optical and Fiber Technology. Vol 20, N°6,pp. 678–687 10.1016/j.yofte.2014.06.010 [9] W.Renard et Al. “Beyond 10 km range wind-speed measurement with a 1.5 µm all-fiber laser source”, Conference on Lasers and Electro-Optics (CLEO 2014). San José (USA). 08-13/06/2014 [10] W. Renard et Al. “High peak power single frequency efficient Erbium-Ytterbium doped LMA fiber” Conference on Lasers and Electro-Optics (CLEO 2015). San José (USA). [11] L.P.Thobois et Al. “Wind and EDR Measurements with Scanning Doppler LIDARs for Preparing Future Weather Dependent Separation Concepts” AIAA Technical conferences 2014 [12] R.Frehlich et Al. “Measurements of boundary layer profiles in an urban environment”. Journal of Applied Meteorology, n°45, pp.821–837, 2006 [13] V.A.Banakh et Al. (1997). “Estimation of turbulent energy dissipation rate from data of pulse Doppler LIDAR”. Journal of Atmospheric and Oceanic Optics 10: 957–965. [14] R.Frehlich et Al. (1998). “Coherent doppler LIDAR measurements of wind field statistics”. Boundary-Layer Meteorology 86: 233–256. [15] Scott M. Spuler et Al."Optical fiber-based laser remote sensor for airborne measurement of wind velocity and turbulence", Applied Optics/Vol. 50, No. 6 / 20 (February 2011). [16] H. Inokuchi et Al., "Development of a long range airborne Doppler Lidar", 27Th International Congress of the Aeronautical Sciences (2010). [17] J.-P Cariou et Al., “All-fiber 1.5 µm CW coherent laser anemometer DALHEC. Helicopter flight test analysis”, 13th Coherent Laser Radar conference , Kamakura (2005). [18] B.Augere et Al. 1.5µm Lidar anemometer for True Air Speed, Angle Of Sideslip and Angle Of Attack measurements onboard Piaggio P180 aircraft; Measurement Science and Technology Journal, 2015. MST-102092.R1

Creative Commons Aucune (Tous droits réservés) Aucune (Tous droits réservés)
contenu protégé  Document accessible sous conditions - vous devez vous connecter ou vous enregistrer pour accéder à ou acquérir ce document.
- Accès libre pour les ayants-droit

ETTC 2015– European Test & Telemetry Conference New Upstream Rotating Measurement System for gases turbine exhaust gas analysis – NURMSys project A. Sylvain LOUME1 , B. Bertrand CARRE2 , C. David LALANNE3 , 1: AKIRA Technologies, ZA St Frédéric, Rue de la Galupe, BAYONNE France (64100) 2: AKIRA Technologies, ZA St Fréderic, Rue de la Galupe, BAYONNE France (64100) 3: AKIRA Technologies, ZA St Frédéric, Rue de la Galupe, BAYONNE France (64100) Abstract: Analysis and measurement of combustion efficiency is one key point for improvement of fuel consumption and pollutant emissions of aircraft engines. The thermal constraints due to combustion process leads to no possibility for direct measurement. A specific device has been developed to improve state-of-the art situation in terms of measurement precision and modularity to increase experimental capacities of our partner. Keywords: Combustion efficiency and pollutants emissions measurement, Gases turbine, Severe measurements environment, Integrated measurement device 1. Introduction In 2001, the ACARE (Advisory Council for Aeronautic Research in Europe) was founded to establish and maintain a technical roadmap [SRA – Strategic Research Agenda] outlining the orientations which should be taken to meet society's needs for aviation as a public mode of transport as well as noise and emissions reduction requirements in a sustainable way. This roadmap defined in particular very ambitious target for 2020 in terms of reduction of CO2 emission [-50%], NOx emission [-80%] and noise reduction [-50%]. All the stakeholders of the aeronautic market, and especially the engine manufacturers, are fully focused on research and development programs to reach those targets. Obviously, one key lever for emissions reductions is the improvement of aircrafts engine, and especially the optimization of efficiency of fuel energy conversion, namely the combustion process. To study combustion and though its efficiency and resultant pollutants emissions, it is essential to characterize exhaust burnt gases, measuring chemical species concentration, exhaust gases velocity, pressure and temperature in particular harsh environment. A new measurement device, dedicated to exhaust gases analysis and characterisation, was developed by AKIRA technologies. This project was part of the CLEANSKY program, the most ambitious aeronautical research program ever launched, and has been performed with our partner TURBOMECA – SAFRAN Group – which is the final user of this system. As the world leading helicopter engine company, TURBOMECA is deeply involved in the fuel conversion process efficiency optimisation. 2. Measurement device – concept description 2.1 State-Of-The-Art The measurement device developed is especially dedicated to the analysis of combustion taking place in small reverse flow gases turbines. The complex behaviour of this kind of combustors, linked to their compactness leads to even more difficult measurement precision and robustness. Nowadays, system architecture consists in positioning a rotating shaft equipped with racks downstream the combustor, surrounded by hot gases flow. Exhaust gases sampling system, thermocouples, velocity and pressure sensors are directly mounted on these racks and so moving compared to the combustion chamber. Displacement system is subjected to high level of thermal stress; typically exhaust gases temperature can reach 1600K. Despite continuous water cooling, this configuration correctly ensures neither good sensors operation nor measurement accuracy. This global approach also leads to very bulky installation as the sampling system and thermocouples need to be far enough the combustors to keep acceptable temperatures. Usual values can reach several meters between the combustion chamber and the measuring device. 2.2 Proposed solution The new system is completely embedded and useable with different combustors – combustion chambers. The measurements performed are the following ones: Exhaust gases composition Exhaust gases temperature The system allows to measure those characteristics on the complete area of the exit of the combustion chamber [annular shape] in order to established complete maps of temperature and gases composition at the exit of the combustors. The measurements are not only means values but also high frequency measurements in order to catch all the combustions dynamic behavior. ETTC 2015– European Test & Telemetry Conference In order to establish those 2D maps of gases composition and temperature, a rotating shaft fitted with 4 measurement rakes every 90 degrees, 2 for temperature measurements (typically 5 thermocouples each) and 2 for gases sampling, (one averaged and one at 5 discrete radii), is placed in the combustor axis. With the rotation of the shaft, this system allows to define the above maps in the racks plan, directly at the combustor exit. The breakthrough proposed is to completely change the concept of the measurement device and to move the gases sampling system, rotating shaft and motion controller upstream the combustor – in the intake area of the combustion chamber. By this manner the environment temperature for the measurements system and motion device is greatly reduced. Indeed the maximum temperature reached by the inlet air is never above 750K. Another aspect of the new concept is that the gases analyser and acquisition system is now fixed compared to the measurements racks and can be placed in more friendly environment This new concept leads to a much more compact system, more easily transportable from one test cell to another. This increased flexibility allows higher and faster testing capacity for different combustion chambers, so that research and development process is improved. Figure 1: New measurement system concept On the other hand, the new concept leads to face new difficulties. The complete system that is placed upstream the combustion chamber shall not interfere with the combustion process itself, means that the upstream flow does not have to be modified compared to the engine situation. This means that the system shall be integrated in the ogive of the engine, which is once again very tiny area especially for small gases turbines. The exhaust gases that are collected from the racks and moved to the gases analyser have to be maintain to a temperature of 190°C to avoid any water or unburnt hydrocarbons condensation that would leads to measurements deviation and mistakes. This temperature management system of the exhaust gases shall not interfere with the global cooling system through the rotating shaft. Because the gases analyser is now fixed compared to combustion chamber, an innovative sealing system that allows tightening of the pneumatic path of the gases from the racks to the analyser shall be implemented. This dynamic sealing has to be ensure through severe thermal conditions, means 450°C for the complete module at the combustion chamber inlet. A specific electrical rotating collector is installed for the thermocouples wiring. At the end, the cooling system and exhaust gases thermal management system, based on fluid transportation [air and water], also needs to implement dynamic sealing in severe thermal environment. All the exhaust gases path are located in one of the arms of the ogive, the water cooling system in another one and the cooling air and electrical wires in a third one. A specific attention has been put to ensure proper guidance of the rotating shaft, submitted to high temperature gradient and which is 400mm long. The driven shaft from the electrical motor to the rotating and collecting shaft is located through the fourth arm of the ogive. Because of external and geometrical constraints, the shaft diameter is 60mm, and all the exhaust gases path [6 in total], water path [2 for inlet and outlet], and the electrical wires for thermocouples are integrated is this tiny room, with corresponding tightening system. Figure 2: Superposition of the NURMSys and a typical gas turbine. Figure 3: Ogive arms with all pathes to the rotating shaft. ETTC 2015– European Test & Telemetry Conference 3. Measurement device – functional and technical description 3.1 Driven system for rotating shaft The shaft is driven by and external electric motor which is once again in a fixed compared to the combustion chamber. The motor is driven with an encoder that allows choosing the working mode: continuous motion or step by step. In continuous mode, the constant speed is 1 rotation in 12 minutes, and the “exploration” to establish the 2D maps is made over 370° to take into account gases transfer and analysis response time. The step by step mode allows making measurements for a given angular area of the combustion chamber. In this mode, a quicker rotating speed of 1 revolution per minute is available. One specific point is that the position measurement and servitude of the shaft and racks is absolute, means that there is no information losses in case of electricity cut and bench shut down. The coupling between the driven shaft through the ogive arm and the rotating shaft is made through a conical gear, and specific ball bearings are used to ensure rotating shaft guidance. Those ball bearings are specially studied in order to support severe thermal stress and proper precision for shaft guidance at high temperature. Figure 4: External electrical motor. 3.2 Measuring racks The measuring module is based on four racks, placed every 90 degrees around the shaft axis. The racks are made using 3D printing method with Nickel based metal. Two of those racks, placed in opposite position, are dedicated to the temperature measurement of the exhaust gases. 5 thermocouples per rack in different radial positions are integrated [K or B types] and another one for water return is installed to monitor cooling temperature. The two other racks are dedicated to exhaust gases composition analysis. On one rack 5 pathes located at different radial position are installed to monitor the gas composition in the 2D dimension. On the other rack, only one path is installed for mean measurement. 3.3 Thermocouples wiring The thermocouples wiring is located through one gallery drilled into the shaft. The electrical rotating collector is located inside the ogive of the system. The temperature of the collector has to be kept below 100°C, so that a specific insulation system with ceramic based material associated with a dedicated water cooling system have been implemented. Figure 5: Insulation of electrical rotating collector. 3.4 Air cooling A specific air cooling system is implemented for thermal management of the conic gearing and the rack holder at the bottom of the rotating shaft. This air is coming from one arm of the ogive, goes to the gearing and through the shaft, and is then expend with the exhaust gases downstream the racks, so that it has no impact on the complete combustion process. The maximum static pressure inside the module is 20 bars, and the pressure for the air cooling system is 15 bars. Finite elements computations have been performed to evaluate strength of the mechanical parts submitted to high pressure and temperature constraints. In addition to device resistance and durability, those computations were needed to ensure controlled deformation of the parts and so precise position of the racks and so proper measurements. The multi material nature of the complete assembly and the high thermal gradient environment led to even more complex computations that were needed to avoid any risk of locking during motion of the system, lack of tightening of the pneumatic and liquid collectors, of wearing/clearance inside the conical gearing – proper position serviture and measurement or the racks. ETTC 2015– European Test & Telemetry Conference Figure 6: Air cooling outlet at bottom of the module. 3.5 Water cooling A dedicated water cooling system is installed to ensure cooling of the racks and to stop any chemical reactions inside the exhaust gases during the travel from the racks to the analyser. The dynamic interface is based on the same principle as the pneumatic collector. Figure 7: Principle of the fluids rotating collector. The implemented solution does not lead to long exhaust gases pathes so that overcooling of the gases shall not happen. Anyway, and additional thermal management system located at the exit of the ogive is installed to maintain the gases temperature to the desire threshold. This system is based on an oil-gases heat exchanger, with a dedicated oil network with a temperature regulated to 190°C – Water or Unburnt hydrocarbons condensation. 3.6 Control and acquisition system AKIRA technologies has developed a complete turnkey measurement device. In addition to the mechanical constraint and implemented solutions presented above, an embedded software and electrical hardware system dedicated to control and acquisition is developed. A modular sensors architecture is proposed based on a collection of data acquisition cards. The control software ensures: Actuators control Acquisition and signal treatments Monitoring/safety of the device Recording of the datas in specific files Specific computations [especially combustion efficiency based on gas composition measurements] Interface with user and on site system [network, etc..] Figure 8: Acquisition device. All the software has been developed using LabVIEW© In continuous mode, the data recording is performed every degree, from a finite and absolute angular position value. Measurements consist in : 10 temperatures from thermocouples that can be recorded at a maximum frequency of 20ksamples/s. Gas concentration [CO2, CO, NO, NOx, O2, UHC (unburnt hydrocarbons) Additional measurements as instantaneous pressure [5 maximum] and lightning intensity at a maximum rate of 20ksamples/s are included in the acquisition system. Those signals are coming from additional sensors not included in the presented module. In the step-by-step mode, an adjustable delay fixed by the user is implemented. In order to avoid any electrical disturbance on the signals, the data acquisition cards are located in a dedicated frame and linked to the computer with a dedicated optical connection not sensible to electrical disturbances. ETTC 2015– European Test & Telemetry Conference 5. Conclusion The measurement of exhaust gases characteristics in “on chamber” conditions is one of the key point for improvement of combustion process analysis and improvement. The environmental conditions in a combustion chamber are not compatible with direct measurement. Also the precision in terms of measurement itself and measurement location in very tiny environment is of first order importance for gas turbine research and development. In collaboration with its partners TURBOMECA – SAFRAN Group, AKIRA Technologies has developed a complete embedded measurement device available to allow the final user to make the analysis of combustion chambers with the required precision and robustness. To reach those measurements accuracy, a specific concept and associated mechanical breakthroughs have been developed to overcome the harsh thermal environment.

Creative Commons Aucune (Tous droits réservés) Aucune (Tous droits réservés)
contenu protégé  Document accessible sous conditions - vous devez vous connecter ou vous enregistrer pour accéder à ou acquérir ce document.
- Accès libre pour les ayants-droit

