Saliency Tracking-based Condition Monitoring for Electromechanical Actuators

DOI : You do not have permission to access embedded form.
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


Saliency Tracking-based Condition Monitoring for Electromechanical Actuators


Détails de l'article
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

Saliency Tracking-based Condition Monitoring for Electromechanical Actuators


378.01 Ko


Creative Commons Aucune (Tous droits réservés)





<resource  xmlns:xsi=""
        <identifier identifierType="DOI">10.23723/10638/20152</identifier><creators><creator><creatorName>Chris Gerada</creatorName></creator><creator><creatorName>Jesus Arellano-Padilla</creatorName></creator><creator><creatorName>Mark Sumner</creatorName></creator></creators><titles>
            <title>Saliency Tracking-based Condition Monitoring for Electromechanical Actuators</title></titles>
        <resourceType resourceTypeGeneral="Text">Text</resourceType><dates>
	    <date dateType="Created">Sun 1 Oct 2017</date>
	    <date dateType="Updated">Sun 1 Oct 2017</date>
            <date dateType="Submitted">Sat 24 Feb 2018</date>
	    <alternateIdentifier alternateIdentifierType="bitstream">79bb493e6bd4f078f6eabd401f5b924ffe0d78c7</alternateIdentifier>
            <description descriptionType="Abstract"></description>

