Experimental Study of Low Speed Sensorless Control of PMSM Drive Using High Frequency Signal Injection

Transcription

1 VOLUME: 4 NUMBER: 6 MARCH Expeimental Study of Low Sensoless Contol of PMSM Dive Using High Fequency Signal Injection Jyoti AGRAWAL, Sanjay BODKHE Depatment of Electical Engineeing, G. H. Raisoni College of Engineeing, CRPF Gate No.3, Hingna Road, Digdoh Hills, Nagpu, Mahaashta - 446, India Depatment of Electical Engineeing, Shi Ramdeobaba College of Engineeing & Management, Ramdeo Tekdi, Gittikhadan, Katol Road, Nagpu India jyotigovindagawal@gmail.com, bodkhesb@knec.edu DOI:.5598/aeee.v4i.564 Abstact. Conventional techniques fo sensoless contol of pemanent magnet synchonous moto dive (PMSM), which equies infomation on oto position, ae eviewed, and ecent developments in this aea ae intoduced in this pape along with thei inheent advantages and dawbacks. The pape pesents an impoved method fo sensoless speed contol of PMSM dive with emphasis placed on signal injection method. This signal injection method examines the contol pefomance of sensoless PMSM dive by injecting signal extenally and theeby sensing the oto position. The main objective of this dive system is to have speed contol at standstill and low speed egions. Seveal tests ae caied out to demonstate the ability of poposed models at diffeent opeating conditions with the help of simulation esults in Matlab/Simulink envionment. Simulation esults confim that the poposed sensoless contol appoach of PMSM can achieve high pefomance at standstill and low speeds but not at vey high speeds. An expeimental setup is implemented using a HP suface mounted (SM) PMSM and dspicdem T M MCHV- development boad, to check the validity of simulation esults. Keywods DsPICDEM T M MCHV- development boad, high-fequency signal injection, MAT- LAB/Simulink, MPLAB X IDE, PMSM dive, sensoless, vecto contol.. Intoduction Pemanent Magnet Synchonous Moto (PMSM) dives have been used widely due to its numeous advantages such as highe efficiency, highe powe facto, ugged constuction, eliable opeation, high degee of contol flexibility, high toque to inetia atio, high toque to cuent atio, etc. These advantages have attacted the inteest of eseaches and industy fo industial dive applications [], [], [3]. Fo the implementation of any contol stategy two cuent sensos and knowledge of oto position is equied. In most vaiable speed dives an electo-magnetic esolve o optical encode is fitted to the oto shaft fo this pupose. Though this method is pecise, the pesence of this shaft senso intoduces exta cost, deceases eliability and highe numbe of connections, thus making the total system cost vey non-competitive compaed to othe types of moto dives. Hence, all of the above limitations make elimination of these devices vey desiable []. Fom liteatue, it is appaent that thee exist diffeent techniques elated to the sensoless contol of PMSM []. An oveview of the conventional sensoless contol techniques fo PMSM is pesented in Fig.. Sensoless techniques ae classified into thee classes as those using the fundamental excitation models o model of the machine, those using saliency and signal injection methods and atificial intelligence methods []. This pape pesents a pefomance of oto position sensoless contol of PMSM dive in which the oto position is sensed by injecting extenal high fequency (HF) voltage signal in otating q-and d-axes [4], [5]. With this scheme the difficulties in sensing the oto position at standstill ae eased. This pape tests the poposed method by vaious simulations and c 6 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 9