ETTC 2015 – European Test & Telemetry Conference Page 1 The Falcon 5X Measurement System Jean Pierre Rouby Dassault Aviation Flight Test Directorate - Istres (France) Abstract: The Falcon 5X is the latest Dassault Aviation business aircraft. Its detailed design began in 2011 and it is about to make its first flight. Three test aircraft will be used to carry out all the certification tests with first deliveries by year 2017. This paper presents the architecture of the measurement system of the Falcon 5X and its specificities as well as the methodology applied during its design and testing. Keywords: Flight Test Installation, Measurement System, digital buses, sensors, telemetry, Ethernet 1. The Falcon 5X program The Falcon 5X (or F5X) is the latest addition to the range of Dassault Aviation business aircraft. It incorporates many technological changes and new features which make it the most advanced aircraft in the range, and one of the best performing and most comfortable of business aviation:  All-new wings designed to achieve new aerodynamic efficiencies  All-new Safran-Snecma Silvercrest engines  New generation of Digital Flight Control System (DFCS) able to manage all moving surfaces including flaperons and nose wheel steering  A new Head-Up display (Combined Vision System) allowing information to be presented from both the Enhanced and Synthetic Vision Systems  3 rd generation EASy avionics suite  Bigger and more comfortable flight deck  The widest cabin in a purpose-built business jet and among the largest in business aviation  Skylight ceiling window providing natural light from above in the entryway and galley zone in forward fuselage  28 large, expansive windows providing unbeatable luminosity The Falcon 5X has a maximum range of 5,200 nautical miles at its long-range cruise speed of Mach .80. Top speed is Mach .90. Typical cruise altitudes are of 43,000 to 47,000 ft on long range missions. The maximum operating altitude is 51,000 ft. The detailed design phase started at the beginning of year 2011. The first flight of the first F5X test A/C is expected in the very next months. The objective is to obtain the certification before the end of year 2016. The corresponding flight tests will be performed mainly in Istres (France) with 3 development A/C. ETTC 2015 – European Test & Telemetry Conference Page 2 2. An overview of the F5X Measurement System 2.1 Development process On test aircraft, the measurement system (MS) of the flight test installation (FTI) includes the equipment used to collect, record and broadcast the data required for monitoring the test in the Flight Test Room (FTR) and required for post-flight analysis. These data can be collected from many sources: physical parameters measured with specialized sensors (temperatures, vibrations, pressures…), functional or FTI digital buses, video cameras, microphones. The MS core includes the pieces of equipment which actually collect, transmit and record the FTI data. The MS core doesn’t include the FTI data sources which are referred as the actual instrumentation installed on the various systems of the aircraft. During the design phases of the F5X program, the MS core has been managed in the same way as the other functional aircraft systems. In that capacity, the development methodology applied to the functional systems has also been applied to the F5X measurement system: same design documents (ICD…), same formalism required for the documents, same schedule, same design reviews (PDR, FDR and CDR) and lay-out in the digital mockup. The definition of the measurement system began with the detailed design phase of the A/C, at the beginning of the year 2011, leading in parallel:  The collection of the need for each of some 40 functional systems to be instrumented and writing interface specification documents (FTI ICD's) with industrial partners  The beginning of the design of the core of the measurement system and validation testing of new equipment and architectural principles 2.2 Requirements from test engineers We mention below a list of initial requirements specified by our test engineers taken into account during MS core design:  Capacity to record the FTI data during 10 hours in flight  Capacity to transmit FTI data and one Video signal by telemetry during standard test flights, and FTI data through SATCOM in case of remote test flight  Ability to deal with the following input: o A large number of Analog sensors :  ~1200 on F5X#1  ~700 on F5X#2  ~400 on F5X#3 o About 80 ARINC buses o Honeywell EPIC eASCB avionics bus: full acquisition for post flight analysis and limited set of parameters for telemetry (TM) o Ethernet buses from various systems (DFCS, Electrical Generation, Maintenance System, IPPS instrumentation) o Cockpit Audio (to be recorded with videos and transmitted by TM)  Ability to process about 20 video streams (FTI cameras and 4 avionics display units video output): o To be recorded on board o To be transmitted by TM and displayed in FTR o To be displayed in the cockpit and in the cabin for the crew  Have an embedded test monitoring station for a flight test engineer (FTE) in order to have the same features as in the FTR  Have in the cockpit 2 FTI color video displays (Pilot & Copilot) to display, in real-time, FTI data synoptic and videos  Precise time correlation for the acquisition of all the parameters 2.3 Technical choices The Falcon 5X program was an opportunity to make technical choices to modernize the measurement system of our future civil aircraft. This concerns in particular the Falcon 8X program whose development A/C have a measurement system with architecture identical to that of F5X. The main changes introduced on F5X and F8X are the following:  MS Core based on Ethernet technologies  Data Acquisition Units (DAU’s) synchronization system based on the use of PTP v2 time protocol  Analog DAU’s with Ethernet link for configuration, real-time data output and clock synchronization  New generation time server able to provide NTP and PTP v2 messages and IRIG-B signals  New generation data recorder  Complete redesign of the video acquisition and recording system based on digital technologies  For the crew, independent FTI displays (also called VIP displays), with touch panel, able to manage selectable data synoptic or video stream  FTI remote command in the FTR available to the Test Engineer to reduce the workload of the crew 2.4 Measurement system description A synoptic of the F5X MS architecture is given at the end of this paper. The F5X MS consists of the following sub-assemblies: ETTC 2015 – European Test & Telemetry Conference Page 3  DACQNET is responsible for the acquisition and recording of analog sensors and digital buses. It is also responsible for the transmission of a portion of this information by telemetry. It also integrates the GPS receiver with its own antenna and the time server  VIDEONET is responsible for the acquisition, recording and transmission of all video streams. These streams are transmitted to the INFONET subset and by TM to the FTR  INFONET is the embedded computer system capable of providing the services of a FTR monitoring station for a flight test engineer (FTE) and supply the information to be displayed on the Pilot & Copilot VIP displays installed in the cockpit  The telemetry (TM) system is used to transmit data and a video image in real-time to the FTR. It uses the S-BAND and SATCOM  The MS control system consists of the control panels in the cockpit and in the cabin and a remote control in the flight test room  Sensors & Buses : All analog sensors added in the A/C to perform the requested measures and taps added to acquire in safety functional digital buses and aircraft systems sensors 3- DACQNET – Sensor and data buses acquisition DACQNET is the main subset in the MS responsible for the acquisition and recording of analog sensors and data buses. It is also responsible for building and sending the real-time PCM message sent by telemetry to the FTR. Figure 3.1 F5X#1 DACQNET rack It consists primarily of acquisition boxes connected to an Ethernet network. Equipment Manufacturer Function KAM-500 Acracontrol, Ir Analog sensors DAU SARI-NG ADAS, Fr Analog sensors DAU DataTap-10 ICS,U.S.A eASCB Avionic bus DAU DIANE AMESYS, Fr Digital buses (Ethernet, Arinc & Serial) DAU PCM telemetry output MDR Zodiac Data Recording MAR-1040 Eth. switch Hirschmann Ethernet data streams distribution Table 3.1 –Main DACQNET equipment All the acquisition boxes transmit collected data in the form of Ethernet UDP/IP streams. These boxes are all connected on Ethernet switches whose main role is to direct each of these streams to the data recorder and/or DIANE boxes which select parameters to be included in the TM PCM message. All the Ethernet flows pass through the DACQNET Ethernet switches. The number of SARI-NG and ACRA DAU’s, as well as the number of Ethernet switches, varies according to the considered test A/C, depending on the number of analog sensors to be processed, as shown below. F5X#1 F5X#2 F5X#1 Analog sensors ~1200 ~700 ~400 SARI-NG 25 10 7 ACRA 3 2 1 DataTap-10 1 1 1 DIANE 2 2 2 MDR 1 1 1 MAR-1040 4 3 2 Table 3.2 – MS boxes distribution A DAU data stream may contain data to monitor during the flight. It is then routed to both a DIANE box and to the MDR recorder. When this is not the case, the flow is routed only to the recorder. The DataTap-10 (ICS, USA) makes it possible to collect avionics parameters on the main eASCB buses (Pilot and Copilot). First of all, it produces an Ethernet stream containing the entire eASCB traffic which is only recorded. It also produces an Ethernet stream containing a selective list of parameters that is inserted by DIANE box into the TM PCM message. DIANE boxes are primarily responsible for building the TM PCM message. They therefore receive all streams containing parameters to be monitored in the FTR. These Ethernet streams come from MS DAU’s but they can be also Ethernet streams generated by functional aircraft systems (DFCS, Power Generation, and Multipurpose Maintenance System). The ARINC buses and serial links (RS232, RS422 …) are directly connected to the DIANE boxes which collect real-time parameters. They are then fully duplicated on an Ethernet output (called HD-Channel) which is connected to a DACQNET switches and routed to the recorder. All the DAU’s are synchronized to the UTC time extracted from the GPS message and distributed in various forms (PTP V2 and IRIG-B in particular) by a LANTIME M600 time server. This allows "à la source" data time stamping. ETTC 2015 – European Test & Telemetry Conference Page 4 Details on the main Ethernet MS network and synchronization system will be found further in this paper. 4 – VIDEONET – Video acquisition, recording and broadcasting VIDEONET is the subset responsible for the acquisition, recording and broadcasting of all video streams. It consists of entirely new equipment developed in the context of the Falcon 5X program by TDM Company (Merignac, France). The main equipment is a rack which integrates an Ethernet switch, a Video encoder and a Video server for processing up to 20 Video streams. The encoder consists of a processor board and up to 5 SDI acquisition boards with 4 input channels each. It also performs broadcasting to the server and to external devices in low or high resolution. The server builds the video files which are recorded on a NAS for post flight analysis. Figure 4.1 – TDM video rack At any time, one of the video streams is available for the FTR and sent by telemetry thanks to an ETH/PAL converter also provided by TDM Company. This device performs the conversion of an RTP stream to a PAL video signal. The corresponding input video stream is selected from the FTR by the test engineer with a remote control. The code sent by the remote control activates a combination of 5 input discreet on the TDM ETH/PAL converter. Video flows are also broadcast to the INFONET subset, allowing their display on the VIP screens in the cockpit. The video streams come first from FTI cameras positioned to film various parts of the aircraft (activity in the cockpit, wings, landing gears...). These cameras are MRCC from ADIIS Company. VIDEONET is also responsible for acquiring the information displayed to the crew on 4 MDU and PDU avionics displays. It is then to acquire the DVI output of 4 avionics graphics modules (AGM) through 4 DVI/SDI converters also developed by TDM. These converters can select the active output from the 2 outputs of the corresponding AGM. The encoder of the video rack is able to produce in real- time each video stream in the form of HR and LR (High/Low Resolution) Ethernet RTP stream. For reason of lower latency, the LR flows are used in TM for restitution in the FTR and by INFONET for display in the cockpit for the crew. The HR flows are used by the server to build the video files recorded on the NAS. 5 –INFONET– Embedded computing The MS embedded computer system has 2 functions: • provide the services of a FTR monitoring station for a Flight Test Engineer inside the cabin • control the Pilot and Co-Pilot FTI displays installed in the cockpit to allow the display of synoptic with real-time FTI parameters and FTI video It consists of a group of four computers organized around a MAR-1040 Ethernet switch (same model used in DACQNET). One of these computers analyses in real time the TM PCM message output by DIANE box and broadcasts the decoded parameters to the 3 other computers which are only used for display management. The first computer manages the display screen of the FTE. Each of the two others manages a VIP display. Figure 5.1 –Videonet/Infonet Rack The input of INFONET subset is: • TM PCM message generated by DIANE master box (duplication of the PCM message sent to the FTR) • Low-resolution Ethernet Video streams produced by the TDM Video encoder The application software tools available in INFONET for flight monitoring are exactly the same as those used in the FTR. The FTE can thus use all the synoptic set for the FTR and follow the test in the same way. It also has the same remote control keyboard. The embedded test monitoring station is therefore equivalent to a single station FTR. The pilot and co-pilot have independent VIP display running in Video mode or FTI data mode. The VIP screen is a commercial off-the-shelf 7’’ color video screen with a touch panel. Its controls (ON/OFF, brightness) have been adapted for use in the cockpit. Figure 5.2 –VIP displays position ETTC 2015 – European Test & Telemetry Conference Page 5 In Video mode, the crew can select, through a dedicated menu, one of the video channels managed by the TDM encoder. In Data display mode, the crew can select, through the “Synoptic” menu, to a set of synoptic whose ergonomics is adapted for use on a limited screen. The VIP displays are independent each other and controlled by a dedicated computer. The pilot can for example choose to display the image of a camera while the copilot is monitoring individual FTI parameters on the corresponding synoptic. Note that the output of a VIP PC can also be sent to a functional display unit of the EASy III system. The control of the display is then performed with a dedicated USB trackball which replaces the touch panel. This operating mode can for example be used during acoustic certification tests for displaying a Real-Time guidance page. 6- Routing the Ethernet data flows The MS has to deal with a large number of Ethernet flows (coming from FTI DAU’s or from external systems). Every stream must be directed to the recorder (in any case), and to DIANE boxes when it contains data to be monitored in the FTR. They are mostly UDP/IP streams using Multicast addressing mode. This is always true for FTI data coming from MS DAU’s (SARI-NG, ACRA, DIANE). The Multicast mode facilitates the switching of flows within an Ethernet switch. It’s used in conjunction with static routing tables for associating a Multicast address to a set of physical output ports of the switch. These ports are the ones that are connected to the acquisition channels of the recorder and of the DIANE boxes. The principle of routing Multicast frames by the switches internal static tables is used both within DACQNET and INFONET subset. In the case of the A/C Multipurpose Maintenance System (MMS), all the frames are basically sent in Unicast to a functional recorder. MMS frames are sent to the MS via the mirroring ports of the MMS Ethernet switches. In this case, frames are routed in a DACQNET switch through dedicated VLAN’s. Flows distribution on the various acquisition channels of the recorder and DIANE boxes has been adjusted gradually during the design of the MS with the constant concern of balancing the load on the various acquisition channels. 7 – MS Equipment synchronization Many MS equipment are synchronized with the PTP v2 network protocol. This is the case for example for SARI- NG and ACRA DAU’s. This is also true for INFONET computers. Other pieces of equipment, sometimes less recent, do not have this feature and must be synchronized with an IRIG-B signal. This is the case for DIANE boxes and MDR recorder. Furthermore, the TDM Video rack is synchronized through NTP network protocol. The time server used is a Meinberg LANTIME M600. It can handle all PTP v2, NTP and IRIG-B protocols. It synchronizes itself on the UTC time through a GPS antenna dedicated to the MS. Hirschmann MAR-1040 are PTP v2 switches that are configured in Boundary Clock mode and therefore play the role of PTP Grand Master for the DAU’s connected to them (SARI-NG, ACRA mainly). KN-Systems AES is new equipment, specially developed to allow control of the good timing of all MS boxes. It uses as input the LANTIME PPS signal (as the reference) and the PPS signal of every DAU’s (SARI-NG, ACRA, DIANE). It verifies that the maximum difference between the various DAU’s PPS and the reference is included in a programmable time interval (50 usec currently) to decide the right synchronization of the whole. The global and detailed synchronization status is indicated through LEDs on the equipment and also transmitted in a digital message decoded and displayed in the FTR. This box is very helpful to the FTI engineer during the initialization phase of the MS and often watched during the flight in the FTR. 8 - Control System and Telemetry Telemetry allows receiving in real time, in the FTR, stream of FTI parameters (sensors or data taken from digital buses) and a video stream among those managed by the VIDEONET sub-system. Telemetry uses S-band and the PCM message is transmitted at 1.4 Mbps. The PCM message can also be sent via SATCOM. We then use an additional box that converts the PCM message into an Ethernet UDP/IP stream. In the FTR, the remote control of the test engineer allows in particular, the following: • Selecting the Video stream sent by TM • Choosing the DIANE acquisition program (P1/P2) • Resetting SARI-NG clusters • Resetting DIANE boxes Some control panels are available in the cockpit, allowing the crew the following: • Turning ON/OFF the Measuring System general power • Turning ON/OFF the VIDEONET power • Start/Stop the video recording • Resetting vital equipment (Diane boxes & Flutter SARI- NG cluster) Finally, the FTI engineer also has a control panel in the cabin which allows MS equipment configuring and the check of the smooth functioning of the whole before flight departure. ETTC 2015 – European Test & Telemetry Conference Page 6 9 – Progress and Prospects This new measurement system was designed and developed in about four years. A laboratory test rig has been specially set up to test as soon as possible the new principles of architecture and operating modes, concerning in particular the networking and the global synchronization process. An MS of the same type has already been used for several months on 2 Falcon 8X test aircraft and gives complete satisfaction. It has also been used operationally during ground tests and engine runs on the Falcon 5X#1. We now look forward to the first flight of the F5X#1 for its final validation. 10. Glossary A/C Aircraft AGM Advanced Graphic Module CDR Critical Design Review DAU Data Acquisition Unit DFCS Digital Flight Control System FDR Final Design Review FTE Flight Test Engineer FTI Flight Test Installation FTI core Equivalent to MS core FTM Flight Test Measurement System FTR Flight Test Room GPS Global Positioning System ICD Interface Control Document IP Internet protocol IPPS Integrated Power Plant System IRF Interface de régulation FADEC (IPPS FTI device) MDU Multifunction Display Unit MMS Multipurpose Maintenance System MS Measuring system MS core Equipment to collect, record and transmit FTI data NAS Network Attached Storage NTP Network Time Protocol PCM Pulse Code Modulation PDR Preliminary Design Review PDU Primary Display Unit PPS Pulse Per Second PTP Precision Time Protocol RTP Real-Time Protocol TM Telemetry TS Transport Stream UDP User Datagram Protocol UTC Universal Time Coordinated VIP Very Important Parameters (Pilot/Copilot FTI display) ETTC 2015 – European Test & Telemetry Conference Page 7 11-Annex Figure 11.1 – The F5X#1 Measurement System - General architecture

Creative Commons Aucune (Tous droits réservés) Aucune (Tous droits réservés)
contenu protégé  Document accessible sous conditions - vous devez vous connecter ou vous enregistrer pour accéder à ou acquérir ce document.
- Accès libre pour les ayants-droit

High speed development of a temperature remote acquisition system to reduce instrumentation heat sink in an aircraft engine - Jean-Christophe Combier - AIRBUS Operation SAS – France. For new aircraft certification, engine contains a lot of instrumentation in a small confine area. New high performance engine tend to big external volume but less empty slot for instrumentation. Even if acquisition systems are more powerful, compact and generic, the heat sink from these systems is too high for device around and thermal aspect become a major issue. For the A320 NEO engine, we had 7 months to produce a new solution: A very low consumption device for the major measurement (Thermocouple) with no impact on the definition. The selected solution had been based on: - A partnership selected on his technical skill, technology watch and his agility to drive this kind of subject, - A system based on the best mass market component, - A real-time decision in AIRBUS or with the supplier, to manage the risks, - Demonstrator develops early in the development to derisk all technical topics in the first month. - Flexibility from power (28V DC or power over Ethernet) with 2 kinds of output (ARINC429 or IENA Ethernet packet) and +/-2°C of global uncertainties upon -55°C to +105°C. The result was 64 channels thermocouples, compliant with the mechanical and electrical definition from the generic acquisition system, going from 70W to less than 2W.

Creative Commons Aucune (Tous droits réservés) Aucune (Tous droits réservés)
contenu protégé  Document accessible sous conditions - vous devez vous connecter ou vous enregistrer pour accéder à ou acquérir ce document.
- Accès libre pour les ayants-droit