Saliency Tracking-based Condition Monitoring for Electromechanical Actuators Jesus Arellano-Padilla, Chris Gerada, Mark Sumner University of Nottingham; University Park, Nottingham, UK;; Abstract This paper presents a monitoring scheme to determine and track the condition (in real time) of PM- servodrives intended for Electro Mechanical actuators (EMAs). The improved sensibility of the proposed scheme allows quick detection of incipient faults traditionally undetectable by conventional monitoring methods; this is the case for rotor demagnetisation and certain faults in the stator windings such as insulation degradation, and the presence of inter-turn short circuits. The proposed approach is suitable for the More Electric Aircraft (MEA) applications since can be used in drives operating in the low speed region including standstill conditions such as in the case of EMAs when controlling the aircraft control surfaces. A saliency tracking scheme based on High Frequency (HF) injection is used here not only to determine the condition of the drive but to track any change due to the presence of electromagnetic faults. Experimental results validate that the proposed scheme has a high sensitivity and quick response for the detection of inter-turn short circuits, including those at an early stage of development. I. Introduction With the increasing move to More Electric Aircraft there is a need to ensure safety critical equipment is not only designed to satisfy the conflicting specifications of low volume/weight, but to provide excellent performance/reliability at low costs. Service costs can be reduced by prioritising preventive maintenance especially when initial signs of performance degradation are observed. For highly complex systems such as EMAs, unattended incipient faults may degrade into major faults and propagate to other components which will degrade the overall performance of the actuator. Implementation of schemes for condition monitoring with an improved sensitivity for the detection of incipient faults is very desirable not only to guarantee reliability but help to reduce service costs. Early detection of electromagnetic faults such as in the case of inter-turn short circuits in the stator windings is one of the most difficult challenges for condition monitoring; this is because these faults (including incipient faults) easily pass unnoticed. Conventional methods fail to detect these problems due to poor sensibility since only rely on the detection of abrupt changes as a consequence of severe faults. This paper evaluates the detection of inter-turn short circuits in the stator windings of a PM machine intended for EMA applications. Fig.1 shows the evaluated machine which contains a multi-pole rotor with surface mounted magnets as shown in Finite Element (FE) representation. Saliency tracking (which is commonly associated to sensorless control operation) is used here for purpose of condition monitoring; the aim is to detect any change in the drive’s magnetic signature due to the presence of electromagnetic faults. In this paper the presence of inter-turn short circuits will be evaluated. Fig. 1: Machine designed for EMA applications; top: FE model, bottom: experimental rig The proposed scheme combines saliency detection and a Saliency Modulation Profile (SMP) [2] to expand the capabilities of the monitoring scheme not only for the detection of incipient faults, but to allow uninterrupted motoring even in the case of standstill operation; this is relevant since it is not possible at present by conventional schemes currently used. II. Saliency Tracking Concepts The tracking of anisotropic properties in PM machines has become a standard mechanism for sensorless control applications, particularly those in the low speed range, including standstill operation [1]. Sensorless drives are attractive since reliability is improved considering that a usually -fragile and expensive- rotor position sensor is not required for purposes of control; instead this information is estimated by a dedicated technique aimed to calculate the rotor position from the drive’s saliency characteristics [1]. A representation of the saliency tracking scheme used in this work is shown in Fig.2. HF voltage (1khz) dq  dq  PI PI         vi vi va vb vc Id Iq 2/3 3/2 Iq * Id* P VSI PM e ia ib ic I q * In1 In2 out e Pos* Saliency tracking BP filter Esaliency Healthy Warming 1 Warming 2 Fault 1 Fault 2 Fault 3 Pos Pos Diagnostics Pos* out In1 In2 Condition monitoring scheme High frequency injection signal Fig. 2: Saliency tracking based condition monitoring For the case of EMA drives where a surface-mounted PM machine is expected, the relative constant effective airgap length results in a geometrically symmetric structure (see top of Fig.1); this produces lower saliency levels respect to rotors with buried magnets [2]. However saliency is usually enhanced by saturation in the stator core since it is the main flux that causes the spatial anisotropy that modulates leakage inductances; which at the same time modulates the overall HF stator impedance. A) High Frequency Injection Scheme HF injection is based on the principle that for a high frequency voltage applied to the machine terminals (carrier signal c), the leakage inductances (L) will be modulated by the spatial anisotropy produced by the rotor geometry and saturation levels [3]. For the case considered in this work, a - injection scheme was selected due its simplicity; in this case a rotating low magnitude vector (which can be observed in the upper section of Fig.2 by vi and vI) is applied to the machine terminals in feed forward connection to the inverter after the current controllers (see Fig.2). According to [3] the resulting currents denominated as ic, can be represented by equation (1); where uc is the magnitude of the injected signal, Ld and Lq are the d-q transient leakage inductances respectively with L = (Ld + Lq)/2 and L = (Ld - Lq)/2 [4].     r c c t j t j q d c c c Le e L L L u i        2      icp + icn (1) As observed from (1), ic comprises of two components: a negative sequence denominated as icn, and the positive sequence icp. Note that it is icn who carries information about the saliency angle (r), therefore icp should be removed [4]. In this work a homodyne demodulation technique is selected, which consists of a synchronous filter tuned at the carrier signal c; this demodulator (which is represented in Fig.1 as “saliency tracking”) is depicted in Fig.3. BP Filter Synchronous filter Ipos I c HP Filter 1/S I c I c I c j(-ct) e j(ct) e c c  POS  POS Fig. 3: Demodulation of the carrier currents After demodulating the HF currents, the saliency components are obtained which according to [4] are:   r j q d c c e L L L u Pos        2 (2) Note that two saliency components (Pos and Pos) are obtained; their magnitude and waveform are determined by the carrier signal (uc and c), and most importantly by the saliency-modulated leakage inductances (Ld, Lq), while L has a monor effect. B) Space Modulation Profile Detection of faults in the windings such as insulation degradation or demagnetization is a complex task especially at early stages of development. We propose a combined technique between saliency tracking and a Saliency Modulation Profile (SMP) [5] to enhance the detection capabilities. This technique requires a “healthy profile” of the drive operating under healthy conditions which is stored in a lockup table for later use. The healthy