2 VOLUME: 4 NUMBER: 6 MARCH finally some useful conclusions ae dawn. The desied pefomance is achieved by implementing vecto contol technique in Matlab/Simulink and tested unde diffeent opeating conditions [6]. The main aim of this dive system is to have a closed loop speed contol at low speed ange without using a speed tansduce and the simulation esults demonstate that the poposed stategy is effective. This pape is oganized into seven sections: Section. povides a geneal intoduction and liteatue eview of the PMSM and a study of vaious existing sensoless contol stategies with focus on the HF signal injection fo the extaction of oto position. In Section., the simplified dynamic model of the PMSM is developed and is shotly examined. In Section 3. the signal injection technique used fo extaction of oto position is pesented. Section 4. pesents a newly developed position sensoless speed contolled PMSM dive. In Section 4.8. detailed simulation esults unde vaious opeating conditions ae pesented emphasizing the meits and demeits of the method. Expeimental setup and esults ae pesented in Section 5.. The pape is concluded in Section 6... PMSM Model Sensoless Contol of PMSM the system. With some assumptions, the d- and q-axes stato voltages in oto efeence fame ae [6]: [ ] [ ] [ ν ds RS L = d p ω i ds ω L d R S p ν qs i qs ] [ ω λ af ]. () Fom Eq. (), it is obseved that the voltage equations ae equal to the poduct of the impedance matix and the cuent vecto, with an additional component due to the motionalelecto-motive foce (emf) of the oto flux linkages. The electomechanical dynamic equation is given by Eq. (), T e = J dω m dt T Bω m. () The dqo cuents ae obtained fom abc cuents using Pak tansfomation as Eq. (3): In a balanced thee-phase system, the sum of thee phase cuents is zeo. This is given by: i as i bs i cs =. (4) Fom Eq. (4), it is seen that if two cuents ae measued, the thid-phase cuent can be econstucted fom the othe two phase cuents [7]. This eliminates the need fo additional cuent sensos.in ode to have a meaningful intepetation in the modeling, analysis and simulations, the powe input to thee phase machine has to be equal to the powe input to two phase machine. Fom Eq. () and Eq. (), dynamic model of PMSM in oto efeence fame is deived and is expessed by Eq. (5). Fundamental excitation models NonAdaptive methods Atificial Inteligence methods Saliency and Signal injection methods. HF Signal injection methods. LF Signal injection methods 3. Implementation of Sensing Roto Position by Extenal HF Rotating Voltage Signal Injection. Estimatos using monitoed stato voltages/cuents. Flux based position estimation 3. Back e.m.f. based position estimatos Estimation based on Model Refeence Adaptive System (MARS) Adaptive Methods Obsevedbased Estimatos. Luenbege Obseve. Reduced Ode Obseve 3. Sliding mode obseve (SMO) 4. Extended Kalman filte Fig. : Common sensoless contol techniques fo PMSM. The detailed modeling of thee-phase PMSM dive system is equied fo pope simulation and analysis of The basic scheme in signal injection is to geneate a evolving voltage phaso by applying thee-phase voltages at a signal fequency that is diffeent and mostly highe than fundamental fequency [8]. This method has been used fo extacting oto position using an obseve. In stationay efeence fame, injected voltages at fequency ω i ae given as: whee: ν asi sin θ i ν bsi = V i sin ( ) θ i π 3 ν csi sin ( ), (6) θ i π 3 θ i = ω i t. (7) c 6 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 3

3 VOLUME: 4 NUMBER: 6 MARCH i qs cos θ cos ( ) θ π i 3 ds = sin θ sin ( ) θ π 3 i cos ( ) θ π 3 sin ( ) θ π 3 i as i bs. (3) i cs i qs p i ds ω = θ Rs L d ω λ af H L d ω Rs λaf L d (L d )i qs H B H i qs i ds ω θ L d ν qs ν ds T. (5) Tansfoming the voltages into q- and d-axes in the same efeence fame as: [ ] [ ] νqsi sin θi = V ν i. (8) dsi cos θ i The injected flux linkage vecto in estimated oto efeence fame can be given by: [ ] [ ] λqsi cos θe = V λ si sin θ i. (9) dsi sin θ e The injected fequency component of the stato cuents can be obtained as: [ ] [ ] iqsi I cos θ = sin θ e I cos { (θ θ e )} i i, dsi I sin θ e I sin { (θ θ e )} () whee: ) I = V i ω i I = V i ω i ( LqL d ( LqL d ( Lq L d ( LqL d ) ( ), () Lq L d ) ) ( Lq L d ). () Tansfoming the d-axis cuent to the estimated oto position efeence fames yields [9]. i e dsi = i qsi sin (θ e ) i dsi cos (θ e ). (3) Equation 3 can be ewitten as: i e dsi = I sin θ i [sin { (θ θ e )}]. (4) Fom Eq. 4 it can be seen that it contains the infomation elated to oto position eo. This useful signal can be extacted using a bandpass filte (BPF) which sepaates the fundamental component fom the HF component followed by an obseve. This position eo can be demodulated by any demodulation schemebut in this model heteodyning pocess is usedfo the estimation of position and speed [5]. 