P a g e 1 | 6 Combining GPS based Precise Timing and Accurate Navigation requirements, benefits Emmanuel Sicsik-Paré (1) , Gilles Boime (1) , John Fischer (1) (1) Spectracom – Les Ulis, France ; Rochester NY, USA Emmanuel.sicsik-pare@spectracom.orolia.com Gilles.boime@spectracom.orolia.com John.fischer@spectracom.orolia.com 1 Astract It has been a continuous trend that all aerospace and payload programs require more and more parameters to be measured during flight tests, at increasing sampling rates. Contextual data, associated with measurements – like timestamp, geolocation, attitude - are instrumental to performing a relevant analysis of measured data. Therefore flight test teams, in charge of engineering high speed measurement systems, need to ensure proper time alignment amongst all on-board systems, facing two challenges: - Distribute precise time (better than 1 µs), even in case of GPS loss, over the whole test mission duration - Distribute precise position and attitude, which are time-stamped consistently with the distributed time In this paper, we demonstrate how a combined position & attitude measurement sensor and precise time server can meet all “Positioning, Navigation and Timing (PNT)” needs for a complex observation payload like an on-board test system and a SAR imagery radar, with benefits in terms of architecture simplification, and overall performance increase in terms of time and attitude accuracy. We also review the benefits of associating a high stability clock with a GPS receiver, in terms of improvement of some GPS reception performances. Keywords: Position Navigation and Timing, GPS performance improvement, Interference Detection and Mitigation. 2 Introduction Position / attitude measurement instruments on one hand, time and frequency generation instruments on the other hand, have been traditionally very distinctive products or solutions, handled by different teams and specialists within companies or research institutions. As a “time based” positioning system, Global Navigation Satellite System (GNSS) in essence provides simultaneously a clock reference and the means to elaborate a position solution. GNSS generalized usage therefore introduces a dramatic opportunity for gathering together positioning and timing techniques. Thanks to the on-going miniaturization of all associated components (GNSS receiver, high performance frequency oscillator, inertial measurement unit), it is possible to integrate both position / attitude measurement and time / frequency generation functions in a single instrument, using GNSS signal as the common reference. On the timing side, the use of a high performance oscillator disciplined by GNSS combines the high short term frequency stability from the oscillator with the high long term stability from GNSS. It allows also to deliver low phase noise frequency signals which are important in many radio communications and radar applications. The implementation of network timing protocols, like NTP or PTP (IEEE-1588) provides an elegant way to transfer precise time through an IP network, thus avoiding the need for dedicated media (like IRIG). In addition, the timing distribution is immune to temporary GNSS loss, as the frequency oscillator, in holdover mode, is used to maintain the local timescale, with some long term drift - which is actually a key performance feature for the oscillator. On the position and attitude side, tight coupling between GNSS (in standalone or in differential correction mode) and IMU, also combines the short term “stability” from the IMU with the absolute accuracy from the GNSS reception, in order to provide navigation solutions that include all the parameters of interest: position, orientation, speed, rotation rate, acceleration, etc.. Temporary loss of GNSS is handled by the IMU, which maintains navigation solutions in dead reckoning mode – with a drift depending on the IMU performance. High dynamics can be captured by using a high sampling rate from the IMU. Such a combined approach provides benefits in terms of Size, Weight and Power (SWAP) as applications requiring both position/attitude and timing can access them through a single instrument, single antenna solution avoiding discrepancy resulting from separate sources. 3 Examples of applications requiring PNT In order to illustrate how a single instrument can efficiently provide all critical Position, Attitude and Timing information, we chose two applications, one in the Intelligence Surveillance Reconnaissance (ISR) area, the other one in the flying test bench area. P a g e 2 | 6 3.1 On-board test bench New aircraft or modernization programs require more and more data to be recorded and analyzed in view of qualification and certification. Distributed sensors are operating at increasing sample rates in order to capture transient phenomena or high frequency vibrations. Those data must be acquired and recorded in real time, with IP network topology being well adapted to cope with such large streams of data. Along with measurement data, contextual data are needed in order to perform relevant analysis. Timestamps provide time alignment of samples and allow to correlate measurements made by different sensors at the exact same time. Position, attitude (relative to the body frame), speed, and acceleration measurements can be used to directly determine relationships between the measured data and some of the flight envelop parameters. 3.1.1 Typical architecture PNT Sensor Sensor Proxy IP Recorder Legacy recorder IRIG B 1 PPS NTP, PTP Master On board IP LAN sensors NTP client PTP slave Sensor Fig 1: Typical timing architecture for an on-board timing system In a typical on-board flight test system, the measurement data are time-stamped by the recorder. The recorder timescale and clock is itself disciplined thanks to 1 pps and / or IRIG B signals for legacy recorders, and thanks to a Network Timing Protocol (NTP) client or Precise Timing Protocol (PTP - IEEE1588 v1 & v2) slave for recent recorders. The iNET standard recommends the use of PTP protocol, as the way to transfer precise time on an Ethernet network from a master clock to a slave clock, thanks to the exchange of PTP messages that contains ingress and egress message timestamps, allowing the PTP slave to adjust its clock to synchronize with the PTP master clock. Required Position and Navigation (PN) data are stored within the recorder along other sensor data, but at a lower rate (typically 1 to 100 Hz). 3.1.2 Timing and positioning requirement The time accuracy required for data time-stamping depends on the sampling frequency, but typically ranges from 100 ns to 10 ms. The appropriate techniques for time transfer can be summarized as below: Required time transfer accuracy Appropriate time transfer method 1 – 10 ms NTP (network) IRIG B AM (dedicated media) 10 µs - 1 ms PTP (network) IRIG B DCLS (dedicated media) 100 ns – 10 µs 1 pps (dedicated media) Fig 2 : time transfer methods according to required time-transfer accuracy Position and Navigation data are strongly application dependent. However, attitude measurement is a common requirement with heading accuracy ranging between 0.1° and 1°. 3.2 Synthetic Aperture Radar for imagery All recent analysis confirm the need for ISR capabilities, whether it’s on land, air, sea, and space. Synthetic Aperture Radar (SAR) is now a mature technology that provides imagery of vast ground or maritime zones along the trajectory of the vehicle. It’s an interesting complement to optical observation, thanks to its capability to see camouflaged objects through any weather. P a g e 3 | 6 Fig 3: SAR imagery radar principle and SAR image 3.2.1 Key stakes In a SAR radar, a synthetic antenna is created thanks to the straight movement of the vehicle. The antenna’s virtual length is roughly equal to the traveled distance during the signal integration period. This synthetic antenna is therefore very long, resulting in very good resolution along the movement axis. Like for optical observation, the important criteria for SAR performance are related to: - Resolution: ability to resolve an object of interest within several pixels, to allow reconnaissance (and identification) - Geometric conformity: a square on the ground must be reported as a square on the SAR image (without echo migration) - Contrast: ability to distinguish between objects that have small reflectivity difference It has been shown that these key features are adversely impacted by many PNT aspects: On the time & frequency side: - Slow frequency emitter instability: generates echo migration and decreases image conformity - Phase noise increases the post-correlation spurious level and tends to decrease the contrast On the navigation measurement side, if not properly measured and compensated for: - Longitudinal position variations impact geometric conformity - Longitudinal, transverse and vertical velocity components variations affect both geometric conformity as well as resolution - Transverse and vertical accelerations impact resolution The following example from [5] provides a numerical calculation of the constraints applicable on standard deviation for position, velocity and acceleration, in order to maintain: - Geometric conformity criteria: echo shifts by less than half a resolution cell - Resolution criteria: size of resolution cell increases by less than 10 % For a X band lateral SAR radar ( = 3 cm), 1 m resolution, -10° elevation, 0.5 s integration time: Geometric conformity criteria Resolution criteria Longitudinal move STD on longitudinal position < 0,5 m STD on longitudinal velocity < 2 ms-1 Transverse move STD on transverse velocity < 1,6.10-2 ms-1 STD on transverse acceleration < 1,2.10-1 ms-2 Vertical move STD on vertical velocity < 8.10-2 ms-1 STD on vertical acceleration < 8.10-1 ms-2 STD: standard deviation Fig 4: Requirements on navigation data accuracy for a SAR radar In addition, for vehicles (helicopters, drones) which have significant parasitic yaw and roll movements, it is necessary to adjust the antenna steering direction, based on attitude measurements. Following radar processing, the available SAR images (strip map, or focalized map) must be properly geo-referenced. Such referencing – a classical operation in surveying – requires to know both the position and attitude of the observation vehicle with the appropriate accuracy. P a g e 4 | 6 3.2.2 Typical architecture The below diagram shows the typical (simplified) architecture of a SAR radar, and the PNT information needed by each subsystem of the radar. Fig 5: PNT requirements for a SAR radar subsystems 4 State of the art PNT instrument Geo-PNT is an all-in-one box PNT instrument that provides: 1. Time & Frequency reference - Low phase noise, high stability frequency signal, based on either Oven Compensated Quartz Oscillator (OCXO) or Chip Scale Atomic Clock (CSAC) - Configurable pulsed signals, including 1 pps and IRIG B, referenced to UTC 2. Navigation solutions (serial or LAN interface) - Position, velocity, accelerations, yaw, pitch, roll, rotation rates The Geo-PNT, with internal MEMS Inertial Measurement Unit (IMU) can be configured to work in standalone mode or in RTK mode. Accuracies are provided in Fig. 6. Horizontal / vertical position Velocity Acceleration Attitude Roll, pitch / Heading Standalone 1.5 m / 2.5 m 0.1 ms-1 0.15 ms-2 0.2 ° / 0.5° RTK 0.05 m / 0.1 m 0.02 ms-1 0.1 ms-2 0.1 ° / 0.3° Fig 6: Geo-PNT position & navigation performances 5 Improving GNSS receiver operation thanks to a high performance oscillator Having a good oscillator obviously contributes to good timing performances (short term stability, phase noise) which are required by applications like flying test bench applications, as well as radar and other ISR applications. But it also contributes to the improvement of the GPS receiver performances. GPS reception requires that the receiver’s clock aligns on the transmitted satellite clock. This alignment needs to be initialized at receiver startup, but needs also to be maintained all along receiver operation. As most GPS receiver use a poor short term stability oscillator (typically a TCXO), clock adjustment of the receiver oscillator needs to be done for each, as clock error is one of the four variables to be calculated along with the three position components (of course, if the receiver is fixed at a well-qualified position - which is often the case for timing receivers - then a single satellite allows to discipline the receiver clock). P a g e 5 | 6 Fig 7: compared ADEV for GPS, and various types of oscillators Allan Deviation (ADEV) [1] is the widely used metric for clock stability characterization over different periods of observation (Tau). The lower the ADEV, the more stable the oscillator is. Fig 8 shows the typical ADEV as communicated by various oscillator providers (Rubidium : LPFRS from Spectratime, CSAC : SA45 from Microsemi, OCXO from Rakon). In addition, it shows ADEV of GPS recovered clock, as well as GPS disciplined rubidium clock (SecureSync from Spectracom); the latter illustrates the effect of disciplining, which combines the short term stability of the rubidium, with long term stability of GPS. It can be seen that, below a Tau of 100 s, GPS recovered clock is less stable than all oscillators. For Tau higher than 100 s, GPS becomes better than an OCXO; and slightly better than CSAC. It becomes better than rubidium only for Tau higher than 20 000 s. The use of a good quality oscillator, e.g. featuring good long term stability, provides interesting options for improving some receiver features, depending on the application. 5.1 Improvement of vertical position and velocity measurement accuracy Krawinkel et al, [2] from Erdmessung Leibniz Universität Hannover, created a receiver clock model, using real ADEV measurements of a few oscillators, which was then input to an Extended Kalman Filtering, as a way to determine the influence of clock process noise in code-based GPS single point positioning. This work concluded that vertical position and vertical velocity accuracies could be improved respectively by 58 % and 66 % respectively when using a rubidium oscillator. 5.2 Integrity monitoring Bednarz et al [3] from MIT, made some laboratory measurements and observed accuracy improvement of vertical position accuracy ranging from 34 to 44% using also an atomic reference. Going further, Bednarz proposes to use the good external clock reference (instead of processing received signal to extract clock error) in order to determine the three position parameters. Then extract the clock error from pseudo-range measurement, which is a good predictor of the vertical position error. By setting a range of acceptable clock errors, it is possible to establish a Vertical Protection Level (VPL), as the main input of a clock-aided integrity monitoring mechanism. Such integrity monitoring adapts to changing atmospheric conditions or multipath or other clock sources. P a g e 6 | 6 5.3 Multipath mitigation In a similar approach, Preston et al [4] developed a CSAC clock model, allowing to solve the three geometric coordinates parameters equation, relying on the atomic clock, thanks to only three satellites in view (which is appreciable in urban canyons for example). A suddenly growing GPS recovered clock error, extracted from pseudorange measurements, probably reflects the presence of multipath. The satellite affected by multipath can then be excluded from 3D position calculation, as a multipath-mitigation mechanism. It provides an Interference Detection and Mitigation (IDM) reliable timing source to improve confidence in estimated PNT solution that are computed within autonomous computation. 5.4 Three satellites operation We have seen earlier that clocking the GPS receiver with a high stability oscillator allows to perform GPS processing limited to solving the three position parameters, with only three satellites in view. This in itself is an interesting feature for applications where only a small portion of the sky is accessible, either momentarily (aircraft manoeuvers) or permanently (masked GPS antenna, canyons). 6 Conclusion In this paper, we have illustrated how applications like airborne radar surveillance, and test benches require accurate time and frequency as well as navigation data to be provided to their various sub-systems. System performances depend (amongst others) on the accuracy of PNT data. With Geo-PNT, Spectracom offers a solution which combines both timing & navigation within a single enclosure, easing the integration of this function as well as optimizing its Size Weight and Power. In addition, combining a GPS receiver with a high stability oscillator can contribute to the improvement of GPS reception and IDM mechanism. Depending on application requirements, a high performance external clock – OCXO, Rubidium or CSAC - can be used to either increase vertical position and velocity accuracy, to implement clock based integrity monitoring mechanism – including multipath mitigation, or simply to increase GPS reception reliability when a limited number of satellites can be viewed. 7 Acknowledgment The present study has been solely possible thanks to great material provided by Geodetics Inc. and warm advice of Dr. Jeffrey A. Fayman, Vice President of Geodetics Inc. 8 References [1] D. Allan, “Time and Frequency (The Domaine) characterization , Estimation and Prediction of Precision Clocks and Oscillators”, vol 34, pp647-654, 1987 [2] T Krawinkel and S. Schön, “Applying Miniaturized Atomic Clocks for Improved Kinematics GNSS Single Point Positioning”, proceedings of 27th International Technical Meeting of the ION Satellite Division, Tampa, Sept. 8-12, 2014 [3] Sean G. Bednarz, “Adaptative modeling of a Global Positioning System Receiver clock for integrity monitoring during precision approach”, thesis, Massachusetts Institute of Technology, 2004 [4] Sarah E. Preston and David M. Bevly, “CSAC-Aided GPS Multipath Mitigation”, Proceedings of 46th Annual PTTI, Boston, Dec.1-4 2014 [5] Jean-Philippe Hardange, Philippe Lacomme, Jean-Claude Marchais, « Radars Aéroporté et Spatiaux », Masson, Sept 1995.

Creative Commons Aucune (Tous droits réservés) Aucune (Tous droits réservés)
contenu protégé  Document accessible sous conditions - vous devez vous connecter ou vous enregistrer pour accéder à ou acquérir ce document.
- Accès libre pour les ayants-droit

ETTC 2015– European Test & Telemetry Conference Optimizing Bandwidth in an Ethernet Telemetry Stream using a UHF Uplink Gonzalez-Martin, Moises Rubio-Alvarez, Pedro (moises.gonzalez@airbus.com , pedro.r.rubio@airbus.com ) Abstract: A conventional Flight Test demands a big amount of Real Time Data Information. Nowadays is natural to send Video Signals and Data in the same telemetry Stream with a limited bandwidth. The big challenge for the Flight Test Telemetry Engineer is being able to select the set of information needed for each test because is frequent need to transmit more information than bandwidth available on the telemetry stream. Up to now, it was a regular task to define a static set of information for each group of tests. That implies rebuild the Telemetry Stream, change the Acquisition Mapping and perform the telemetry checks in order to verify all the information needed will be transmitted. The solution developed by Flight Test Spain allows choosing, On Ground and in real time, which information is needed during a Flight Test Execution and transmitting it to the Airplane dynamically by means of a UHF uplink. Whenever a request is made from On Ground it produces an On Board Telemetry Stream with the new set of information to be transmitted. Keywords: Telemetry, Uplink, FxS, Packetizer, Bandwidth, IENA. 1 Introduction The flight test community faces issues with insufficient bandwidth available to support telemetering requirements. The amount of spectrum available for aeronautical telemetry is inadequate today, and demand is growing exponentially. Aeronautical telemetry is used to transmit real-time data during flight tests, and the availability of such data is integral to the productivity and safety of live flight test programs. Enough telemetry bandwidth is critical to maintaining rigorous system testing. Each country manages its own electromagnetic regulations. In some cases Technology research initiatives offer the prospect of increasing the bandwidth efficiency and, if they reach their intended capability, may partially offset telemetry spectrum demand until more spectrum access may be secured. 2 Technical Background In recent years there has been a shift from proprietary and closed solutions for Flight Test Instrumentation (FTI) networks towards more open standards-based systems using Ethernet technology. The trend towards Ethernet is further driven by the CTEIP Integrated Network Enhanced Telemetry (iNET) initiative that is pushing the adoption of Ethernet technology for the future of FTI[1]. Today there are some technologies based on Ethernet as transport protocol. Flight Test Spain is focused in these running technologies for the future Flight test Instrumentation: The concept of iNET[5] is to use internet-like architectures (Transmission Control Protocol/Internet Protocol (TCP/IP), Space Communications Protocol Standards (SCPS), and Consultative Committee on Space Data Systems (CCSDS)) to form a wireless network to supplement point-to-point telemetry capabilities. While some critical/safety data will always need a dedicated point-to-point reliable link, a significant portion of the data may be more efficiently handled by a network topology. iNET is currently in the architectural definition phase. iNET is a huge project (The economic model estimates that cost impacts of inadequate telemetry spectrum at a test range complex over a twenty year period will range from almost $23 billion) that is still in a previous phase. iNET-X[4] is the complete framework for iNET test articles that has been developed around the core recommendations and technologies outlined in the iNET standard ensuring interoperability for instrumentation networks. iNET-X extends the iNET standard ensuring high performance, network coherency, ease of setup, and management. IENA is another protocol widely used in Airbus for flight test Instrumentation. Airbus Defence & Space Flight Test Spain uses in the standard instrumentations Flight Test – Airbus Defence&Space, Avd. John Lennon s/n, 28906 Getafe (Spain) ETTC 2015– European Test & Telemetry Conference due the flexibility and the easy adaptation to Ethernet UDP protocol. 3 Ethernet acquisition and packetizer bus monitor, positive and negative aspects. Packetizer bus monitors are designed for networked data acquisition systems where the acquired data from the avionics buses is captured and re-packetized in Ethernet frames for transmission to an analysis computer or network recorder. The packetizer bus monitor encapsulates all messages on the bus and packages the message in the payload of a UDP/IP packet. The application layer contains bus identifiers, sequence numbers and timestamps. The packetizer mode has clear advantages between the parser mode because there is no need no select messages in the bus. Packetized mode is a very easy mode to program an acquisition system. No configuration error and the programming of the system is universal (no need to reprogram the system depending on the user information requirements or any change in the bus messages information). The drawback of this mode is the increase of the bandwidth in the acquisition stream due to all the messages are capture. But, as stated above, it is a great benefit not having to select any parameter during the elaboration of FTI Map. 4 FxS Protocol as parameter request concentrator FxS[2] (Flight Test Data Exchange Service), a platform independent protocol for the transmission of data between clients and servers within a local area network (LAN). The use of FxS Server-Client architecture allows to the server to know in real time what are the parameters and messages that all the users are demanding in a certain time of the flight Test. Picture 1 FxS Server Client Connection Protocol FxSIENA Server is an implementation of the FxS Server protocol and IENA[3] protocol as FTI transport layer. FxSIENA receives the IENA Data coming from the telemetry downlink and servers this information in real time to the FxS Clients connected in real time to the telemetry station. FxSIENA Server has been modified to collect all the information and send dynamically to the software responsible of Upload the parameter list through the UHF Uplink. FxSIENA Server is also responsible for the bandwidth control. The FxSIENA receives a parameter indicating the current used bandwidth and notify to user the bandwidth utilization. The protocol is based on a client server communication. When a on ground client demands a new parameter to be monitored, the new telemetry message is serialized and sent through the UHF Uplink. On Board, the UHF Receiver decommutes the frame and regenerates the list of parameters. After that the Software in the Telemetry Gateway is reconfigured to filter the new set of parameters. The telemetry message is, essentially, the list of parameters to be sent. Additionally, it can contain information of how to identify the message in the acquisition stream. Uplink Request Telemetry Result Picture 2 Communication Process On Ground On Board Transmission Period 1 sec ETTC 2015– European Test & Telemetry Conference Picture 3 Telemetry Message with acquisition information Telemetry Message can be significantly large (for a list with 10000 messages the telemetry Message is 100 Kb). For that reason, this can be compressed or partially sent using a delta mechanism. That means to send only the list of parameters to be added or removed since the current operating list. The confirmation check can be added using a standard error detecting code. For that case, has been tested using CCITT CRC-16. Data processing configuration is the same in both on ground and on board. This is due to the telemetry gateway software only filters messages inside the IENA packet maintaining the header and resizing the payload according the messages requested. Picture 4 IENA payload before and after passing through telemetry stream 5 UHF Telemetry Uplink UHF transceiver has been tested for use as telemetry uplink. The equipment incorporates a narrowband AM/UHF/VHF receiver and a FSK decoder. The demodulation is done in amplitude. It will integrate the FSK decoding circuit and restore a frame with a baud rate set to 1200 bauds. On Board the Uplink receiver can be connected to the UHF/VHF Aircraft radio. 6 Joining the chain links If the set of parameters needed during a flight test change during the flight Test, the crux of the matter is to change the telemetry stream dynamically according the real time requirements. Behind the idea there is an Acquisition System based in packetizer buses, an FxS Server/Client architecture, a UHF link for communications between Ground and On Board applications and a small process who decides which information to send through telemetry stream. Picture 5 Telemetry Gateway General Architecture 7 Proof of concept The pair receiver/transmitter was tested during the A400M Flight Tests where there was a predecessor system designed for manage predefined lists. The lists were elaborated On Ground (previous to perform the flight) with the requirements of the specialists. This way of work had some drawbacks: - Mistakes in the elaboration of the lists cannot be solved during the flight. For example, if one parameter is missing or is suddenly needed in the list, there is no possibility to add this parameter to any list On Board. - It is necessary to make a previous analysis of the parameters needed for each test point of Flight Tests. Sometimes this task can be extremely cost and difficult to perform in the analysis phase. No coverage tests were performed but the theory states the coverage range is above telemetry coverage. Once the HW part had been tested, a software test bench was developed for test the behaviour. Below there is the list of software components used in the test bench: On Board Simulation process: FTPlayer to simulate the acquisition stream (On Board) based in IENA Packets. The Telemetry Upload Message to filter (passes every ARINC429 message but Message number 2) ETTC 2015– European Test & Telemetry Conference simulation uses a recorded Flight in pcap format (packet capture). NetFxS to simulate a client to request dynamically parameters to FxS Server. FxSIENA Server receives IENA packets and sends request parameter to the client. Telemetry Gateway receives the list from the UHF receiver (using a RS232 connection) that filter the acquisition stream according the dynamic parameter list. On Ground Simulation process: NetFxS to simulate a client on ground to request dynamically parameters to FxS Server. FxSIENA Server that receives IENA packets from telemetry stream. This is a modified version that generates the list of requested parameters on ground and sends the list to the RS232 transmitter. A proof of concept has been successfully done on the lab. Test results managed to send a 100 Kb of information between transmitter and receiver for a 10 Mbits acquisition stream and 10000 parameters requested, producing up to 2 Mbits of filtered telemetry stream. 8 Next Steps Once the system has been tested in the lab, next stage is to test the system in a Flight Test Bed. Over the next year, a C295 aircraft will be equipped with the HW and software to test the complete system. The system will be improved with the following capabilities: Bandwidth control. The FxSIENA will control and notify that the maximum bandwidth limit is not exceeded. Improvement of the Uplink communication protocol (compression, retransmission and checksum). 9 Conclusions While the Flight Test community is waiting to next generation of Telemetry leaded by iNET project, Airbus Defence & Space Flight Test Spain is exploiting the maximum capabilities of the last Ethernet based technologies. With this improvement it will make possible to achieve more flight test efficiency, to have a better response time to telemetry requirements and overcoming Bandwidth constraints. 10 References [1] Toms Grace, “Telemetry of the Future” [2] Michael W. Dillard, “FXS – A bridge between Worlds”, Society of Flight Test Engineers 2004 [3] MARTIN S., “IENA & Ethernet Format Overview” Internal Airbus Group Distribution. [4] CWC-AE, White Paper “Packet header structures and payload structures for iNET-X application layer packetization protocols” [5] “iNET system architecture, version 2007.1.” Central Test and Evaluation Investment Program (CTEIP), July 2007 [6] Abdul Jabbar, Erik Perrins, James P.G. Sterbenz, “A Cross-Layered Protocol Architecture for Highly-Dynamic Multihop Airborne Telemetry Networks” 7. Acronyms FXS Flight Test Data Exchange Service LAN Local Area Network UDP User Datagram Protocol IP Internet Protocol IENA Installation d’Essai Nouveaux Avions UHF Ultra High Frequency HW Hardware FSK Frequency Shift Keying AM Amplitude Modulation VHF Very High Frequency CRC cyclic redundancy check LAN Local Area Network FTI Flight Test Instrumentation iNET-X Extended iNET TCP Transmission Control Protocol