profile for the machine under consideration is shown in Fig.4. Fig. 4: Saliency profile for a healthy machine; top: Pos*, bottom: Pos* In order to improve sensibility of the apprach and to extend performance for standstill conditions, the measured saliency (Pos) is compared against the healthy profile (identified now as Pos*). Any relevant differences between them are combined to get an indication of a fault (explained below). Fig.5 shows the procedure to obtain the SMP scheme; note that due to space constrains technical details are maintained to a minimum, however full details can be found at [5] and in the full version of this paper. i / ni Pos_f 2Fe Pos_ave Pos* i / ni Pos Pos_f Non-casual filters tuned at expected at 2Fe 2Fe P r Iq* Pos_ave Pos* Pos Saliency cycle averaged interval: i=1 to 256 Lookup table Pole pairs Rotor position Iq demand (from controller)  Fig. 5: Procedure to obtain SMP C) Detection of the saliency error The procedure to determine the condition of the drive by using the SMP scheme can be observed in the lower part of Fig.2. Note that the measured saliency (Pos, Pos) is compared against the ideal saliency (Pos*, Pos*) and that any difference is used to determine the drive’s condition. Note that Pos*, Pos* are obtained from a lookup table according to two inputs: r which requests information about the expected saliency for a particular rotor position; and Iq*, which is used to provide information about the load condition. These errors are combined to provide Esaliency which is used as an indication of the error in saliency respect to a healthy system. For the case of a healthy drive Esaliency  0 is expected while for a faulty case a significant variation is expected. III. Experimental evaluation: Detection of inter- turn short circuits This section contains experimental results for the proposed scheme in the presence of inter-turn short circuits intentionally applied to the machine windings. The evaluated machine (shown in Fig.1) consists of a surface mounted PM machine with 14 pole pairs specially designed for EMA applications. The machine has links to turns in the stator winding brought out, so different combinations of winding faults can be applied. Each winding consist of 4 coils with 46 turns each connected in series (total number of turns per phase is 184). The machine will be rotated to very low speed (Fe=2Hz) and at standstill conditions to demonstrate that the scheme is able to determine the condition of the drive under these conditions (also to reduce the amount of induced current in the short circuited turns). Fig.6 shows a comparison for Pos against the Pos* for two conditions: a) healthy system (shown in Fig.6a), and a faulty system where an inter-turn short circuit of 12 turns has been applied to one of the coils of phase A (the latter shown in Fig.6b); note that Pos is very similar to Pos* for the case of a healthy system (as expected) while changes considerably in the presence of the applied fault. 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5 Time Saliency Pos and Pos Pos* Pos Pos Pos* Pos* a) 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5 Time Saliency Pos and Pos Pos* Pos Pos* Pos b) Fig. 6: Comparison of saliency components for: a) healthy; b) faulty (12 turns s.c. applied) As it was stated, this is scheme is suitable for drives operating under standstill conditions where the rotor position is maintained independently of the load; this is demonstrated in Fig.7. Note that Pos are kept constant (as expected in standstill) and that change considerably in the presence of a fault (again, a 12- turns s.c. has been applied). Note that Pos gets back to healthy values when the fault is removed. 0 0.5 1 1.5 2 2.5 3 3.5 4 -0.15 -0.1 -0.05 0 0.05 0.1 0.15 0.2 0.25 0.3 Pos Pos Time Saliency Pos and Pos Faulty (12 turns S.C.) Healthy Healthy Fig. 7: Saliency components at standstill when a fault (12 turns s.c.) is applied and removed Fig.8 shows Esaliency for the measurements shown in Fig.6. Note that for a healthy system Esaliency is very close to zero as expected and that vary considerably when the fault is applied (12 turns s.c. applied). The bottom of Fig.7 shows the Loci plot which shows Esaliency respect to the rotor position (Loci plots are relevant since can provide information of the fault’s location within the stator). E saliency 0 1 2 3 4 5 6 7 8 0 0.05 0.1 0.15 0.2 0.25 Time Healthy Faulty (12 turns S.C.) a) 0.05 0.1 0.15 0.2 30 210 60 240 90 270 120 300 150 330 180 0 Esaliency 0.0 Healthy Faulty (12 turns S.C.) b) Fig. 8: Transition of Esaliency: healthy to faulty system It was stated that Esaliency can be used not only to detect the presence of faults, but to determine their magnitude; this is evaluated in Fig.9 where two fault conditions are evaluated. First, a one-turn s.c. was applied to one of the coils. Note that this can be assumed as an incipient fault considering that is very small respect to the total number of turns (i.e., 1/184). The second case corresponds to the previously considered fault (12-turns s.c: 12/184). Both faults are compared agaist a healthy system as a reference. Note that for the case of a single turn s.c., a significant variation in Esaliency is observed respect to a healthy system (this despite being considered as a incipient fault). The variation of Esaliency for the case of the 12 turns s.c. is significantly higher respect to one single turn which demostrates than Esaliency can be used to determine the magnitude of the fault. 0.05 0.1 0.15 0.2 30 210 60 240 90 270 120 300 150 330 180 0 0.0 Esaliency Healthy Faulty (1 turns S.C.) Faulty (12 turns S.C.) Fig. 9: Comparison of Esaliency for a healthy and faulty system (1- and 12-turns s.c. fault applied) Conclusions This paper has shown that saliency tracking (normally associated to sensorless control applications) can be used as a mechanism to implement condition monitoring in drives intended for EMA applications. In this paper HF injection is combined with a SMP to implement a monitoring scheme with improved detection capabilities. A major advantage is that the proposed approach can be used in drives operating in the low speed range and during standstill conditions. The presented experimental results validate the performance of the scheme not only to detect the presence of several electromagnetic faults, but to determine their magnitude. The approach was able to detect the presence of an incipient fault represented by one single turn short applied in a high number of turns per phase coil. References 1 P.L. Jansen, and R.D. Lorenz, “Transducerless position and velocity estimation in induction and salient AC machines,” IEEE Trans. Ind. Applicat.,1995, 31, (2), pp. 240-247. 2 J.F. Gieras, M. Wing, “Permanent magnet motor technology: design and applications,” Electrical and computer engineering. Marcel Deckker Inc, New York Basel, 2nd edition, 2002. 3 C.A. Silva, G.M. Asher, M. Sumner and K.J. Bradley, “Sensorless rotor position control in a surface mounted PM machine using HF voltage injection,” in Proc. EPE- PEMC’02. 4 C.A Silva, “Sensorless vector control of Surface mounted permanent magnet machines without restriction of zero frequency,” Ph.D. Dissertation, Dept. Electrical & Electronic Engineering, University of Nottingham, 2003. 5 J.Arellano-Padilla, M. Sumner and C. Gerada, “Winding condition monitoring scheme for a PM machine by using HF injection, ” IET Electric Power Applications Journal, 2011, vol.5, Iss.1, pp.89-99.