4. Implementation of -Contolled Sensoless PMSM Dive by Estimating Roto Position Using Signal Injection The poposed model developed fo sensoless vecto contol of PMSM dive using signal injection, as pesented in Fig., is implemented using the softwae Matlab/Simulink. It consists of a speed loop, cuent and toque angle calculato, stato cuent synthesize, invete with SPWM cuent contol, PMSM model, HF signal injection, demodulation pocess, contolle and obseve fo oto position and speed estimation. Each functional block is elaboated in detail in the following sections. The pinciple fo this speed contol stategy is based on compaing the efeence speed and the estimated speed []. The speed eo thus obtained is then pocessed though the speed contolle and the output of this block epesents the efeence toque (Te ) fo the cuent and toque angle calculato. The stato cuent synthesize geneates the phase efeence cuent commands. The thee phase efeence cuents ae compaed with thei espective actual cuents esulting in the cuent eos in a PWM cuent contolle which is then used to geneate the switching signals fo the invete. In the poposed scheme the feedback signal i.e. oto position (θ e ) is estimated using extenal signal injection scheme. Once the oto position is estimated, the oto speed (ω e ) can be estimated by using Eq. (6) []. The inability with most of the methods to povide accuate oto position estimation at low speeds, which diectly affects the contol pefomance, would not be a poblem with this technique []. 4.. Contolle The speed eo between command and estimated values of speed is pocessed in the popotional and intec 6 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 3

4 VOLUME: 4 NUMBER: 6 MARCH ω * ω e ω _e Regulato ω e V i ω i T * e Cuent and Toque angle calculato High Fequency Voltage Injection i * s δ * Stato Cuents Synthesize i E as i E bs i E cs θ e i e i bs Sinusoidal PWM Technique i as Gate Signals V dc 3-Phase Invete v E as v E bs v E cs abc/dq θ e v v qsi qs v dsi v ds PMSM Load θ e /s ω e Obseve Contolle the_e Demodulation qsi i e dsi Estimated oto efeence fame tansfomation θ e Fig. : Oveall schematic of the sensoless vecto contol fo a speed contolled PMSM dive by using extenal high fequency signal injection scheme. gal (PI) speed contolle to geneate a efeence toque []. The main aim of the speed contolle is to contol the electomagnetic toque so that the speed eo can be minimized The d- and q-axes Stato Cuent in Roto Refeence Fame By using Eq. (5), the d and q-axis model is constucted which is shown in Fig Electomagnetic Toque The expession fo the electomagnetic toque (T e ) developed on the oto can be obtained fom the input powe and its vaious components as given by Eq. (5) [3]. T e = 3 P [(L d ) i ds λ af ] i qs. (5) 4.3. SPWM Cuent Contolled Invete Model Figue shows that the PMSM is fed fom a voltage souce invete with cuent contol. The switching fequency is usually fixed at caie fequency (f c ). The switching signals fo the powe devices ae detemined by the intesection of a tiangula caie wave of desied switching fequency with eo of the contolled signal obtained fom the efeence and actual phase cuents. The compaison will esult in a signal that will demand the phase voltage to follow in such a way that the cuent eo of the espective phase is educed to zeo [4] Ceation of HF Revolving Voltage Vecto Injection A balanced set of thee phase voltage vecto given in Eq. (6) with low amplitude (V i ) and high fequency fiae tansfomed into d- and q-axes by using Eq. (8). These voltages ae tansfomed into the estimated position efeence fames (efe Fig. 4) and then added to the contolled dq output voltages Demodulation Pocess Figue 5 consists of BPF which aims at sepaating the injected signal fom the fundamental component. In ode to extact oto position eo needed fo the obsevethe BP filteedsignal i.e. Eq. (5) is demodulated and then multiplied by e j(ωit) [5]. This signal is then cleaned by a low-pass filte which emoves the second hamonic component leaving the desied eo tem only. The oto position thus can be estimated fom this signal using an obseve [6]. Simulink model developed fo this pupose is shown in Fig Contolle and Obseve In ode to extact oto position and speed fom the position eo a tacking obseve is equied. The intenal stuctue of the contolle is developed in c 6 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 3

5 VOLUME: 4 NUMBER: 6 MARCH v qs v ds / R s l amaf Ld /L d Rs i qs i ds ω Fig. 3: Dynamic model of PMSM in oto efeence fame. V i θ i Equation (6) v asi v bsi v csi θ i ω /s ie *pi*f i V i sin θ i /ω i abc/dq V i cos θ i v e qsi ω e v e dsi Fig. 4: Common Revolving voltage injection scheme PMSM. i dsi i qsi sin(θ e ) cos(θ e ) e idsi BP Filte sin θ i LP Filte the_e opeating conditions. The plots of speed, toque, cuents, oto position and position eo ae given. The plotted vaiables ae in nomalized units (p.u.). The test esults of the sensoless contolled PMSM dive employing HF signal injection scheme fo five diffeent opeating conditions ae shown in Fig. 7, Fig. 8, Fig. 9, Fig.. The paametes fo PMSM used in the simulation ae given in the appendix. Sampling time is set to e 6 s. The ated speed of 68.6 ad s is selected as base speed. The injected fequency (f i ) is khz and the amplitude (V i ) is V. In ode to veify the accuacy of poposed estimated method and also to study the esponse of position sensoless dive unde diffeent dynamic conditions it is impotant to compae the two diffeent methods. In the fist method the oto speed and position ae estimated using the poposed model as descibed in Fig. while in the second method, the same machine vaiables i.e. oto speed and position ae calculated using the dynamic model of PMSM which is teated as actual model Fee Acceleation Chaacteistics In this test, the moto dive is allowed to acceleate fom ad s to ated speed i.e ad s ( p.u.) consideing % of the ated load. Fom Fig. 7(a), it is concluded that though the estimated oto speed matches the efeence speed nealy at.9 s. The simulation esultsincluding a compaison of the actual and estimated oto angle ae shown in Fig. 7(d), while the position eo is plotted undeneath it. As obseved fom the Fig. 7(e), thee is no appeciable deviation in oto position tackingas long as the oto speed is not vey high, till the coect oto position is detected i.e. afte about.9 s. Fig. 5: Implementation of demodulation pocess. Simulink to pefom this function (see Fig. 6. The detected oto position eo can be used fo estimating the oto position and speed []. Once the oto speed is estimated accuately, the oto position can be obtained by using an integato. The electomagnetic toque signal is added at the output of contolle fo faste dynamics []. Fig. 6: Simulink model of oto position and speed estimatousing oto position eo tacking obseve Simulation Results and Discussion A sensoless speed-contolled PMSM dive has been developed in MATLAB envionment to validate the poposed speed and position estimation method. The functionality of the implemented sensoless contol method is veified by testing the dive unde diffeent 4.. Low and Standstill Opeation One of the biggest challenges in sensoless vecto contol of PMSM is estimating oto position at low speed including standstill in which most of the methods fail. In this test the speed contol pefomance at low speed including standstill unde no load condition is shown c 6 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 33