Creative Commons Aucune (Tous droits réservés) Aucune (Tous droits réservés)
contenu protégé  Document accessible sous conditions - vous devez vous connecter ou vous enregistrer pour accéder à ou acquérir ce document.
- Accès libre pour les ayants-droit

ETTC 2015– European Test & Telemetry Conference Flexible Switching for Flight Test Networks A. Diarmuid Collins1 1: Curtis-Wright, Dublin Abstract: The network switch is a critical element in the flight test network. All devices in the network are configured, synchronised and managed via the switch. In addition to this all acquired data is routed through the switch. For these reasons, the flight test network switch has always needed to be rugged and reliable with high throughput and simple intuitive setup. Ethernet technology and the move towards open standards within FTI systems have enabled flight test networks to become increasingly flexible and heterogeneous. Modern FTI networks may have different synchronisation and data transmission protocols running simultaneously. It is also important to quickly switch network configurations for different flight profiles and to enable new features to be easily added to existing installations. This paper examines the increasing network interoperability and flexibility challenges and discusses how the network switch is best placed to provide solutions. Keywords: Ethernet, switching, FTI, PTP, SNMP 1. Introduction In Flight Test Instrumentation (FTI) as the acquired volume of data increases, the industry is migrating from IRIG 106 chapter 4 PCM to Ethernet networks. Ethernet has a long history in the commercial and industrial markets. Since the initial definition in the 1970s to the first agreed IEEE 802.3 standard in 1983, Ethernet has grown both in commercial market size to a multiple billion dollar market and the technology has developed to be capable of transferring data rates in excess of 100Gbps. Using Ethernet in FTI networks brings a number of significant advantages: • Wide range of off-the-shelf commercial Ethernet products, from switches, network interface cards, recorders amongst other equipment • Mature standards build around Ethernet for the transmission of data, the configuration of networking equipment, synchronisation of network elements. • Wide range of software, both commercial and open sourced, for interfacing and manipulating Ethernet data and equipment. • Scalable network infrastructure and data rates from 10Mbps to 100Gbps with a future path to higher rates. FTI networks have a number of requirements that necessitate specific consideration and place constraints on Ethernet networks. Traffic on FTI networks tend to be heavily asynchronous, that is to say that the data rates and volume of traffic on an FTI network in one direction are far higher than in the opposite direction. Determinism and loss-less transmission are two highly desirable features in an FTI network. To ensure the transmission and recording of all the acquired parameters, packet loss on the network is not acceptable, regardless of the network layer or application layer protocol being used. In many commercial implementations, the reliability of the data transfer is handled by the transport layer, requiring retransmission for lost packets. Figure 1: FTI network elements 2. The Network Switch The core of an Ethernet network is the switch. An Ethernet switch may operate at one of more layers of the OSI network model [1]: • Layer 1. The lowest layer switch is known as a repeater or a hub. It is simple device which does not manage the traffic through the device. • Layer 2. A network bridge which switches Ethernet packets based on MAC addresses. • Layer 3 / 4. Commonly known as routers. Switches network traffic based on IP, TCP, UDP and application layer data. FTI network switching typically requires layer 3 and 4 switching, at a minimum where traffic is routed and ETTC 2015– European Test & Telemetry Conference switched based on UDP ports, IENA [2] and iNET-X [3] stream identifiers. 3. Switch Design COTS Ethernet switches and switch cores support a wide range of features and requirements driven by commercial Ethernet networks. Dynamic switching and self-learning of network topologies are required to support dynamic and changing networks in benign environmental conditions. FTI networks on the other hand may not require all of these features but instead need to be rugged and very reliable. Ruggedness is dictated by a number of environmental standards, specifically DO-160 [4] and MIL-STD-704 [5]. These two standards define a minimal set of environment test conditions, covering: temperature, humidity, shock, vibration, power interface to the aircraft, among others. Network switches can be implemented primarily using two approaches. Application Specific Integrated Circuits (ASIC) can be implemented to perform switching and configuration of the network switch, usually with on chip microprocessors (MCU). These MCUs run management and configuration firmware on an RTOS or even embedded OS such as Linux. ASIC development is very expensive undertaking and as a result, a very limited number of large companies such as Marvell [6] and Intel design very flexible switching products which are then sold off the shelf. OEM manufactures integrate these products, customising the firmware to implement the feature set of interest for their product. Customisation to the lower level hardware is not possible without commercial justifications in the hundreds of millions of dollars range. A second approach is implementing the switching and management functionality in Field Programmable Gate Arrays (FPGA). FPGAs have the advantage of much shorter and cheaper development cycles, with some trade off in the volume of supported features. The switch manufacturer can design the feature set of interest for their product line and exclude the unwanted functionality that the more general purpose ASIC switch cores support. For an ideal FTI switch the latter approach has significant advantages. Within the FPGA, a store and forward switch fabric can be implemented using state machine based code. Dynamic learning algorithms for routing and on board OS are not required due to the more limited set of requirements reducing the time from power up to operation, simplifying the design and consequently, increasing the reliability of the switch. The static forwarding and filtering configuration, stored in on-board non-volatile memory, allows the switch to start routing based on a pre-defined set of rules as soon as power is applied. As an example, the NET/SWI/101 from Curtiss-Wright powers-on, achieves link up and is transmitting within 2 seconds [7]. In certain FTI networks, the ability to tap an Ethernet link for monitoring purposes can be a very useful feature. Most switches can be configured to perform such functionality, however minimising the latency through the switch can be challenging. The FPGA based designs can be configured to bypass the core keeping latency to the minimum for “tap”-like performance. 4. Time Synchronisation in FTI Network Ethernet networks support a number of well-defined and supported time synchronisation protocols. The two most widely known and used are Network Time Protocol (NTP) [8] and Precision Time Protocol (PTP) 4.1. NTP NTP is a time synchronisation protocol widely used to synchronise desktop computers on packet switched networks, most famously the Internet. Sub second accuracy is possible, with simplified implementations known as SNTP also available. The accuracy is good enough for consumer applications. 4.2. PTP PTP is an IEEE standard used to synchronise clocks in a network, using similar principles to NTP. Unlike NTP, it was designed to achieve sub-microsecond accuracy. This accuracy makes it more suitable to FTI networks than NTP. The original standard was agreed in IEEE 1588- 2002 [9] and is known as PTPv1. The second revision of the standard was agreed in IEEE 1588-2008 [10], improving accuracy precision and robustness. However PTPv2 is not backward compatible with PTPv1. 4.3. PTP in FTI Networks Synchronisation of all data acquisition units in an FTI network is a key requirement. The time correlation of the data on the network is a function of the synchronisation accuracy. Clearly the time synchronisation protocol of choice is PTP. This raises the requirement for the support of a number of PTP related features in the ideal FTI switch. PTP Grandmaster In a PTP-synchronised network, one element in the network acts as the master to all the time slaves, this is the Grandmaster (GM). The GM acquires time from an external time source such as GPS, IRIG Analog and Digital or a battery backed Real Time Clock (RTC) and synchronises the slaves to this time source. With non- backward compatible standards in PTPv1 and PTPv2, support for both grandmasters is required PTP Transparency ETTC 2015– European Test & Telemetry Conference In a larger network where the switch is not a PTP grandmaster, to improve on the synchronisation accuracy, the switch should appear invisible to the PTP conversation. This is known as PTP transparency. The propagation time of the PTP packets through the switch is measured and the timestamps are adjusted accordingly, removing the propagation delay. Support for PTP transparency is required in both PTPv1 and PTPv2 modes of operation. Bridging of PTP Protocols With non-backward compatible protocols, it is not uncommon for network devices supporting either PTPv1 or PTPv2 to co-exist on the same network. Ensuring that these PTP clients are synchronised to the one time source requires the FTI switch to support the translation or bridging between the two protocols. This is a very powerful and useful feature allowing the network designer to mix clients comfortably on the one network. Figure 2: Mixed PTP clients in one network with translation Multiple Time Sources While PTP is the time synchronisation mechanism on the network, the absolute time needs to be acquired by the Grandmaster to allow for accurate absolute time synchronisation. Historically IRIG-B was a standard created by the US military defined in 1960, the latest revision of the standard published in 2004. This standard is widely supported in FTI networks both in analog and digital formats. The Global Positioning System (GPS) is another very popular space-based location and time synchronisation system. If the FTI switch supports GPS, it allows the time to be synchronized to the satellite based atomic clocks. This is a very accurate and cost effective mechanism for acquiring absolute time. 4.4. Time Synchronisation in the ideal FTI Switch As described, there is a long list of time synchronisation features that the ideal FTI switch should support, to a high level of accuracy. Acting as a PTP v1 or v2 GM, taking time sources from GPS, IRIG or free running from an on- board RTC. The bridging of PTP protocols allows the FTI engineer to define a per port PTP protocol selecting between v1 and v2. 5. Traffic Filtering in an FTI Network As previously mentioned, switches typically route data based on a certain level of the OSI model. In a level 2 switch, the Ethernet MAC address are used to automatically route traffic to ports on which that particular networking interface is connected. A level 3 or 4 Ethernet switches can route traffic based on IP address or UDP ports as an example. FTI networks are heavily asymmetric with a large number of sources but a limited number of sinks. These sinks may have very different requirements. Network recorders typically have a very large bandwidth and storage space so will generally record all the traffic on the network for later analysis and archiving. As a result the network switch will generally route all traffic to ports on which the record is connected, filtering none of the traffic. In certain applications it may also be desirable to separate certain high volume traffic to a dedicated recorder. One example of this may require all video traffic to be recorded on a dedicated recorder. Such selective switching could be implemented using a dedicated multicast IP address for video traffic. The FTI switch is then required to switch this traffic to the dedicated video recorder. Transmitters on the other hand have a very limited bandwidth but give engineers on the ground very valuable insight into key information on the FTI network. In this scenario the switch is required to filter based on very specific parameters from the Ethernet traffic. In IENA traffic, a specific stream identifier in combination with a UDP port may contain parameters of interest. The FTI switch therefore requires the ability to switch based on multiple header fields at all layers of the OSI model. ETTC 2015– European Test & Telemetry Conference Figure 3: Dynamic routing in a configurable crossbar An on-board data processing unit connected to the network would have similar requirements to the transmitter, in that it could only process a subset of the traffic during the flight. However in addition to filtering the traffic, the required switching configuration could change during the flight to allow the engineer to perform analysis at different phase of the flight. For example on take-off, the switch could be configured to pass iNET-X stream identifiers in the range 0x1 to 0xF to the data processing unit, then at altitude, filter this traffic, allowing all traffic on UDP port 4444. The FTI switch therefore, should have a rich set of filtering and switching functionality built into the switch core. A standard set of switching based on layer 2 and 3 header fields should be supported. In addition to this, it is desirable that custom switching rules can be implemented at the application layer. Even more powerfully, the engineer could define fields within the payload of the packet and filter based on these values. This level of flexibility results in a very powerful switch. All these filters should be stored on the switch in non-volatile memory, in an efficient lookup table to allow the traffic to be filtered at line speed through the switch, avoiding any bottlenecks in the data path. 6. Configuration of Network Switches With the expanding features and configuration options available on FTI switches, ease of configuration is more important than ever. The configuration of such devices can be implemented either proprietary configuration software or open standards which have been adopted for such purposes. TFTP [11] is a file transfer protocol that has been developed specifically for light weight file transfer. The server can be implemented with a minimum of CPU and RAM requirements making it suitable for embedded devices. For the transfer of large configuration binary files, it is a widely adopted protocol used on switches. Simple Network Management Protocol (SNMP) [12] is an open internet standard for managing devices on an IP network. It is typically used to monitor and configure switches and recorders. It is a self-documenting protocol that is used to configure smaller volumes of configuration data. Off the shelf SNMP managers are widely available for Windows, Linux and OSX operating systems which can then be used to manage the networked devices. This configuration phase itself can be split into two distinct phases, dynamic, on the fly configuration and static configuration prior to acquisition. FTI network topologies are generally relatively static and as a result prior to flight, these networks can be defined and the switches configured with the routing and filtering tables. With a broad range of options available, the configuration at this point can be significant with settings for different filtering options to be setup for a number of phases of the flight. The configuration would ideally be stored in a local setup file on the engineers PC, to make iterative changes to the network configuration simple. An example of such a file format is XidML (eXtensible Instrumentation Definition Markup Language). [13] Figure 4: Simple crossbar configuration Once this configuration has been completed, programmed and stored locally the switch is configured and ready for flight. Once in flight, this static configuration will not be modified, however now the dynamic configuration aspect takes place. The FTI engineer may want to monitor network traffic, link status and the health of the network as well as switch, between the various phases of the flight. SNMP managers running on PCs connected to the network can select between the various configurations that were pre-configured, in a seamless manner, with little or no packet loss. Monitoring using an SNMP manager allows the FTI engineer to query the health of the network switch however, it can be useful to have automatic or passive health reporting. Such a facility would allow the switch to periodically report on various metrics in a status packet. This status packet could easily be telemetered to the ground as well as recorded. The advantage of such approach is that a query/response mechanism is not ETTC 2015– European Test & Telemetry Conference required, which in many telemetry links is not possible, and the information density of a status packet makes it a very efficient use of the limited telemetry bandwidth. An example FTI switch, the NET/SWI/101 has 254 configurable filters per “mode”, which can filter traffic based on 8x16bits fields in both the header and payload of the Ethernet traffic. Each “mode” represents a selectable mode of operation, of which there are 16. This allows the FTI to design a very powerful set of configurations for their network and dynamically switch between pre- defined configurations using SNMP. 7. Future-Proofing Network Switches Over time, as the size and complexity of airborne networks continues to increase, the demand for additional features and performance upgrades continues. Some of these upgrades will require replacing existing hardware in the instrumented airplane, however there is significant scope for FPGA-based designs to incrementally upgrade the programmed “firmware” of the FPGA. This mechanism allows the user to remotely upgrade the feature set of the switch without physically removing or even accessing the switch. The upgrade process can be implemented in a similar mechanism to the static programming of the device over the Ethernet interfaces using TFTP. Such an upgrade could feasibly be executed in minutes between flight tests, if the demand arose. Naturally, such a process has potential to be interrupted so significant effort and measures need to be taken to ensure that the process cannot result in a non-working or unusable switch. “Fall- back” firmware images are untouched on the switch to ensure any interruptions in the programming cycle do not result in a non-working switch. 8. Conclusion FTI network switches, which sharing some commonality with COTS Ethernet switches have specific demands of their own. The relatively static nature of FTI networks in combination with stringent reliability and rugged requirements places specific demands that many switches cannot meet. Flexible switching and filtering requirements, advanced time synchronisation mechanism and rugged, deterministic and scalable performance are currently met by the Curtiss-Wright NET/SWI/101 airborne switch, among others. 9. References [1] ISO/IEC, “Open Systems Interconnection”. [2] F. Abadie, “A380 IENA Flight Test Installation Architecture,” ETTC, 2005. [3] Curtiss-Wright, “iNET-X Packet header structures and payload structures”. [4] Radio Technical Commission for Aeronautics (RTCA), “DO-160F Environmental Conditions and Test Procedures for Airborne Equipment,” 2007. [5] Department of Defense, “Aircraft Electric Power Characteristics”. [6] Marvell, “Marvell Prestera 98DX2101,” [Online]. Available: http://www.marvell.com/switching/assets/Marvell_Pre stera_98DX21xx-41xx-001_product_brief.pdf. [7] Curtiss-Wright, “NET/SWI/101 Datasheet,” [Online]. Available: http://www.cwc- ae.com/product/netswi101. [8] Internet Engineering Task Force, “Network Time Protocol Version 4: Protocol and Algorithms Specification,” [Online]. Available: http://tools.ietf.org/html/rfc5905. [9] 1588_WG - Precise Networked Clock Synchronization Working Group, “1588-2002 - IEEE Standard for a Precision Clock Synchronization Protocol for Networked Measurement and Control Systems”. [1 0] 1588_WG - Precise Networked Clock Synchronization Working Group, “1588-2008 - IEEE Standard for a Precision Clock Synchronization Protocol for Networked Measurement and Control Systems”. [1 1] IETF, “RFC1350 - THE TFTP PROTOCOL,” [Online]. Available: http://tools.ietf.org/html/rfc1350. [1 2] IETF, “ Management Information Base for Network Management (RFC 1213),” [Online]. Available: http://tools.ietf.org/html/rfc1213. [1 3] “Welcome to xidml.org - a website for the XidML community,” [Online]. Available: http://www.xidml.org/. 10. Glossary NTP Network Time Protocol PTP Precision Time Protocol FTI Flight Test Instrumentation ASIC Application Specific Integrated Circuit FPGA Field Programmable Gate Array TFTP Trivial File Transfer Protocol COTS Commercial Off The Shelf CPU Central Processing Unit RAM Random Access Memory GM Grand-Master RTC Real Time Clock SNMP Simple Network Management Protocol iNET Integrated Network Enhanced Telemetry iNET-X iNET-Extended

Creative Commons Aucune (Tous droits réservés) Aucune (Tous droits réservés)
contenu protégé  Document accessible sous conditions - vous devez vous connecter ou vous enregistrer pour accéder à ou acquérir ce document.
- Accès libre pour les ayants-droit

Evolving Embedded Electronics Testing in HIL Simulation and Large- Scale Test Cells Through Sub-ns synchronization systems via Time Sensitive Networks in Ethernet Veggeberg, Kurt1, Daurelles, Olivier2 1: National Instruments, 11500 N. Mopac Expwy, Austin, TX 78759 USA 2: National Instruments France, 2 rue Hennape 92735 France Abstract: Time Sensitive Networks in Ethernet are improving computer based measurements from sub- microsecond to sub-nanosecond precision. By using message based protocols that synchronize clocks, it is possible to generate this required precision signaling locally. This solution is scalable from within the same node to multiple nodes throughout the world. To fulfill the more demanding needs of test & measurement applications, IEEE 1588 (PTP) has been developed, which is able to provide sub-microsecond performance. Much better performance is becoming possible. Keywords: Ethernet, IEEE-1588.2, IEEE 802.1, White Rabbit, CERN 1. Introduction Many of the research activities concerning IEEE 1588 have been targeted at Ethernet. Already, CERN is able to achieve sub-ns synchronization in the timing, trigger and control system for the Large Hadron Collider (LHC), the biggest machine in the world based on Ethernet. This is an example of what Time Sensitive Networks and standards such as IEEE-1588, IEEE 802.1 will be bringing to COTS Ethernet. This allows time sensitive and best data effort to coexist. The key ideas of the White Rabbit (WR) technology used by CERN can be adapted and included in the next revision of PTP (Precision Time Protocol)enabling the standard compliant PTP devices to achieve high accuracy synchronization using methods prototyped and tested in WR.. 2. Time Sensitive Networks (TSN) are revolutionizing monitoring, control & test applications Measurement and automation systems involving multiple devices often require accurate timing in order to facilitate event synchronization and data correlation. To achieve this synchronization, devices in the system must either have direct access to timing signals from a common source, or the devices must synchronize their individual clocks in order to share a common time base. There are advantages and disadvantages to both methods of device synchronization. In a time based synchronization scheme for data acquisition, we still have to provide the same timing signals as in a signal based scheme, but the way in which we do so is different. All of the timing signals, triggers and clock, are based off of a common time reference. Examples of time references are GPS, IRIG-B, 802.1 and IEEE-1588. Devices on the network, including switches and routers, can be synchronized very precisely via the IEEE 1588 and IEEE 802.1 "precision time protocol" standards. Some examples of applications where this type of synchronization can prove extremely valuable are in both large, distributed hardware-in-the-loop (HIL) test systems and test cell measurement systems, especially in industries like aerospace and defense. Often times, an ‘iron bird’ is created where the full electronics inside a plane are tested by simulating the environment to make them think they are actually in an operating plane. When performing this type of HIL test, the large electronic system of the plane can be broken down into several different components representing separate subsystems performing different functions. For example, a plane’s flaps, slats and rudder engines can be treated separately and in order to provide accurate test results, distributed processing power over the area of the plane may be required. This increase in computational power can then enable efficient testing by aiding in the execution of large, complex simulation models, but these architectures may also be needed to allow for testing of high-channel count systems in application areas like structural tests of wings. Depending upon how the system is set up, it can be necessary to provide shared trigger and timing signals between nodes while also requiring deterministic data sharing. Each subsystem may generate a local reference clock signal (local to the subsystem), Which may be aligned and locked With respect to one or more similar respective reference clock signals of other subsystems, via a high- level precision time protocol (PTP) such as IEEE-l588 or a global positioning system (GPS) protocol. For instrumentation systems, each DAQ card (i.e. device) within a given subsystem may generate a local sample clock (local to the DAQ card) based on the local reference signal, and generate a local trigger clock (local to the DAQ card) based on the local sample clock. The trigger ETTC 2015– European Test & Telemetry Conference clocks may be synchronized with respect to each other, and each DAQ card may then use its trigger clock to synchronize any received trigger (or trigger pulse), resulting in received triggers being synchronized across all participating DAQ cards across all participating subsystems. Sharing a common timing signal becomes unfeasible when the distance between devices increases, or when devices frequently change location. Even at moderate distances such as 50 meters, a common timing signal may require significant costs for cabling and configuration. Up to 100 meters you can match trace length. Once you can’t, you lose sync at 1.5 ns/ft so over 200m you can lose approximately 900 ns of sync . IEEE-1588 can meet 200m through only a single switch (single subnet) with what we have today. 3. Using message based protocols allows synchronization of clocks by compensating for path delays In general, distributed measurement and control systems often require their composite parts to be aligned to the same timebase. One useful result of synchronization in these applications is the sharing of synchronized periodic signals, which can be used to take measurements at the same time or to provide known relationships between control units in a distributed environment. In these situations, distributed clock synchronization becomes necessary. Using this approach, devices act on timing signals originating from a local clock which is synchronized to the other clocks in the system. Examples of distributed clock synchronization include devices synchronized to a GPS satellite, a PC’s internal clock synchronized to an NTP time server, or a group of devices participating in the IEEE 1588 protocol. Instead of sharing timing signals directly, these devices periodically exchange information and adjust their local timing sources to match each other. The synchronization of distributed clocks requires a continuous process. A clock is essentially a two part device, consisting of a frequency source and an accumulator. In theory, if two clocks were set identically and their frequency sources ran at the exact same rate, they would remain synchronized indefinitely. In practice, however, clocks are set with limited precision, frequency sources run at slightly different rates, and rate of a frequency source changes over time and temperature. Most modern electronic clocks use a crystal oscillator as a frequency source. The frequency of a crystal oscillator varies due to initial manufacturing tolerance, temperature and pressure changes, and aging. Because of these inherent instabilities, distributed clocks must continually be synchronized to match each other in frequency and phase. By using message based protocols that synchronize clocks, we are able to generate the required signaling locally. This solution is scalable from within the same node to multiple nodes throughout the world. Devices act on timing signals originating from a local clock which is synchronized to the other clocks in the system. Examples of distributed clock synchronization include devices synchronized to a GPS satellite, a PC’s internal clock synchronized to an NTP time server, or a group of devices participating in the IEEE 1588 protocol. Instead of sharing timing signals directly, these devices periodically exchange information and adjust their local timing sources to match each other. 4. Ethernet-IEEE 1588 Synchronizaton IEEE 1588 provides a standard protocol for synchronizing clocks connected via a multicast capable network, such as Ethernet. Released as a standard in 2002, IEEE 1588 was designed to provide fault tolerant synchronization among heterogeneous networked clocks requiring little network bandwidth overhead, processing power, and administrative setup. IEEE 1588 provides this by defining a protocol known as the precision time protocol, or PTP. IEEE 1588 is designed to fill a niche not well served by either of the two dominant protocols, NTP and GPS. IEEE 1588 is designed for local systems requiring accuracies beyond those attainable using NTP. It is also designed for applications where a GPS receiver at each node is too expensive, or for which GPS signals are inaccessible. 5. Using Time-Based Synchronization Slave clocks synchronize to the 1588 grandmaster by using bidirectional multicast communication. The grandmaster clock periodically issues a packet called a ‘sync’ packet containing a timestamp of the time when the packet left the grandmaster clock. The grandmaster may also, optionally, issue a ‘follow up’ packet containing the timestamp for the ‘sync’ packet. The use of a separate ‘follow up’ packet allows the grandmaster to accurately timestamp the ‘sync’ packet on networks where the departure time of a packet cannot be known accurately beforehand. For example, the collision detection and random back off mechanism of Ethernet communication prevents the exact transmission time of a packet from being known until the packet is completely sent without a collision being detected, at which time it is impossible to alter the packet’s content. ETTC 2015– European Test & Telemetry Conference Figure 1 : Using message based protocols allows synchronization of clocks by compensating for path delays The master periodically broadcasts the current time as a message to the other clocks. Under IEEE 1588-2002 broadcasts are up to once per second. Under IEEE 1588- 2008, up to 10 per second are permitted. Each broadcast begins at time with a Sync message sent by the master to all the clocks in the domain. A clock receiving this message takes note of the local time when this message is received. By sending and receiving these synchronization packets, the slave clocks can accurately measure the offset between their local clock and the master’s clock. The slaves can then adjust their clocks by this offset to match the time of the master. The IEEE 1588 specification does not include any standard implementation for adjusting a clock; it merely provides a standard protocol for exchanging these messages, allowing devices from different manufacturers, and with different implementations to interoperate. 6. Time Based Synchronization Accuracy Results Several factors affect the synchronization levels achievable using IEEE 1588 over Ethernet. During the time between synchronization packets, the individual clocks in a system will drift apart from each other due to frequency changes in their local timing source. This drift can be reduced by using higher stability timing sources and by shortening the intervals between synchronization packets. Temperature-controlled crystal oscillators (TCXOs) and oven-controlled crystal oscillators (OCXOs) provide higher stability than standard crystal oscillators, and atomic clocks provide still higher stability. In addition to stability, a clock’s resolution will affect the accuracy of the timestamps transmitted in the PTP synchronization messages. Devices that have a higher resolution clock are able to more accurately timestamp messages. Also, variations in network delay, caused by jitter introduced by intermediate networking devices such as hubs and switches reduce the achievable synchronization level. IEEE 1588 provides an important alternative for systems requiring sub-microsecond synchronization in geographically distributed systems. Figure 2: Time Based Synchronization accuracy results 7. Time Sensitive Networking brings deterministic Ethernet Ethernet, Wi-Fi and other IEEE 802-based network technologies have been very successful in a large number of connectivity applications, but until very recently, there was no way to provide critical time sensitive services in those networks. The result has been a proliferation of specialized networks and connectivity systems for audio/video and real time control applications. This lack of integration is in the process of being remedied by the creation of an IEEE 802 architecture for "time sensitive networking." It specifies a profile for use of IEEE 1588 for time synchronization over a virtual bridged local area network defining how IEEE 802.3 (Ethernet) and IEEE 802.11 (Wi-Fi), can all be parts of the same timing domain. based on three major advances: 1) Universal time synchronization - or "time awareness" by the network infrastructure. Devices on the network, including switches and routers, can be synchronized very precisely via the IEEE 1588 and IEEE 802.1 "precision time protocol" standards. 2) Time sensitive queuing and forwarding in all devices to provide lower, and guaranteed, delays for time-sensitive data. 3) Bandwidth and latency reservations so that the time- sensitive queues in the network do not overflow and packets are not dropped. 8. CERN’s Timing Triggering and Control System uses optical networks to achieve sub-ns synchronization The LHC at CERN is one of the largest and most complex systems ever built. CERN is a complex of 6 circular and some linear accelerators which are interconnected. The biggest accelerator is the Large Hadron Collider (LHC) which is 27 km long. All the devices which serve the accelerators (magnets, kickers, etc) need to be precisely synchronized and controlled by a central control system. CERN takes advantage of the FPGA to serve as the timekeeper in the NI RIO platform and uses it to move blocks of graphite in place to absorb the protons that are not in the nominal path of the beam or, in other words, go astray. This process is commonly known as “collimation.” ETTC 2015– European Test & Telemetry Conference Since this is a 27 km tunnel, there are more than 100 of these collimators around the tunnel that have to be synchronized accurately and reliably. In a given collimator, the PXI chassis run LabVIEW Real- Time on the controller for reliability and LabVIEW FPGA on the reconfigurable I/O devices in the peripheral slots to perform the collimator control for approximately 600 stepper motors with millisecond synchronization over the 27 km of the LHC. The timekeeper on the field- programmable gate arrays (FPGAs) on these devices give the level of control needed. A decision was made to base the new CERN timing system on PTP. The project is an Ethernet-based network with low-latency, deterministic data delivery and network-wide, transparent, high-accuracy timing distribution. The White Rabbit Network (WRN) extension is based on existing standards, namely Ethernet, Synchronous Ethernet and PTP. The approach aims for a general purpose, fieldbus-like transmission system, which provides deterministic data and timing (sub-ns accuracy and ps jitter) to around 1000 stations. It automatically compensates for fiber lengths in the order of 10 km. Figure 3 : Enhanced Ethernet with White Rabbit Synchronization and Determinism The White Rabbit Project focuses on an open design consisting of: Sub-nanosecond accuracy - Synchronization of more than 1000 nodes via fiber and copper connections up to 10 km apart. Flexibility- creating a scalable and modular platform with redictability and Reliability -allow deterministic delivery obustness- no losses of high prioritized accelerator clock signals over the Ethernet physical yer. This signal can then be made traceable to an oving jitter generated from the lock recovery circuitry before being feed to the evices. The timing aster is syntonized with an atomic clock and is vices may be connected s well, only White Rabbit devices take part in the timing ake has been erformed. Therefore if a non-WR aware device is conn nd GPS. IEEE 1588 is designed for local systems simple configuration and low maintenance requirements. P of highest priority messages. R device control messages. White Rabbit takes advantage of the latest developments for improving timing over Ethernet, such as IEEE 1588 (Precision Time Protocol) and Synchronous Ethernet. Synchronous Ethernet, also referred as SyncE, is an ITU- T standard for computer networking that facilitates the transference of la external clock. Network synchronization in WR is based on clock hierarchy with the highest accuracy clock at the top. Slave clocks are PLL-based, locked to an external reference that is being recovered by the data link. The PLL cleans the recovered clock by rem c transmitting device. The key ideas of the White Rabbit (WR) technology can be adapted and included into the next revision of PTP consequently enabling the standard compliant PTP devices to achieve high accuracy synchronization using methods prototyped and tested in WR. In a White Rabbit timing network, a timing master will provide the master clock. The timing master provides the general timing to be used by all attached White Rabbit d m synchronized with a GPS receiver. Components of a White Rabbit network are White Rabbit Switches and White Rabbit nodes. Both components may be added dynamically to the network. Though conventional Gigabit Ethernet de a by using the timing information. The switch is the core element of the WR network, implementing the standard IEE802.1x Ethernet Bridge functionality and WR-specific extensions. The extensions are enabled only after a proper WR handsh p ected, it sees a standard 802.1x switch. 9. Updating standards provide improved data acquisition and control with Ethernet Upcoming changes to the Ethernet standards can yield a giant leap in performance from nanoseconds to picoseconds. IEEE 1588 is designed to fill a niche not well served by either of the two dominant protocols, NTP a requiring accuracies beyond those attainable using NTP. ETTC 2015– European Test & Telemetry Conference ETTC 2015– European Test & Telemetry Conference ject Authorization Request (PAR) for the revision f the IEEE 1588-2008 Standard was approved on 4-June- work implementations and defines uch mappings, including User Datagram Protocol lues are known. The grandmaster can be ynchronized to a source of time external to the system, if he impossible, possible for recision test and measurement applications. As they are d by the makers of equipment, it should make it easier for end-users. To fulfill the more demanding needs of test and measurement applications, IEEE 1588 (PTP) has been developed, which is able to provide sub-microsecond performance. Many of the research activities concerning IEEE 1588 have been targeted at Ethernet. Research at CERN shows that clock synchronization between the master node, which provides the UTC reference clock, and the slave nodes, which synchronizes to the reference clock, is possible in the ns-range. Currently the P1588 working group is working on a new edition of IEEE 1588. The Pro o 2013 with an expected completion date of 31-December 2017. The standard specifies requirements for mapping the protocol to specific net s (UDP)/Internet Protocol (IP versions 4 and 6), and layer-2 IEEE 802.3 Ethernet. The protocol enables heterogeneous systems that include clocks of various inherent precision, resolution, and stability to synchronize to a grandmaster clock. The protocol supports synchronization in the sub-microsecond range with minimal network bandwidth and local clock computing resources. The protocol enhances support for synchronization to better than 1 nanosecond. The protocol specifies how corrections for path asymmetry are made, if the asymmetry va s time traceable to international standards or other source of time is required. These enhancements make t p incorporate

Creative Commons Aucune (Tous droits réservés) Aucune (Tous droits réservés)
contenu protégé  Document accessible sous conditions - vous devez vous connecter ou vous enregistrer pour accéder à ou acquérir ce document.
- Accès libre pour les ayants-droit