6 VOLUME: 4 NUMBER: 6 MARCH Stato Phase Cuents ias ibs ics.5 ef estimated actual (a) (c) Electomagnetic Toque ef estimated 4 Roto Position estimated actual (b) (d) 4 Position Eo (e) Time Fig. 7: Fee acceleation chaacteisitc: (a) Refeence, estimated and actual oto speed esponse, (b) Toque esponse, (c) Stato phasecuents esponse, (d) Estimated and actual oto position, (e) Roto position eo (δθ). Stato Phase Cuents. ef estimated actual.5 ias ibs ics (a) (c) Electomagnetic Toque ef estimated (b) Roto Position estimated actual (d) Position Eo (e) Time Fig. 8: Step change in efeence speed fom ad s to ad s at.75 s and back to ad s at.5 s unde no load: (a) Refeence, estimated and actual oto speed esponse, (b) Toque esponse, (c) Stato phase cuents esponse, (d) Estimated and actual oto position, (e) Position eo (δθ). in Fig. 8(a). The speed command is vaied fom ad s to ad s at.75 s and back to ad s at.5 s in step manne as it is epesented in Fig. 8(a). The oto speed is acceleated fom standstill to ad s stating fom t = s so that the moto successfully uns at the efeence speed. Figue 8(c) pesents that the stato phase cuents esponse and Fig. 8(b) pesents the toque esponse which is just sufficient to ovecome fiction except at tansitions of the shap speed vaiations (see Fig. 8(a)) [7], [8]. 4.. Step Change in Refeence Figue 9 shows the simulation esult fo speed contol with a step change in efeence speed at ated load. At.5 s, the efeence speed is step changed fom 5 ad s to 5 ad s, then ad s at. s and finally set to 5 ad s at.5 s. As obseved fom the Fig. 9(d) the estimated oto position wavefom follows the actual oto position. Figue 9(e) shows the wavefom of eo in oto position i.e. the diffeence between the estimated and actual oto position (δθ = δ θ e ). Fom the plot of oto position eo it is seen that the eo is diven to zeo which indicates how well the position estimation scheme 4.. Revesal Opeation In this test, the moto dive coves both the diection of otation. The speed efeence is step changed fom 375 ad s to 375 ad s at. s. Fom Fig. (a) and Fig. (b) it is seen that with a positive speed command the electomagnetic toque is diven to posc 6 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 34

7 VOLUME: 4 NUMBER: 6 MARCH ef estimated actual (a) Stato Phase Cuents ias ibs ics (c).5.5 ef estimated Electomagnetic Toque (b) Roto Position 4 estimated actual (d) 4 Position Eo (e) Time Fig. 9: Step change in efeence speed fom ad s to ad s at.75 s and back to ad s at.5 s unde no load: (a) Refeence, estimated and actual oto speed esponse, (b) Toque esponse, (c) Stato phase cuents esponse, (d) Estimated and actual oto position, (e) Position eo (δθ). Stato Phase Cuents.5 ef estimated actual (a). ias ibs ics (c) Electomagnetic Toque ef estimated 4 Roto Position estimated actual (b) (d) 4 Position Eo (e) Time Fig. : speed changes fom 375 ad s to -375 ad s at. s unde no load: (a) Refeence, estimated and actual speed, (b) Toque, (c) Stato cuents, (d) Estimated and actual position, (e) Position eo (δθ). itive maximum and is maintained thee until actual speed matches the efeence speed (ω) [5]. When the estimated speed matches the efeence speed, the toque comes down to match the load toque and the fiction toque [9]. With a negative speed command the electomagnetic toque is diven to negative maximum, which causes the evese otation. The speed eo causes ipples which ae not even noticeable (see Fig. (c)) [9]. 5. Expeimental Setup and Results DEM MCHV- development boad manufactued by Micochip Technology Inc. The caie fequency of the high speed PWM module is set to khz. The thee-phase bidge invete with a powe ating of 4 V / 6.5 A has six sets of IGBTs. The eal test bench employed fo sensoless contol of PMSM is pesented in Fig. The poposed algoithm fo speed estimation is developed using MPLAB X IDE. The test moto is a Y-connected, HP, 4.5 Amp, 4 pm SMPMSM with detailed paametes enumeated in Table (). PMSM is equipped with built-in position encode used only fo monitoing the estimated speed not fo speed contol. To suppot the simulation esults, the complete poposed scheme is implemented and tested using a dspic- c 6 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 35