ETTC 2015– European Test & Telemetry Conference PTPv1 vs PTPv2: Characteristics, differences and time synchronization performances. Guillermo Martínez Morán (guillermo.m.martinez@airbus.com) 1 Abstract: Precise Time Protocol (PTP) based on Ethernet network packets is displacing more and more the traditional time synchronization based on IRIG-B. This paper briefly describes the main characteristics of the two PTP existing versions and underlines the main differences between them. As PTP requires special expensive network equipments to achieve its maximum performance, some tests using standard network switches are performed in order to evaluate if performance achieved is enough for typical flight test applications. Keywords: Time synchronization, PTP Performance, FTI application. 1 Introduction When analyzing data coming from an aircraft, it is mandatory to have all of them dated in a coherent time reference, so all measurements captured under certain event have the same dated time. Very often the number or nature of the measurements requires the installation of many different hardware units. In order to coherently date all the data, time synchronization between the different hardware units becomes essential. Historically, synchronization has been based in a point-to- point schema where a single dedicated cable between two equipments transports a synchronization signal (typically an analog one). The introduction of Ethernet network schemas in instrumentation installations have driven the development of networked synchronization protocols. Networked schemas simplify the installations, allowing more flexible architectures and moreover, best performance can be easier achieved since high-frequency analog signals are not necessary. Currently, the most important networked protocol for precise applications is Precise Time Protocol (PTP), published under the standard IEEE 1558. Section 2 summarizes the legacy synchronization architecture based on the standard IRIG Time Codes and shows their performance. In Section 3 a network synchronization architecture advantages are addressed. Section 4 explains how PTP works. Main differences between the two existing versions IEEE 1588v2002 (PTPv1) and IEEE 1588v2008 (PTPv2) are also addressed. Section 5 presents the results for several PTP performance tests performed using different network switches. The aim of the tests is to evaluate if the synchronization performance achieve with standard network switches can be enough for typical flight test applications, so expensive special network equipment may be avoid. In Section 6 main conclusions reached after testing PTP technology are summarized. Results demonstrate that PTPv1 can be used without any special switching hardware obtaining similar accuracies to those achieved using IRIG-B. It is also shown that PTPv2 must be used with special switching hardware in order to obtain serviceable accuracies. PTP accuracies achieve are highly hardware manufacturer dependant. This fact is more relevant in PTPv2 technology, which seems to be less mature. As the probable survivor standard will be PTPv2, manufacturers must evolve the technology in the medium term. 2 Legacy synchronization technology The Inter-range Instrumentation Group (IRIG) is an USA group defining standards to be used by the instrumentation community. IRIG-200 standard [1] (last reviewed in 2004) addresses the harmonization of synchronization across test ranges by specifying a number of possible signals to be used. IRIG-200 is divided in code format, modulation, frequency and coded information [1, pag. 4-1]. 6 code formats, 3 types of modulation, 6 frequencies and 8 information combinations are possible. By combining these elements, different types of synchronization signals can be generated. The most used among Flight Test Instrumentation (FTI) community is the IRIG-B 120 signal. Therefore, it is a B format: 1 second cycle with 100 bits per second [1, pag. 6-6]. First 1 denote amplitude modulated; following 2 stands for 1 kHz sine wave carrier frequency; last 0 defines the information included: Time_of_Year, Control- Functions and Time Of Day expressed in seconds of the day in binary count. IRIG-B 120 accuracy is guaranteed only to be 1 millisecond (equal to 1 period of the 1kHz carrier signal). Nevertheless, by accurately measuring the carrier phase angle, it is possible to achieve an accuracy of tens of microseconds [2]. 1: Flight Test Means – Airbus Defence&Space, Avd. John Lennon s/n, 28906 Getafe (Spain) ETTC 2015– European Test & Telemetry Conference If better accuracy were required, an unmodulated signal should be used, since it can provide tens of nanoseconds accuracy. Nevertheless, sending this signal to several equipments is quite complex [2]. IRIG-B 120 was widely adopted by FTI community because, as a modulated audio signal, it could be spread much more easily [2]. Even so, impedance adaptation of the cable is crucial in order to get the maximum performance. This fact limits the scalability of an IRIG-B installation. It is common to see scalable-limited daisy-chain distribution schemas. Parallel schemas requiring specific hardware with limited number of outputs can be also found. Daisy Chain Synch IN Synch OUT Synch IN Synch OUT Synch OUT Synch IN … SynchIN Synch OUT #1 Synch OUT #2 Synch OUT #N Paralell Distribution Figure 1.- Legacy IRIG-B Architectures 3 Networked synchronization concept Nowadays, Ethernet based instrumentations in FTI are being used more and more. Three most important advantages are driving the change. In first place, the standard and widely adopted hardware available makes easy and cheap accessing the technology. Secondly, the high degree of scalability of an Ethernet star topology makes easy adding more hardware units. Lastly, the high concurrent data rate possible to exchange in a transparent way for the user. Moreover, by implementing different protocols over Ethernet, it is possible to exchange information for different purposes over the same cabling. Ethernet Switch #1 Ethernet Switch #N Ethernet Switch #2 Sources 1 2 K... Sources 1 2 L ... Sources 1 2 M... Figure 2.- Ethernet Star Topology Therefore, it is possible to reduce cabling complexity in FTI installations by using a synchronization protocol over Ethernet. In addition, using digital information rather than analog signals, drastically reduces performance impact of impedance mismatching and other noise sources. Besides, synchronization would take advantage of Ethernet scalability. Several Ethernet time synchronization protocols have been defined over the years like UNIX daemon timed and Digital Time Synchronization Service. Nevertheless, these protocols are day-time oriented (for IT purposes mainly) and had no clock discipline features. This fact makes its use worthless for instrumentation purposes. Network Time Protocol (NTP) [3] was the first including clock discipline. NTP typical accuracy values are tens of milliseconds. Nevertheless, under ideal Ethernet network conditions, 1 millisecond could be achieve [3]. Higher values are not possible due to the operating system stack latency [4]. 4 PTP explained in depth As NTP performance is not enough for many precision applications, such as FTI, a new Precise Time Protocol (PTP) must be used. In the very basic PTP is similar to NTP but implementing hardware time stamping in each Ethernet port. This eliminates the operating system stack latency, which is the main error contribution in NTP. PTP UDP IP Driver MAC Phy PTP UDP IP Driver MAC Phy NET HW Timestamping points Delay: < 50 ns Fluctuations: < 1 ns Asymmetry: <100 ns SW Timestamping points Delay: 0.1-3 µs Fluctuations: 0.1-3 µs Asymmetry: <3 ms Packets Exchange Master Clock Slave Clock Figure 3.- PTP on the network stack [5][6 pag, 55] 4.1 PTP Basic Operation As NTP, PTP is master-slave protocol based on packet exchange between both ends. Two are the processes that simultaneously takes place: syntonization and synchronization. Syntonization is the mechanism to make the slave clock running at the same speed as the master clock. It is achieve by using a continuous flow of Sync messages from master to slave. Master Clock Slave Clock tk 1 tk 2 tk+1 1 Sync(t k 1)Follow_up(t k 1) Sync(t k+1 1)Follow_up(t k+1 1) tk+1 2 Figure 4.- Syntonization process [5] Sync messages are dated with t1 when leaving the master clock and with t2 when entering the slave clock. The slave clock must adjust its speed until both intervals are equal. ETTC 2015– European Test & Telemetry Conference Master clock time t1 can be sent to slave clock using two different methods. One-step clocks include t1 in the Sync message, while two-steps clocks send t1 in a later message called Follow_up message. Two-steps clocks are easier to design as the heavy duty is in the software, while one-step ones have a more complex hardware part because they must be able to timestamp the Sync message on the fly [5][7]. Sync messages are sent periodically, at a rate of some messages per second. Continuous control is mandatory as oscillators are susceptible to environmental changes. Synchronization is the mechanism that determines the slave’s offset from the master (i.e. difference in seconds between both clocks). This is done by measuring the round-trip time of the packets. Under the assumption of a symmetrical transmission path for Sync and Delay_Req messages, offset and delay are obtained as follow [6, pag. 50-53]. In order to maintain the slave clock synchronize along the time, a continuous correction is mandatory. For this purpose, values obtained from synchronization process are used to feed a Proportional-Integration (PI) control loop [6, pag. 146]. Master Clock Slave Clock t1 t2 Sync(t1)Follow_up(t1) Apparently same time t3 t4 Delay_Req O=Offset D=Delay D O Delay_Resp(t4) t1 t2 t3 t4 Figure 5.- Synchronization process [5] 4.2 Introducing PTP in a network As stated in previous section, PTP works reasonably well when the condition of symmetrical transmission path is comply. If both ends were directly connected through a cable, this assumption would be complied. Nevertheless, in an Ethernet network there is normally a switch in the middle with variable queues introducing jitter in transmission times. To overcome this issue, PTPv1 introduced the concept of Boundary Clock (BC) (see Figure 6). Section 5.3 and Section 5.4 will address the synchronization accuracy impact when using a regular switch. Several boundary clocks may be cascaded, the same way regular switches are. This cascade schema makes errors introduced in each step accumulative downwards. PTP UDP IP Driver MAC Phy PTP UDP IP Driver MAC Phy NET Master Clock Slave Clock PTP UDP IP Driver MAC Phy Master Clock PTP UDP IP Driver MAC Phy Slave Clock NET Sync Boundary Clock (Special Switch) Figure 6.- Boundary Clock Schema Therefore, when many steps are introduced synchronization performance is affected. Although it is not a big issue in an FTI installation (typically consisting of one switch or two cascaded at most) it is a huge issue for telecommunications or energy applications. For this reason new switch types were introduced in PTPv2 definition. 4.3 PTPv2 improvements over PTPv1 [5] This section contains not all but the most relevant improvements introduced in PTPv2. None of them are important for FTI instrumentation but, as PTPv2 will be the probable surviving standard, it is important to know the main differences with PTPv1. 4.3.1 Higher Message Rate Time synchronization for mobile 4G technology requires frequent state actualization. For this reason PTPv2 includes the possibility to send up to 128 sync messages per second, instead of 1 packet/second in PTPv1 4.3.2 Shorter Sync Message PTPv1 Sync message include information related to master clock in order to allow slaves to select de Best Master Clock. In PTPv2 this feature is performed separately, as it is not necessary to update this information 128 times per second. Sync message is then reduced to 44 bytes from the original 128 bytes in PTPv1. 4.3.3 Timestamp resolution PTPv1 maximum representation in the messages is 1ns, while in PTPv2 it has been improved to 2-16 ns. Therefore a sub-nanosecond accuracy is somehow foreseen. In addition, in order to solve the accumulative errors introduced by Boundary Clocks, new switches types are introduced (see Section 4.4) 4.4 Transparent clocks PTPv2 defines new types of switches in order to avoid the accumulative error schema introduced by Boundary Clocks when cascaded. A Transparent Clock (TC) is an Ethernet switch capable to measure the time a PTP message spends in the switch during its transit. This time is called residence time. TCs ETTC 2015– European Test & Telemetry Conference are able to introduce the value of the residence time in the correction field of Sync message in a one-step schema, or in the Follow_up message in a two-steps one. Two types of TCs are defined. Master Clock Slave Clock t1 Sync(t1,c) TC1 Sync(t1,c) c = cinitial TC2 Δs1 Δs2 c = c+Δs1 Sync(t1,c) c = c+Δs2 t2 Timestamping point Δs Residence time c Correction Field Figure 7.- Sync message through two E2E TCs 4.4.1 End-To-End (E2E) E2E TCs allow the slave to know the residence time accumulation in the exchanged packets by adding the residence time to the correction field in each packet going through the TC (see Figure 7). Same schema can be applied to the Delay_Request packet. As this packet is sent to the master clock by each slave clock, the end master clock knows latency time to each slave on the network (see Figure 9). Master Clock Slave Clock t1 Sync(t1,c) TC1 Sync(t1,c) c = cinitial TC2 Δs1 Δs2 c = c+Δs1+ΔL1 Sync(t1,c) c = c+Δs2+ΔL2 t2 Timestamping point Δs Residence time c Correction Field ΔL1 ΔL2 ΔL3 ΔLUplink delay c = c+ΔL3 Figure 8.- Sync message through two P2P TCs 4.4.2 Peer-to-Peer (P2P) P2P TCs measure the link delay to all neighbouring clocks (either slave clocks or other TCs). To measure links delay, PTPv2 introduces a new type of message called Pdelay_Req / Pdelay_Resp. When a Sync message traverses a P2P TC, the correction field is updated both with residence time and the link delay previously measured (see Figure 8Figure ). P2P schema does not overload master clock with information coming from all the nodes in the network, as represented in Figure 9. Additionally, notice link delays are even measured in connections where standard traffic is blocked by i.e. Spanning Tree Protocol. 4.5 PTPv1 and PTPv2 compatibility Because message format is different in PTPv2 from PTPv1, it is not possible for clocks of different standards to synchronize between them. Nevertheless, there are on the market several switches able to use PTPv1 in one port and PTPv2 in another one. Therefore, islands of different PTP versions can be used. Figure 9.- E2E (left) and P2P (right) network schema It is however expected that PTPv2 will be the dominant protocol in the future. For this reason, although FTI installations do not take any advantage from PTPv2, it worth the time knowing how PTPv2 works. 5 PTP performance tests 5.1 Metodology and means Differences between master and slave clocks can be measured by simultaneously monitoring the Pulse Per Second (PPS) signals coming out from them. When synchronized, the difference between both signals tends to be 0 seconds. Master Clock PPS PPS Switch Slave Clock Oscilloscope ETH ETH Figure 10.- Test RIG schema Table I shows the role distribution of the four equipments available to perform the tests. Each equipment has a different manufacturer and all of them are FTI focused. IRIG-B PTPv1 PTPv2 Master A A D Slave B B/C B/C Table I.- Test Means available 5.2 Reference measurements To be taken as reference, synchronization performance without switches between master and slave has been measured. Figure 11 shows typical accuracy achieves using IRIG-B. Master Clock A Slave Clock B µ=40 µs σ=2 µs IRIG-B Figure 11.- IRIG-B Performance Comparing Figure 12 and Figure 11Figure , it is possible to appreciate the improvement in one order of magnitude when using PTPv1 against IRIG-B. In Figure 12, the importance of actual implementation of the protocol can ETTC 2015– European Test & Telemetry Conference be seen. Different equipments from different manufacturers provide quiet different synchronization performance. Master Clock A Slave Clock B µ=300 ns σ=150 ns PTPv1 Master Clock A Slave Clock C PTPv1 µ=20 ns σ=20 ns Figure 12.- PTPv1 Performance As clearly represented in Figure 13, PTPv2 performance is worst than PTPv1 in both tested hardware. Hardware B PTPv2 synchronization is even worst than IRIG-B. Master Clock B Slave Clock C Master Clock B Slave Clock B PTPv2 PTPv2 µ=200 ns σ=25 ns µ=300 µs σ=120 µs Figure 13.- PTPv2 Performance 5.3 PTPv1 performance using regular switch Nowadays, Ethernet port cost for regular switches is 50- 100 times lower compared to FTI PTP compliant switches. Regular switches does not assure the symmetrical transmission path delay require in PTP (see Section 4). However, as PTP performance exceeds FTI requirements, synchronization accuracy achieve with regular switches could be enough for FTI purposes. Table II shows how PTPv1 Boundary Clocks do well their job and synchronization performance is not affected by traffic level in the switch. On the other side, when using a regular switch, performances are 10 times worse. Moreover, as expected, it is affected by the amount of traffic managed by the switch. Nevertheless, with 33 Mbps of traffic (corresponding to a big instrumentation), performance is similar to the one obtained when using IRIG-B. Therefore, using PTP with regular switches is possible in most FTI applications, getting a similar or better accuracy (depending on traffic amount) to the one obtained with IRIG-B. MasterClockA SlaveClockC Network Traffic 0 Mbps 6 Mbps 33 Mbps Regular Switch Mean µ 0 ns 0 ns 0 ns Std σ 500 ns 500 ns 5000 ns Boundary Clock Mean µ 0 ns 0 ns 0 n Std σ 50 ns 50 ns 50 ns Table II.- PTPv1 performance with different types of switch 5.4 PTPv2 performance using regular switch Unlike PTPv1, PTPv2 performance is severely affected when using a regular switch to interconnect master and slave clocks. Figure 14 shows how, even with no additional traffic in the switch, synchronization accuracy is worst than with PTPv1. Mean desviation is 25μs for hardware C and 120μs for hardware B. Again big differences in performance between different manufacturers can be found. When adding moderate level of Ethernet traffic to the system, performance decline. Slave clock B stability is so affected that it cannot be used for FTI purposes. Slave clock C, although working better, presents worst stability than the one achieve using IRIG-B. Thus it is not possible to get an usable synchronization using PTPv2 with regular switches. A transparent clock is mandatory to use PTPv2 in instrumentation arquitectures. Master Clock D Slave Clock B Master Clock D Slave Clock B Slave Clock C Slave Clock C PTPv2 No Traffic PTPv2 6 Mbps Figure 14.- PTPv2 performance using regular switch 6 Conclusions In order to get from PTP the best performance, it is mandatory to use expensive special switches, which assure a symmetric transmission path between master and slave clocks. Using cheaper regular switches with variable latencies in their queues provides worst PTP performance. Nevertheless, it has been demonstrated that PTPv1 accuracy is as good, or even better, as the one provided by the legacy IRIG-B synchronization. Performance is directly link to the amount of traffic in the switch. However, it is acceptable even at rates corresponding to big FTI installations. Therefore, PTPv1 can be used through regular switches for most FTI installations. In contrast, PTPv2 performance when using regular switches is not useful for FTI installations. PTPv2 performance is near to IRIG-B when there is no additional traffic in the switches, but degrades severely under moderate traffic conditions. It is then mandatory to use transparent clocks if using PTPv2 as synchronization protocol. The main aim of the tests was not to compare synchronization performance of different hardware manufacturers. Though, big differences in performance depending on the hardware manufacturer have been found. Differences are especially big when testing PTPv2, which indicates a low mature level in this technology. As the probable future standard is the PTPv2, and it is not compatible with PTPv1, FTI manufacturers must evolve the technology in the medium term. ETTC 2015– European Test & Telemetry Conference 7 References [1] Range Commanders Council, Telecommunications and timing group. “IRIG Serial Time Code Formats”, 2004. Accesed: April 2015. http://www.wsmr.army.mil/RCCsite/Documents/200- 04_IRIG%20Serial%20Time%20Code%20Formats/200- 04_IRIG%20Serial%20Time%20Code%20Formats.pdf [2] B. Dickerson, “IRIG-B Time Code Accuracy and Connection Requirements with comments on IED and system design considerations”, (No date) Accesed: April 2015. http://www.arbiter.com/files/product- attachments/irig_accuracy_and_connection_requirem ents.pdf [3] From Wikipedia, the free encyclopedia, “Network Time Protocol”. (No date) . Accesed: April 2015. http://en.wikipedia.org/wiki/Network_Time_Protocol [4] B. Dickerson, “Precision Timing in Power Industry, How and Why we use it”. (No date). Accesed: April 2015. http://arbiter.com/news/technology.php?id=4 [5] H. Weibel, “The Second Edition of the high Precision Clock Synchronization Protocol”, 2009. Accesed: April 2015. http://ines.zhaw.ch/fileadmin/user_upload/engineering /_Institute_und_Zentren/INES/Downloads/Technology _Update_IEEE1588_v2.pdf [6] J.C. Edison, “Measurement, Control and Communications using IEEE1588”, 2006. Ed. Springer. ISBN-10: 1-84628-250-0. [7] D. Arnold, “One-Step or Two-step?”, 2013. News and Tutorials from Meinberg. Accesed: April 2015. http://blog.meinbergglobal.com/2013/10/28/one-step- two-step/ 7. Acronyms PTP Precise Time Protocol IRIG Inter-range Instrumentation Group FTI Flight Test Instrumentation NTP Network Time Protocol PI Proportional-Integration Control Loop BC Boundary Clock TC Transparent Clock PPS Pulse per Second

Creative Commons Aucune (Tous droits réservés) Aucune (Tous droits réservés)
contenu protégé  Document accessible sous conditions - vous devez vous connecter ou vous enregistrer pour accéder à ou acquérir ce document.
- Accès libre pour les ayants-droit

1 User Programmable FPGA I/O for Real-Time Systems – Combining User Friendliness, Performance, and Flexibility Andreas Himmler1 dSPACE GmbH, Paderborn, 33102, Germany Jürgen Klahold2 dSPACE GmbH, Paderborn, 33102, Germany Field Programmable Gate Array (FPGA) technology has already proven its benefits for a wide range of applications requiring I/O interfaces with highly parallel computation power, short latencies in combination with with very fast, high-resolution signal processing. These benefits are for instance highly appreciated to interfaces real-time systems facing electric drives. Typical real-time systems are hardware-in-the-loop simulators and rapid-control- prototyping electronic control units for electric drives. This paper presents a workflow and an associated toolchain for a real-time technology that fulfills these requirements. The workflow gives users a framework to easily interface their own FPGA code in an FPGA I/O board with the physical I/O on the one hand and the real-time code running in the CPU of the real-time system (typically based on Matlab/Simulink). This framework is readily prepared but nonetheless flexible. The physical I/O of this I/O board is modular, in order to select from a range of off-the-shelf I/O modules complementing the FPGA base board or to build very specific and project dependent I/O modules. An example is given about how this toolchain, I/O boards and real-time system are applied to real-world problems. Nomenclature ECU = electronic control unit FPGA = field-programmable gate arry HIL = hardware-in-the-loop I/O = input/output LVDT = linear variable differential transformer MEA = more electric aircraft PSM = permanent magnet sychonous motor PWM = pulse width modulation RCP = rapid control prototyping RTI = real-time interface VHDL = very high speed integrated circuit hardware description language XSG = Xilinx® system generator 1 Business Development Manager Aerospace, Product Management, dSPACE GmbH, Rathenaustraße 26, 33102 Paderborn, Germany. 2 Product Manager Hardware in the Loop Testing Systems, Product Management, dSPACE GmbH, Rathenaustraße 26, 33102 Paderborn, Germany. 2 I. Introduction IELD Field Programmable Gate Array (FPGA) technology has already proven its benefits for a wide range of applications requiring I/O interfaces with highly parallel computation power, short latencies in combination with with very fast, high-resolution signal processing. These benefits are for instance highly appreciated to interfaces real-time systems facing electric drives. Typical real-time systems are hardware-in-the-loop simulators and rapid- control-prototyping electronic control units for electric drives. In addition, FPGAs give users extreme flexibility when they need to implement specific sensor or actor interfaces (e.g. synchro, resolver, or audio interfaces) or bus interfaces. Users are well aware of these benefits. Nevertheless, they regard the usage of FPGA-based interfaces is complicated because they may need to program an FPGA by themselves and they need to interface the FPGA with the real-time CPU running their algorithms (e.g. simulation or analysis algorithm). Thus, there is a need to build FPGA- based I/O for real-time systems that adds an efficient, intuitive and simple to use workflow to the technology-inherent benefits of FPGAs. This paper will present a workflow and an associated toolchain for a real-time technology that fulfills these requirements. The workflow gives users a framework to easily interface their own FPGA code in an FPGA I/O board with the physical I/O on the one hand and the real-time code running in the CPU of the real-time system (typically based on Matlab/Simulink). This framework is readily prepared but nonetheless flexible. The physical I/O of this I/O board is modular, in order to select from a range of off-the-shelf I/O modules complementing the FPGA base board or to build very specific and project dependent I/O modules. Examples will be given about how this toolchain, I/O boards and real-time system are applied to real-world problems. The first two sections give an overview about real-time systems that require the application of FPGAs for closed-loop operation with very short cycle times. The first kind of real-time system are Rapid- Control Prototyping (RCP) Systems and the second is a Hardware-in-the-Loop (HIL) simulation system. The third section decribes in more detail why there is a need to apply FPGAs in RCP and HIL systems. Section four describes a user-friendly workflow to use FPGAs in a real-time F Figure 1. Rapid Control Prototyping systems: compact, ruggedizd system, modular systems, and a modular, ruggedized signal conditioning system. Figure 2. General RCP development process. 3 system most efficiently and section five contains an application example how FPGAs are used in a HIL system to test electric drives. II. Rapid-Control Prototyping Systems for Model-Based Development Many industries are under pressure to reduce their development times as they produce unique and innovative products. These two factors are indispensable to success in a globalized market, especially for thigh-tech industries such as automotive, aerospace and communication, where electronic controls are a vital part of each new product. Model-based control design is a time-saving, cost-effective approach, because control engineers work with just a single model of a function or a complete system in an integrated software environment. This model-based development process results in an optimized and fully tested system, with no risk that individual components do not fit together optimally. To model controller strategies and the internal behavior of software components, tools such as MATLAB® /Simulink® /Stateflow® from Mathworks and TargetLink® from dSPACE can be used. If a new ECU or a new set of control functions has to be developed from scratch, quick trials have to be run at an early stage to verify the correctness of the control strategy. Tests in the real environment (e.g. vehicle, plane) or on a test bench therefore have to be carried out even before the new ECU hardware becomes available. Producing an application-specific prototype ECU for this purpose, e.g., by modifying a production ECU, would be expensive, time- consuming and inflexible. Instead, developers can use a powerful off-the-shelf rapid control prototyping (RCP) system which acts as an experimental ECU, but which has many advantages compared to other solutions. User requirements on RCP systems are very diverse. Some applications require compact, ruggedized RCP systems for in- vehicle applications while other RCP systems are to be used in the lab. Some users require systems that offer optimum scalability and flexibility while others require compact, all-in-one systems with a common set of well-know I/O interfaces. Examples of such systems are shown in Figure 1 and Figure 3. Such best-in-class RCP systems for model-based development fulfill two requirements: (1) The have high computation power combined with very low I/O latencies in order to provide great real-time performance and (2) they have a perfect Simulink integration to allow faster design iterations and to reduce the overall development time. III. Real-Time Systems for Hardware-in-the-Loop Simulation Hardware-in-the-loop (HIL) simulation is an integral and reliable part of the development process. Hardware-in- the-loop simulation is used for testing ECU functions, for system integration, and for testing ECU communication. The environment of the ECUs to be tested is simulated in real time (Figure 4). The environment can consist of Figure 3. Rapid Control Prototyping system MicroLabBox. Figure 4. Hardware-in-the-loop simulation. 4 interacting system components such as sensors and actuators, other subsystems or complete systems, and the aircraft or vehicle environment. The main advantages of HIL tests are reproducibility, systematic and automated testing also outside of safe system states, and the traceability of problems observed in the field. This makes it possible to conduct tests efficiently (time, costs) and as early as possible in the development process. The trend to test with virtual (i.e., simulated) ECUs that are later replaced with real ECUs highlights the importance of early testing (Ref. 2). A typical HIL system comprises the simulation hardware, such as:  Processor unit for computing the simulation models  Battery simulation as the power supply for the simulation system  I/O interfaces  Other auxiliary components such as load boards or failure simulation Connected to it is the unit under test, which usually is one or more electronic control units (ECUs) containing new functions or ECU software to be tested. The software for configuring and automating the HIL test runs on a PC, as well as the software for parameterizing the simulation model and visualizing the simulation run. Test data management software can also be used. In the following subsection, the HIL technology SCALEXIO [2] is used in order to discuss topics related to HIL simulation. SCALEXIO uses the I/O network IOCNET, which is used for high bandwidths and low latencies and supports the required model synchronization on the real-time processor and FPGAs applied on I/O boards. The configuration of a SCALEXIO HIL system is done by using the dedicated tool ConfigurationDesk® (ref.). The configuration process is roughly divided into three tasks: describing the externally connected devices (control units, real loads, etc.), selecting the I/O functions for each signal, and linking the I/O functions to the plant model. IV. FPGAs for Real-Time Systems Rapid Controls Prototyping (RCP) and Hardware-In-the-Loop (HIL) technologies requiring very fast computations, in particular when electric drives are controlled or simulated. In the prototyping phase of a new electronic control unit (ECU) it might be necessary to use new sensors and therefore to implement new protocols or new control strategies which require a more precise control of the power stages. The implementation of new protocols or controls with direct effect on electric components are not feasible on processor and require a direct hardware implementation, which is only possible on ASICs or FPGAs. As ASICs are not also feasible for prototyping new functions, FPGAs are used. The simulation of electric drives require the simultaneous computation of complex simulation models and highly precise measurement of the electronic control unit (ECU) signals. Generally, time-critical I/O Figure 5. FPGA Interface Library. Figure 6. Signal chain on the HIL system SCALEXIO. 5 computations are described by an FPGA model. Even parts of the plant model are frequently executed on the FPGA to meet the needs of a modern ECU. dSPACE's SCALEXIO provides a convenient solution for both these cases and also for mixed scenarios. If mean value models are used on the processor side, the output signal is often updated only once per ECU control step size (typically 50 µs). FPGA-based model computation offers decisive advantages for highest requirements on dynamics and accuracy. FPGAs reach very high sampling rates so that output signals are calculated and updated considerably more often than once per ECU sampling cycle. The result is an appreciably higher quality of simulation. For example, high-frequency simulation makes it possible to simulate the inductance current ripple caused by pulse width modulation (PWM) control, improve the precision with which higher frequencies are simulated, and ensure high control loop stability. The measurable latency between the hardware input and hardware output is typically reduced from 50 µs to about 1 µs in comparison to processor-based models. The simulated current values are output every 100 ns. Nevertheless, it should be mentioned that the tool chain for FPGAs consumes a lot of time and normally requires specialized hardware developers. Even if the usability can improved to a level that software developers can also design functions for FPGAs, it is still more time consuming effort necessary. Therefore, there is the intention to keep the function parts, which should be executed on an FPGA as stable as possible avoiding to run through the time consuming process too often. V. User-friendly workflow for Model-Based-Design Commonly FPGAs are only used from hardware developers on I/O boards with a fixed set of implemented functions. So special tool chains and special languages (e.g. VHDL) can be used to program the FPGA. With new requirements on turnaround times (below one µs) for models or new interfaces, the inherent given flexibility of FPGAs has be used from function developers to satisfy the shortened development times. Using VHDL in this use case is inconvenient, as the developer would have to learn a new language and development chain, which is completely different from his ordinary work. Therefore, a method to program an FPGA is required, which meets the known workflow of a function developer. Known from the modelling for processors, the most convenient way to configure an FPGA is a graphical method. The established tooling for processors is Simulink® , so it suggests itself to use Simulink as well for the programming of the FPGA. The Xilinx® System Generator (XSG) is a Simulink blockset for configuring Xilinx® FPGAs. It contains simple logic elements as well as complex blocks such as Fourier transforms and FIR filters. With this blockset or additional libraries based on the XSG, it is possible to implement the desired function on the FPGA. However, the model has still to be connected to the environment of the FPGA. This is on the one side the I/O connected to the FPGA in order to measure or generate Figure 7. XSG Utils Library. Figure 8. FPGA-based simulation – illustrating the data transfer from the processor model (top window) to the FPGA model (bottom window). 6 signals and on the other side the real-time processor in order to exchange data. To evaluate the desired behavior an offline simulation within Simulink is very useful, as the build process for an FPGA application requires a long time. It is an advantage if the interface blocks directly support the offline simulation, as no variant from the model is required and some effects like quantization or value ranges can be directly taken into account. A concurrent simulation of the model for the real-time processor stimulates the model for the FPGA with proper values and also verifies the interaction of both model parts. In addition, a behavior verification is much easier within Simulink especially together with other finally used model parts, as the common method from hardware developers writing special testbenches within VHDL. After the offline verification of the behavior, a build of the application is required. As a function developer is unfamiliar with the special tooling for FPGAs, a push button solution is preferred. Therefore, it would be optimal if the model for the FPGA could directly transferred in an I/O function for the processor model. This allows the function developer as well as his colleagues to use the FPGA application with in their common software development environment, also for further projects as well as for several FPGA boards. dSPACE provides both the hardware (e.g. MicroLabBox or DS2655 FPGA Base Board) and the software (RTI FPGA Programming Blockset) to connect the XSG models to the FPGA’s interfaces. The FPGA Programming Blockset provides blocks for implementing the interface between the FPGA mounted on a dSPACE board and its I/O, and the interface between the dSPACE FPGA board and real-time PC of the system. On the I/O side the blockset allows a simple configuration of the basic features of the I/O, e.g. the electrical configuration of a digital output driver (push/pull, high-side voltage 3,3V or 5V etc.). Further functions like the measurement of the duty cycle of a PWM signal can easily be implemented by connecting a PWM measurement block from the utils library to the I/O block. In this matter, it is also possible to add more complex function blocks like resolver simulation or the control of power stages. This allows a comfortable handling of I/O functionalities known form RTI for the standard modelling for processors with the new possibility to adapt the I/O function if necessary, enabling the implementation of new interfaces for example. The I/O functions can be extended with controllers or models (e.g. based on the XSG Electric Components Library) to close loops with turnaround times far below 1 µs. Slower parts can be implemented on the processor. Therefore, also a convenient way to exchange data between the FPGA and the processor is given. The user can choose between a buffer communication to transfer a burst of data or register based communication to transfer single values also from different levels of the FPGA related model. Here it is helpful to group the registers to logical groups so that the Figure 9: FPGA-based simulation at signal level. Figure 10: Simulation of the current ripples of a permanent magnet synchronous motor. 7 consistent sets of data are transferred. The framework directly supports this. Considering the complete workflow, a compfortable toolchain is given. It is fully Simulink based and hides FPGA specific tools. Function devlopers can stay in their common environment, model the desired behavior in their familiar way and verify it, including quantisation and fixed-point effects. The result is an FPGA application which can directly be used in the known model based design flow for processors. VI. Application Example After the functions of the electronic control units (ECU) associated with an electric motor have been developed and implemented on the production ECU, they have to be tested thoroughly. With HIL simulation, it is easy to cover all the different motor varieties and their ECUs. A combination of fast model computations and low I/O latencies is indispensable for simulating highly dynamic controlled system in electric drives engineering, and is also a typical application area for SCALEXIO. With the connection to the XSG Electric Components Library, the following challenges can be handled: The realistic representation of current behavior required for developing analog current controllers runs with a sampling rate considerably higher than once per PWM period. Simulating electrical circuit with frequencies higher than 1 kHz, processor-based simulations exceed their limits. Using FPGA technology increases the range several times over. The enormous FPGA sampling rate makes PWM synchronization unnecessary, so now systems with a variable PWM switch frequency can also be simulated realistically. Highly dynamic applications such as DC/DC converters require higher PWM frequencies. These frequencies are higher than 20 kHz, and the only way to represent the current and voltage realistically is by FPGA-based simulation. When an electric motor is simulated at power level, voltage and current values must be represented as realistically as possible. This is necessary if these reference values are to be used as input to the electronic load. Here too, fast computation on the FPGA is absolutely essential. Figure 10 shows an example for the simulation of current ripples: the current behavior in the stator of a permanent magnet synchronous motor with a torque of 6250 rpm; the PWM frequency is 32 kHz. References 1 Himmler, A., “Modular, Scalable Hardware-in-the-loop Systems,” ATZelektronik worldwide, Vol. 5, No. 2,2012, pp. 36-39. 2 Himmler, A., “Hardware-in-the-Loop Technology Enabling Flexible Testing Processes”, 51st AIAA Aerospace Sciences Meeting including the New Horizons Forum and Aerospace Exposition, DOI 10.2514/6.2013-816, eISBN 978-1-62410-181-6, Grapevine (Dallas/Ft. Worth Region), Texas, 2013. 3 Himmler, A., Allen, J., and Moudgal, V., “Flexible Avionics Testing - From Virtual ECU Testing to HIL Testing”, SAE 2013 AeroTech Congress & Exhibition, Montreal, Canada, 2013, SAE Technical Paper 2013-01-2242, 2013, doi:10.4271/2013- 01-2242, http://papers.sae.org/2013-01-2242/ [cited December 12, 2013]. 4 Himmler, A. “Openness Requirements for Next Generation Hardware-in-the-Loop Testing Systems”, AIAA Modeling and Simulation Technologies Conference, doi:10.2514/6.2014-0636, National Harbor, Maryland, 2014. 5 Schütte, H., Wältermann, P., “Hardware-in-the-Loop Testing of Vehicle Dynamics Controllers – A Technical Survey, SAETechnical Paper [online database], Paper 2005-01-1660, URL: http://papers.sae.org/2005-01-1660 [cited December 13, 2012].

Creative Commons Aucune (Tous droits réservés) Aucune (Tous droits réservés)
contenu protégé  Document accessible sous conditions - vous devez vous connecter ou vous enregistrer pour accéder à ou acquérir ce document.
- Accès libre pour les ayants-droit

N° 9 – Guaranteed end-to-end latency through Ethernet - Øyvind Holmeide and Markus Schmitz - OnTime Networks - Norway, United States. Latency sensitive data in a Flight Test Instrumentation (FTI) system represents a challenging network requirement. Data meant for the telemetry link sent through an on-board Ethernet network might be sensitive for high network latency. Worst case latency through the on-board Ethernet network for such data, might be as low as a few hundred microseconds. This challenge is solved by utilizing the Quality of Service (QoS) properties on the Ethernet FTI switches. This paper describes how to use Ethernet layer 1, layer 2 or layer 3 QoS principles of a modern Ethernet FTI network

Creative Commons Aucune (Tous droits réservés) Aucune (Tous droits réservés)
contenu protégé  Document accessible sous conditions - vous devez vous connecter ou vous enregistrer pour accéder à ou acquérir ce document.
- Accès libre pour les ayants-droit