8 VOLUME: 4 NUMBER: 6 MARCH POWER ENGINEERING AND ELECTRICAL ENGINEERING Tab. : Rating and Paametes of a HP PMSM. Rating Symbol Quantity N f Rated speed Rated fequency Rated line-line ms voltage Rated ms cuent Rated toque Paametes VL L IRM S TL Symbol Quantity P Ls Rs Pole pais Stato winding inductance Stato winding esistance KE Back-EMF constant Pemanent magnet flux-linkage Toque constant Peak toque Quadatue encode λpm Kt Tmax PPR 5.. Value and units 4 pm 5 Hz 4 V 4A 6 N-m Value and units 3.59 (H) 3.83 (Ω) 5.6 (Vpk krpm ) Fig. : Refeence speed changes fom pm to 3 pm at 4 ms, then to 5 pm at 47 ms and finally back to pm at 9 ms unde no load (Refeence speed)..887 (Wb).8554 (Nm A ) 6 N-m 48 PPR Step Change in Refeence Figue, Fig. 3, Fig. 4 and Fig. 5 shows the expeimental esults fo sensoless speed contol of PMSM with a step change in efeence speed command (efe Fig. )unde no load config. 3: Refeence speed changes fom pm to 3 pm dition. At 4 ms, the efeence speed is step at 4 ms, then to 5 pm at 47 ms and finally back changed fom pm to 3 pm, then fom to pm at 9 ms unde no load (Actual oto speed). 3 pm to 5 pm at 47 ms and finally set to pm at 9 ms. As obseved fom the Fig. 4 the estimated oto position wavefom tacks the actual oto position shown in Fig. 3 vey nicely. It is also clea fom the wavefom of oto speed eo as shown in Fig. 5, how well the poposed scheme handles this type of efeence speed command. SMPMSM DSO Host Compute LCR Mete MCHV Development Boad Fig. 4: Refeence speed changes fom pm to 3 pm at 4 ms, then to 5 pm at 47 ms and finally back to pm at 9 ms unde no load (Estimated oto speed). 5.. Powe QEI Cable Cable High Voltage Diffeential Pobe USB/Debug Cables Fig. : Photogaph of expeimental test bench. Isolated QEI Cicuity Low Opeation Figue 6, Fig. 7, Fig. 8, Fig. 9 shows the expeimentalesults fo efeence speed command of 4 pm unde no load. Fom the wavefom of oto speed eo pesented in Fig. 8, it is obseved that thee is an eo of appoximately pm. c 6 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 36

9 VOLUME: 4 NUMBER: 6 MARCH Fig. 5: Refeence speed changes fom pm to 3 pm at 4 ms, then to 5 pm at 47 ms and finally back to pm at 9 ms unde no load (Roto speed eo). Fig. 6: Low speed opeation fo efeence speed of 4 pm unde no load (Actual oto speed). Fig. 7: Low speed opeation fo efeence speed of 4 pm unde no load (Estimated oto speed). Fig. 9: Low speed opeation fo efeence speed of 4 pm unde no load (Estimated oto position). 6. Conclusions The mathematical modeling fo analysis of sensoless vecto contolled PMSM dive and its implementation employing HF signal injection using MAT- LAB/Simulink has been pesented in this pape. It is concluded fom the liteatue that many sensoless scheme cannot be used at low and zeo speed whee back EMF signal is not available.to achieve opeation at such speed the signal injection scheme can be moe efficient than any othe sensoless method. Finally it can be concluded fom the simulation and expeimental esults that the signal injection stategy appeas to be good at low speed and standstill but it cannot give good pefomance at high and vey high speed. Pefomance of sensoless PMSM is investigated unde vaious opeating conditions to veify the validity and feasibility of the poposed model.it must be noted that the HF signal injection method can beextended to a wide speed ange but not by inceasing the injection fequency as it may stat ovelapping with the invete s switching fequency. But the poblem faced by this scheme is, it equies obseve fo successful opeation. Futhemoe, HF signal injection technique is obust to paamete vaiations. Fig. 8: Low speed opeation fo efeence speed of 4 pm unde no load (Roto speed eo). c 6 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 37