ETTC 2015 – European Test & Telemetry Conference Lessons for Onboard Data Storage from Electronic Data Processing Environments and Airborne Video Systems Malcolm Weir Ampex Data Systems Corporation, Redwood City, CA, USA Abstract: When compared to the changes seen in commercial data processing and digital video over past 3 decades or so, onboard data storage techniques have evolved surprisingly little. This paper explores some of the implications on data storage devices used for test instrumentation acquisition and the methods employed to format and store that data. Keywords: Metadata, File Formats, Database 1. Introduction In the early 1980’s, the commercial Electronic Data Processing (EDP) marketplace was starting a gradual transition from off-line batch processing to what was to become known as On-Line Transaction Processing (OLTP). While the operational environment in which the Test and Evaluation (T&E) community has to function is very different from that of EDP shops, many of the goals underlying the data collection are similar, or in some way related, to that of the more mundane commercial industry. There are also, of course, some very significant differences. More recently, the techniques used to manage airborne video acquisition have also evolved. Again, there are similarities between video acquisition and T&E applications. 2. Current Onboard Data Storage 2.1. Legacy Use Cases In the context of this paper, onboard data storage primarily concerns data acquisition operations. In the abstract, acquisition can be characterized as being dominated by write operations, where data is aggregated and stored. Historically, this was the totality of the exercise. Real-time onboard processing (or the absence of same), while not obviously directly associated with onboard storage, is generally a significant part of (or absence from) any data acquisition system. And likewise the data links away from the test article are obviously part of the system, but the data flow is simple and unidirectional. As the diagram shows, the data store was originally tape based, and in terms of logical structure, in many cases this has evolved little beyond changes in the physical media. Increasingly, though, there are demands for retrieval of the acquired data before the acquisition is “complete”. This is coupled with far more complex data flows, with bidirectional data flows, and with a desire for multiple non-overlapping data stores (e.g., crash survivable storage), which typically hold a subset of the data. This all leads to something like the data flow diagram resembling the one on shown in Figure 2. 2.2. Data Storage Formatting As indicated earlier, even though tape recording data storage systems have largely been replaced by ones employing solid state devices, most of which emulate magnetic disk drives, the data is still written in a form that is completely compatible with sequential-access devices like tape drives. In fact, the dominant standard in the field, the “IRIG 106 Chapter 10 Solid State On-Board Recorder Standard”, also known as “IRIG 106 Chapter 10 Digital Recording Standard” (the exact title depending on the version), not only mandates that data be stored in a sequential access “flat file”, but also that the file must be logically contiguous for each recording “session”. While this was a reasonable restriction when the data set for a given recording session ranged in size from a few hundred megabytes up to a few gigabytes, it becomes dramatically unworkable for datasets sized in terabytes. More recent standards (both published and proprietary) that cover onboard acquisition, such as those from the “integrated Network Enhanced Telemetry” (iNET) effort avoid specifying how data items should be stored, but define access methods; the methods are typically fairly elementary, and thus could be implemented by searching and reading a flat, Chapter 10-style file and then filtering out the unwanted data – this sort of approach is in fact the Onboard Processing Data Sources TransmitterTransmitter Data Storage Onboard Processing Control RoomControl Room Data Sources TransmitterTransmitter Data Store Secondary Store Figure 1: Legacy Data System Figure 2: Evolved Data System ETTC 2015 – European Test & Telemetry Conference preferred technique for the iNET goal of filling in “drop outs” on an otherwise unreliable datalink. 2.3. File Format Taking the Chapter 10 file format to be representative of a typical “flat file” recording, the file structure can be generalized as:  A set of variable length records, termed “packets” in Chapter 10.  The records are stored ordered by a counter with a tick interval of 100 nanoseconds.  While the records appear similar to a time series, in practice they are not truly ordered by chronological time (because the payload in a given record may have been delayed at various points prior to writing to the file, and the payload in adjacent records may have not been delayed). However, for a given source, the records will constitute a time series.  The records have “weak typing”: while the definition of a given record is defined by the standard, there is no mechanism to verify the type of data mandated is actually present. For example, compressed video may be stored as some Chapter 10 flags followed by a series of MPEG-2 Transport Packets; as far as the file format is concerned, the Transport Packets can be invalid – for example, the mandatory sync byte may be incorrect – but that is permissible for the Chapter 10 file precisely because there may be value in collecting erroneous data.  Related to the concept of “weak typing” is the observation that generic container formats (“Message Data Packets”) can be used to hold arbitrary data, and such containers are, by definition, opaque from the perspective of the controlling standard: to interpret the Message Data container, you need additional information, and there exists no good method for providing that information.  There are three special record locations, two of which are mandatory and one optional. The first two records in the file must be a “setup record” and a “time packet”, and if “Recording Event indexing” is used, the final record must be the “Root Index”.  Metadata may be placed in the “setup record”, but the absence of metadata does not invalidate the file. In general, the metadata is narrative (“strain gauge on left wing”) rather than descriptive from a data modelling perspective (“integer in the range -127 to +128”). An excellent example of this distinction lies in recording GPS data: many GPS sources provide information in text strings (known as “sentences” by the applicable standard), but the data may more usefully be stored as numbers; so the metadata applicable to that record may indicate that certain records contain GPS time, but not how that time is stored1 . 2.4. Metadata As indicated above, current industry standard practice tends to focus the term “metadata” towards the sources of the data; Chapter 10 defines that the IRIG 106 Chapter 9 standard (“TMATS”) be used to describe the data sources. While this is essential for the interpretation of the information carried by that data, it overlooks the lowest- level metadata that describes how it is stored on board, and indeed how it is handled once it is transferred off the aircraft. Obviously, a schema can be developed for just about any organized dataset, but significantly the schema for typical legacy onboard recording files is only loosely defined. 2.5. Emerging Use Cases As on board systems and the related storage evolve, two factors stress the legacy way of saving data: first is the sheer volume of data being managed, and the second is the sort of environment illustrated in Figure 2 where data retrieval requests are proliferating. It is clearly desirable for the data store to be able to search previously saved records efficiently so that real-time data mining can provide exception-based answers, because the volume prohibits exhaustive scanning of the entire dataset. It is also clear that the handling metadata, both of the narrative and data modelling forms discussed earlier, needs to be incorporated into the design of large systems. There is also yet another layer of metadata that has become (at least) desirable and, in some cases, mandatory: the configuration information. The goal of this metadata is to preserve sufficient information so as to be able to configure the data acquisition system in the same state. Significantly, although these issues have been described for the on board storage system, they are as significant, if not more important, for the processing and archival segment “on the ground”. The concerns also apply to the process of moving data from the on board system to the processing system: it is frustratingly common to require the entire data set to be transferred off the vehicle before analysis can begin; in an ideal world, the “interesting portion” of the data should be transferred first, with the rest following later, or not at all. Fortunately, the task of defining which portion is “interesting” lies well beyond the scope of this paper. 3. Commercial Electronic Data Processing 3.1. Background The one of the fundamental techniques of off-line data processing is known as “Master In, Master Out”, or “MIMO”. In a MIMO process, the file consisting of “yesterday’s” master file dataset is sequentially read along with a sorted list of transactions; if the unique identifier (e.g., account number) matches, the master record is updated (e.g., the account balance is adjusted by the amount of the transaction). In any event, the Trans- actions Yesterdays Master Todays Master Update Process Figure 3: MIMO Processing ETTC 2015 – European Test & Telemetry Conference (potentially modified) record is then written out to create a new master file. The virtues of this sort of approach are obvious: neither the old master file nor the transaction file are modified, so it is trivial to preserve coherent backups. And, since all the files involved are read or written sequentially, the data can conveniently be stored on magnetic tape, which was by far the most cost-effective media of the day. The drawbacks are also clear: regardless of the volume of transactions, the entire master file must be read and written. And the transactions must be aggregated and ordered (sorted) appropriately for the operation (e.g., sorted by account number). Current EDP practice would probably involve a relational database system (RDBMS), and resemble something like this: take a snapshot of the database, and then apply transactions as they come in. At the end of the day, take another snapshot. The obvious advantages of the real-time update perhaps mask a slightly more subtle advantage: since each incoming transaction is processed without sorting or aggregation, it is straightforward to implement multiple dataset updates, so that maintaining searchable records of transactions becomes practical. And, of course, if you can search transactions, you can address the archetypical “Big Data” questions, including the legendary concept of mapping rainfall locations by tracking the sale of umbrellas. Against which is the fact that trying to process each transaction “on the fly” introduces potential issues of “bursty” traffic skewing the performance requirements. 3.2. Observations about Complexity If one equates the workload of receiving a transaction (e.g. over a network connection) with the workload of storing a transaction, then the I/O complexity of the update process as whole can be described, for the “MIMO” case as: O(2n+m) where n is the number of records in the master file, and m is the number of transactions being added. For the modern OLTP approach, assuming a master file significantly larger than the transaction file, the complexity is substantially lower: O(m + m log n) However, while the total complexity is substantially greater with the MIMO case compared to the OLTP one, a critical observation is that the "momentary" complexity with MIMO is much, much lower: there is no updating of index trees, transaction locking or any of the other sophistication that one expects with a relational database. Stated another way, while the I/O complexity of the OLTP case is lower than that of the MIMO one, other measures of complexity favor the latter. Perhaps more important, though, is the recognition that if there is no pre-existing master file, then the MIMO case degenerates into O(2m) while the OLTP version becomes O(m + m log m) For the modern OLTP approach, therefore, it’s clear that several times as much I/O will be performed, specifically of the order of log m times that of a simple linear file approach like that of the MIMO operation. But since OLTP systems exist (to be quite accurate: they dominate the universe of installed systems), it’s clear that the additional I/O complexity is not inherently a bar on using these types of solutions. 4. Airborne Video 4.1. Background Beginning with the invention, in 19562 , of the practical video tape recorder right up to the widespread introduction of digital TV, almost all video, whether airborne or commercial, was recorded in the same manner: as a single “file’ on a linear storage medium (i.e. a tape). With the adoption of digital video, largely driven by digital broadcasting, nothing much changed; the video, not digitized, was stored as a single file on some media (initially tape, but non-linear media – e.g., “disks” – have been rapidly adopted). This form of digital video has minimal manageable metadata, usually limited to fairly generic date-and- location information, possibly with a few “event” tags stored as flashes or audio tones. Significant additional metadata may exist within the file, as video overlays, which are, of course, not amenable to machine searching (and therefore might be more properly considered “data” rather than “metadata”), and which by definition obscure parts of the video frame. However, following the creation within the US Department of Defense of NIMA (now NGA) in 1996, a concerted effort was made to standardize video formats and introduce a comprehensive digital metadata “ecosystem”. These efforts are managed by the Motion Imagery Standards Board (MISB). Largely as a result this effort, the dominant format for airborne video has been steadily evolving towards a common structure:  A set of fixed length records / “packets” containing all information in the file (or stream).  The records are a time series, subject to the limitations of the data sources: a video frame is a snapshot, while an audio channel is a continuous source; the snapshotted data trickles out over the interval between frames; this makes it hard to make conclusions about the “correct” ordering of packets.  Structural metadata for things like drop-out detection.  A defined 27MHz clock (actually, a counter) that synchronize amongst the various types of data (video, audio, etc).  Provision for multiple streams, both semi- independent and related, including video, audio and user-defined metadata. (The underlying use case for this capability originated with broadcast video, where ETTC 2015 – European Test & Telemetry Conference multiple camera angles and microphones might be broadcast over a single logical link.) 4.2. “User” Metadata One of the driving imperatives from the MISB was to eliminate “burned in” metadata, and replace it with machine-readable structures. By design, a system that permits arbitrary labels, or “keys” was selected, based on the “KLV” (Key – Length - Value) scheme adopted by digital broadcasters [1]. To avoid multiple keys being selected for the same logical entity, dictionaries are maintained (in particular, one by SMPTE and one by MISB); there is no requirement at the file level that any given key be used, but for interoperability the appropriate dictionaries should be employed. The presence of this comprehensive metadata has resulted in a sea-change in how the acquired video is archived. While previously it might have been cataloged by date and mission, now the whole metadata set can be indexed. So with a product like General Dynamics Mediaware’s “D-VEX” [2] system, an analyst can call up all video, from whatever source, that (for example) includes a particular latitude and longitude, which was shot in the past 6 months, from an altitude less than 20,000 feet. Critically, this capability is achieved by reading the file captured on board, and then building and maintaining a parallel database that provides positioning information (such as file name and offset) so that the specific location of the metadata can be found in the original file(s). There is no requirement to modify the source file, so the integrity of the original recording is preserved. 5. Evolving Onboard Storage Techniques 5.1. Lessons from Onboard Video It is clear that it would be relatively straightforward, in the abstract, to combine the digital video approach as used by the “D-VEX” software package, as described above, with files created by in the T&E realm. However, while this obviously has significant value for the processing of T&E datasets, it falls rather short in a number of areas. First, and possibly most importantly, it should be noticed that in the digital video case, what is being indexed is the metadata rather than the video itself (note that it is not uncommon to also index a thumbnail consisting of a reduced-size still image from the video, but this can be thought of as just another type of metadata). The metadata represents a significantly smaller volume than the related data (the video). But in the abstract T&E case, it is possible that one would want to index a significant proportion of each record, so it may well be more beneficial to simply important the onboard recording into a full-fledged RDBMS. So the likelihood is remote that comprehensive T&E recordings might achieve the same sort of dramatic improvements in data management that have been seen with video coupled with this sort of metadata. However, it is certainly possible, and it is eminently practical, to use the digital video metadata as a mechanism for embedding untraditional forms of metadata in a video stream. For example, a flight test monitoring ice build-up may employ a high definition video camera trained on the monitored aerodynamic surface. To the video being produced additional metadata can be added as custom “KLV” fields, effectively placing flight test measurands into the video stream, instead of placing the video into a flight test file format (such as Chapter 10). Which approach to take depends entirely on the circumstances of the test, and the infrastructure that exists to support the test program. Certainly, in many instances obtaining long-distance transport of a video stream might be easier to manage simply because of the presence of a mature broadcast market that routinely streams video around the world, via cable and satellite. But perhaps the most significant lesson from onboard airborne video collection is that there is significant value in designing a system where arbitrary metadata can be added to an otherwise fairly rigidly structured data stream. Such metadata could include diagnostic, statistical and “system health” data, but would generally be limited to metadata that more-or-less follows the data flow from the sources to the onboard recorder – -- this mechanism would be poorly suited for the configuration / setup data mentioned earlier. To avoid anarchy, discipline must be observed in the use of such metadata (e.g. the metadata must not be used to store data that has a predefined record format; and if a particular metadata key already exists, it must be used in favor of crafting a new, application specific key), but all of these issues are manageable: for new video metadata keys, the MISB claims to be able to approve a new key in a timeframe of the order of one week [3], which is the sort of agility this capability would need in order to ensure adherence to “the rules”. 5.2. Lessons from EDP At the end of the day, the primary objection to using an RDBMS is the performance penalty resulting from the increased complexity of database update operations compared to the sort of simple sequential operations used by the MIMO approach. Two strategies exist to address this objection: a) Factor into the design of the onboard storage system sufficient excess capacity that the performance penalty will be absorbed unnoticed. b) Alternatively, use a “lighter weight” database system, possibly one omitting capabilities such as the full relational data model. Considering the first point, it can be noted that recorder design has evolved over the past few generations so that architectures that were previously built precisely to deliver the required performance using “hard real-time” operating systems are now being replaced with systems that use general purpose hardware with carefully designed software. So it is not unreasonable to predict that onboard storage systems could be built to handle typical T&E applications. ETTC 2015 – European Test & Telemetry Conference But it is axiomatic that instrumentation systems are constrained by size, weight and power requirements. And therefore, for a given set of performance requirements the smallest, lightest and least power hungry design will generally be preferred over a more versatile one. But even if the performance issue is rendered moot, there is an additional issue with using a full-fledged RDBMS for the onboard data store: once a test is completed, it is typical that the acquired data be immediately transferred to a ground processing segment where it is merged with data from other tests. For this operation, using an RDBMS typically means that the data would have to be exported from the onboard database and re-imported into the consolidated one. This is of course a straightforward task, but it is almost always preferable to be able to use the original recording files, in much the same way as the “D-VEX” application can use the original video recording files. The second option is, for a number of reasons, rather more practical. One of the alternatives to a full RDBMS that has been considered is a hierarchical data format, specifically the HDF5 [4] package and the related NetCDF-4 [5] variant. 5.3. Hierarchical Data Formats HDF5, the related NetCDF-4 and their predecessors were designed by the scientific computing community: HDF5 at the National Center for Supercomputing Applications (NCSA) and several US Department of Energy laboratories, while NetCDF-4 originated with the Unidata organization at the University Corporation for Atmospheric Research (UCAR). While some of the design goals have less direct applicability to the T& E community than others (e.g., the abstraction of data types to allow portability across different computer architectures), the existence of a standardized format designed for large I/O volumes has significant value. This design background ensures that the format is well suited to acquiring time series data, such as from the sorts of experiments that major and national laboratories conduct: it is optimized for high-speed raw data acquisition. As government-funded developments, not only are the specifications available under a variety of “open source” license, but also published is a range of software libraries and tools to exploit the file formats. For the T&E community, a possible approach might be to use a single HDF5 “file”, where an HDF5 file is defined as a container for storing a variety of data, and contains two core types of objects:  Groups: a structure containing zero or more HDF5 objects, along with supporting metadata  Datasets: a multidimensional array of data elements, again with supporting metadata (Note that an HDF5 “file” is typically implemented as a system-level file, but can span multiple system files if required). Both HDF5 groups and datasets may have associated with them an attribute list, where an attribute is a user-defined HDF5 structure that provides additional information about an object. Obviously, the acquired data will be saved as a dataset, but perhaps less obviously the configuration metadata information (of the data sources and the overall system) could comfortably be represented in a separate dataset within the same HDF5 file. Attributes, meanwhile, provide a convenient mechanism to store the narrative and descriptive metadata (such as that traditionally stored in TMATS records and the like). Critically, this sort of metadata storage is orthogonal to the array of acquired data; this allows for the design of the metadata structures to be independent of the design of the data structures, thereby permitting the use of, for instance, the Chapter 10 record formats with enhanced metadata. Of course, the fact that an HDF5 file includes embedded structures to manage the various components of the format means that the process of writing into an HDF5 file will be more complex than writing to a simple flat file. However, it is equally true that writing to a flat file in a file system is more complex than writing to raw storage (such as would be the case with magnetic tape), so the decision boils down to what level of performance penalty is acceptable for a given level of enhanced functionality. One significant achievement of these hierarchical file formats is that information is stored in usable data types, and the types are described as part of the file format. So instead of having to know how to map a particular sequence of bits into a useful value (such as an integer), the file itself contains that information. As the data is transferred off the test vehicle, therefore, a fully coherent “package” containing:  the test data  the configuration metadata  the structural metadata  the descriptive metadata. All of this would then be exported intact. This has obvious advantage for the consumer of this data following the completion of the test. 6. Conclusion The emerging demand that onboard data storage support more sophisticated operations than simply writing (and occasionally reading) sequential data imposes a requirement for more sophisticated storage structures than are currently employed. The interest in onboard data mining suggests that file architectures that support fast ad hoc searching have significant value to the T&E design engineer. In-band and out-of-band metadata are also a key feature to be considered in onboard storage design. As the video exploitation community has demonstrated, adding metadata A key feature of the evolved data storage solution is broad support for comprehensive metadata at several different levels, including a level familiar to users of airborne digital video systems. ETTC 2015 – European Test & Telemetry Conference 7. Acknowledgements The author acknowledges with thanks the openness of the flight test community, in both industry and government across the civil and military marketplace, and their willingness to share not only the challenges of their work but also to entertain wide-ranging potential solutions to those challenges. The author would also like to thank his colleagues and predecessors at Ampex for the innovations that have helped make recording in harsh environments a solvable problem. 8. References [1] SMPTE 336-2007: Data Encoding Protocol Using Key-Length-Value [2] http://www.mediaware.com.au/D-VEX-video-exploitation, retrieved April 2015. [3] http://www.gwg.nga.mil/misb/faq.html#section3.5, retrieved April 2015. [4] http://www.hdfgroup.org/HDF5, retrieved April 2015. [5] http://www.unidata.ucar.edu/software/netcdf, retrieved April 2015. 9. Glossary EDP: Electronic Data Processing HDF: Hierarchical Data Format iNET: integrated Network Enhanced Telemetry I/O: Input/Output IRIG: Inter-Range Instrumentation Group KLV: Key-Length-Value MIMO: Master In, Master Out MISB: Motion Imagery Standards Board MPEG: Motion Picture Experts Group NCSA: National Center for Supercomputing Applications NGA: National Geospatial-Intelligence Agency NIMA: National Imaging and Mapping Agency OLTP: On-line Transaction Processing RDBMS: Relational Database Management System SMPTE: Society for Motion Picture and Television Engineers T&E: Test and Evaluation TMATS: Telemetry Attribute Standard UCAR: University Corporation for Atmospheric Research 1 Chapter 10 introduced a specific packet type with defined storage formats for this sort information in the 2011 version of the standard, but the point remains valid. 2 April 14, 1956: Ampex introduces videotape recorder at NARTB (precursor of the NAB) show in Chicago

Creative Commons Aucune (Tous droits réservés) Aucune (Tous droits réservés)
contenu protégé  Document accessible sous conditions - vous devez vous connecter ou vous enregistrer pour accéder à ou acquérir ce document.
- Accès libre pour les ayants-droit

ETTC 2015– European Test & Telemetry Conference Cabin Comfort Flight Tests Installation - ETTC 2015 GALIBERT Joël1 , PLO Aymeric1 , GARAY Stephane1 1: AIRBUS Operation SAS, Toulouse, France Abstract: The verification of the cabin comfort requests an important amount of temperatures (surface and ambient) and air velocity measurements in the cabin at seat level, dado panels, air outlets, floor, lining, in the crew rests at the bunks, in galleys at attendant seats, on the monuments. On A350 MSN2, an amount of 1800 measurements was requested by the design office specialists. The innovation in a wireless data transmission was to satisfy the request of the specialists by proposing different standard kits (seat, aisle, bunk, cockpit, dado panel, air outlet) in order to cover a great density of measurements in the complete cabin (~400 measurements) and to reach simplicity in installation and removal of the kits. The agility to perform a new configuration and to relocate the different kits, allows the specialists to complete the tests in minimizing the number of flight tests. The development of powerful visualization screens increased the simplicity of the real time analysis for the specialist performing the test (thanks to the transmission of the datas from the central receiver to the flight test engineer station). Keywords: Flight Tests, Wireless, Comfort, 1. Introduction A350 XWB cabin comfort, Simpler and faster. The resources used to test the A350 XWB’s cabin comfort – in other words the wellbeing of passengers and crew owing to the temperature and speed of air flow – have made a technological leap forwards. 2. Comfort measurements requested One thousand eight hundred ambient air and surface temperature and air speed tests are needed for a cabin comfort test campaign, held over two weeks. By comparison, the flight test installation – excluding cabin comfort – counts 2,000 measurement points for a campaign that lasts several months. 3. Innovation based on wireless FTI A transnational flight test project team proposed several innovations designed to meet the requirements of the A350 XWB test campaign. "We needed simpler and faster means," to perform the measurements requested by the design office for cabin comfort tests. The main innovation consisted in a wireless solution which replaced the wired installation for transmitting collected data. For the first configuration on board A350MSN002, we have installed 140 kits to collect ~400 measurements, around the windows, at the level of the air outlets, in the galley and crew rest areas, in the cockpit and on the seats. Each kit consists in one or more airspeed measurement probes and/or a temperature sensor. The seat kits, clipped on to tubes simulating the heat given off by passengers, are equipped with four sensors located above the head and at head, knee and foot level. As the kits are mobile and fitted with a transmitter, it’s easy to move them around the aircraft for each configuration depending on what tests are required as the flights are progressing. Figure 1: Seat kit clipped on PAX heat dummy and surface temperature kit around the window ETTC 2015– European Test & Telemetry Conference Each kit equipped with a small transmitter (10mW, frequency band used: ISM 868 MHz IEEE 802.15.4 standard, communication protocol TDMA) sends the data sequentially to a central box located in the centre of the cabin. Then data are merged (IENA protocol) and transmitted via Ethernet link to the main Flight test installation for being displayed in real time and recorded for post data processing. Figure 2: Wireless measurement chain 4. Real time analysis Data can be viewed in real time at the test engineer’s station, making possible the correlation of outputs from aircraft systems and the results in terms of cabin comfort, even while the aircraft is in flight. Figure 3: Powerful visualisation screens for better real time analysis 5. Other kits developed To improve agility, simplicity and speed, different standard kits have been defined to cover a great many of measurements and to easily perform new configurations by relocating the different kits. Another improvement deals with the development of a solution for fastening the kits in place, without leaving any marks on the cabin lining. Figure 4: Air outlet kit Figure 5: bunk kit Figure 6: Attendant seat and aisle kit ETTC 2015– European Test & Telemetry Conference 6. Outlook The reactivity has been demonstrated with the application of 4 configurations performed during the cabin comfort flight tests between March and October 2014 on A350XWB-900 msn2 development aircraft. As feedbacks from Flight test Engineers and design office specialists were very good, this Cabin comfort Flight test installation will be re-used on the next A350XWB-1000 program and the wireless will be extended to other flight test measurement chains (other domains). 7. Conclusion All together, these advances have strong positive consequences for time, money and quality. The kits are configured off-plane and the time taken to install them has been reduced by 60%. As for financial aspects, the wireless system allows savings up to 30% per measuring channel. The kits can be reused for other programs, generating significant savings. Last, from quality point of view, configuring the installation is fast and simple, measurement points in the cabin are traceable and flight tests optimised. But one of the most important things is that these innovations contribute to improving the cabin product’s maturity at the “Entry Into Service” in companies. 8. Acknowledgement We would like to thank Bremen, Hamburg and Toulouse AIRBUS teams who contributed to this project and brought it to success. 7. Glossary CCFTI: Cabin Comfort Flight Test Installation IEEE: Institute of Electrical and Electronics Engineers IENA: Test Installation for New Aircraft, in French “Installation d’Essais Nouveaux Avions” ISM: Industrial, Scientific and Medical TDMA: Time division Multiple Access

Situation

Centre des Congrès Pierre Baudis, Toulouse (France)

11, esplanade Compans Caffarelli
31000 Toulouse - France