10 VOLUME: 4 NUMBER: 6 MARCH Refeences [] VAS, P. Sensoless Vecto and Diect Toque Contol. st. ed. Oxfod: Oxfod Univesity Pess, 998. ISBN [] KRISHNAN, R. Pemanent Magnet Synchonous and Bushless DC Moto Dives. st. ed. Floida: CRC Pess,. ISBN [3] BOLDEA, I. and S. A. NASAR. Electic Dives. nd. ed. Floida: CRC Pess, 6. ISBN [4] TOUDERT, O. M., H. ZEROUG, F. AUGER and A. CHIBAH. Impoved oto position estimation of salient-pole PMSM using high fequency caie signal injection. In: IEEE Intenational Confeence on Mechatonics. Vicenza: IEEE, 3, pp ISBN DOI:.9/ICMECH [5] CORLEY, M. J. and R. D. LORENZ. Roto position and velocity estimation fo a slient-pole pemanent magnet synchonous machine at standstill and high speeds. IEEE Tansactions on Industy Applications. 998, vol. 34, iss. 4, pp ISSN DOI:.9/ [6] PILLAY, P. and R. KRISHNAN. Modeling, simulation, and analysis of pemanent-magnet moto dives, Pat I: The pemanent-magnet synchonous moto dive. IEEE Tansactions on Industy Applications. 989, vol. 5, iss., pp ISSN DOI:.9/ [7] MISHRA, A., J. A. MAKWANA, P. AGARWAL and S. P. SRIVASTAVA. Modeling and implementation of vecto contol fo PM synchonous moto dive. In: Intenational Confeence on Advances in Engineeing, Science and Management (ICAESM). Nagapattinam: IEEE,, pp ISBN [8] CHEN, Z., X. DENG, K. HUANG, W. ZHEN and L. WANG. Sensoless contol of wound oto synchonous machines based on highfequency signal injection into the stato windings. Jounal of Powe Electonics. 3, vol. 3, iss. 4, pp ISSN DOI:.63/JPE [9] LI, Y., Y. ZHANG and T. ZHANG. Simulation and expeimental studies of speed sensoless contol of pemanent magnet synchonous motos fo mini electic locomotive dive. Jounal of Contol and Automation. 4, vol. 7, iss., pp ISSN DOI:.457/ijca [] KRISHNAN, R. Electic Moto Dives: Modeling, Analysis, and Contol. st. ed., Uppe Saddle Rive: Pentice-Hall,. ISBN [] PILLAY, P. and R. KRISHNAN. Contol chaacteistics and speed contolle design fo a high pefomance of pemanent magnet synchonous moto dive. IEEE Tansactions on Powe Electonics. 99, vol. 5, iss., pp ISSN DOI:.9/ [] MISHRA, A., J. A. MAKWANA, P. AGARWAL and S. P. SRIVASTAVA. MRAS based estimation of speed in sensoless PMSM dive. In: IEEE Fifth Powe India Confeence. Muthal: IEEE,, pp. 5. ISBN DOI:.9/PoweI [3] PILLAY, P. and R. KRISHNAN. Modeling of pemanent magnet moto dives. IEEE Tansactions on Industial Electonics. 988, vol. 35, iss. 4, pp ISSN DOI:.9/ [4] WOOD, P. Theoy of Switching Powe Convete. st. ed. New Yok: Van Nostand-Reinhold, 98. [5] ISHAK, A. D., S. IQBAL and M. KAMAROL. A speed sensoless field oiented contol of paallel- connected dual PMSM. In: IEEE Intenational Confeence on Contol System, Computing and Engineeing (ICCSCE). Penang: IEEE,, pp ISBN DOI:.9/ICCSCE [6] SUL, S. K. Contol of Electic Machine Dive Systems. st. ed. Wiley-IEEE Pess,. ISBN [7] HOLTZ, J. Acquisition of position eo and magnet polaity fo sensoless contol of PM synchonous machines. IEEE Tansactions on Industy Applications. 8, vol. 44, iss. 4, pp ISSN DOI:.9/TIA [8] ZHAO, Y., W. QIAO and L. WU. Compensation algoithms fo sliding mode obseves in sensoless contol of IPMSMs. In: IEEE Intenational Electic Vehicle Confeence (IEVC). Geenville: IEEE,, pp. 7. ISBN DOI:.9/IEVC [9] BRANDSTETTER, P. and T. KRECEK. Estimation of PMSM magnetic saliency using injection technique. Jounal Electonics and Electical Engineeing. 4, vol., iss., pp. 7. ISSN DOI:.5755/j.eee c 6 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 38

11 VOLUME: 4 NUMBER: 6 MARCH About Authos Jyoti AGRAWAL eceived the M.Tech degee and is cuently pusuing the Ph.D. degee both in electical engineeing fom GHRCE, Nagpu, India. He eseach inteests include electic moto dives and powe electonics. Sanjay BODKHE eceived the Ph.D. degee in electical engineeing fom VNIT, Nagpu, in. His eseach inteests ae electic moto dives, powe electonics and smat gid. c 6 ADVANCES IN ELECTRICAL AND ELECTRONIC ENGINEERING 39