FPGA IMPLEMENTATION TO MINIMIZE TORQUE RIPPLES IN PERMANENT MAGNET SYNCHRONOUS MOTOR DRIVEN BY FIELD ORIENTED CONTROL USING FUZZY LOGIC CONTROLLER 1 B.ADHAVAN, 2 Dr.C.S.RAVICHANDRAN 1,2 Department of EEE (PG), Sr Ramakrshna Engneerng College, Combatore, Tamlnadu, Inda: 641022. E-mal: 1 adhav14@yahoo.com, 2 enyanrav@gmal.com ABSTRACT The Permanent Magnet Synchronous motor s a rotatng electrcal machne where the stator produces a snusodal flux densty dstrbuton n the ar gap and the rotor has permanent magnets. A substantal ar gap magnetc flux generated by permanent magnet makes t relable to desgn hghly effcent motors. However, the man dsadvantage s the non-unform varance n the developed torque. These torque rpples causes speed oscllatons and vbratons and perverts the system performance. Snce the constructon of permanent magnet synchronous motor lacks rotor col whch wll provde mechancal dampng durng transent condtons, these motors are neffcent wth open-loop V/Hz control and rely on Vector Control for better dynamc response. Fuzzy logc controllers whch does not requres any modelng of a system based on mathematcs and are they are workng on the lngustc rules. It also mproves the performance of PI controllers whch are affected by load turbulence, parameter varatons and speed dsturbances. Ths project presents the Fuzzy logc approach for a vector controlled PMSM drve for mnmzng torque rpples. Also, Space vector modulaton s employed to overcome the perodc torque pulsatons generated by hysteress controllers. The desgn analyss and the hardware control are mplemented by usng FPGA controller and dscussons on the hardware results are done. Keywords: Permanent Magnet Synchronous Motor (PMSM), Feld Orented Control (FOC), PI controller, Fuzzy Logc Controller (FLC), Feld Programmng Gate Array (FPGA), Space Vector Pulse Wdth Modulaton (SVPWM). 1. INTRODUCTION Permanent magnet synchronous motor (PMSM) s a hybrd of an AC nducton motor and a Brushless DC motor. They have rotor whch contan permanent magnets and t s smlar to that of a BLDC motors. These PMSMs are extensvely utlzed n small and medum power applcatons such as Robotcs, Computer perpheral equpment s, Adjustable speed drves and Electrc vehcles due to the advantages lke hgh effcency, hgh power factor, hgh power densty, compactness and mantenance free operaton. Also these motors are preferred over the tradtonal brush-type dc motors because of the absence of mechancal commutators, whch reduces mechancal wear and tear of the brushes and ncreases the lfe span of the motor. However, the man dsadvantage of PMSMs s the parastc torque pulsatons. Presence of these torque pulsatons results n nstantaneous torque that pulsates perodcally wth rotor poston. These pulsatons are reflected as perodc oscllatons n the motor speed, especally for low-speed operaton.there are varous sources of torque pulsatons n a PMSM such as the coggng, flux harmoncs, errors n current measurements, and phase unbalancng. In vew of the ncreasng popularty of PMSMs n ndustral applcatons, the suppresson of pulsatng torques has receved much attenton n recent years. Many technques based on both motor desgns and control technques have been proposed n lterature to dmnsh the torque rpples n the PMSM [10]. Nonlnear torque controller centered on flux/torque estmate s ntroduced to reduce the nfluence of the flux harmoncs. The nfluence of the coggng torque s sgnfcantly condensed at low motor speed by means of nternal model prncple and adaptve feed forward compensaton technque. The dsadvantage 369
s that the practcal mplementaton of the model requres addtonal work [9]. The torque rpples are parastc and can lead to torque pulsatons, vbratons and nose. The complex state varable technque s used for modelng and analyzng the effects of parastc torque. The compensaton of parastc torque rpples s by producng hgh frequency electromagnetc torque components through current control system [8]. A numercal predetermnaton of the current waveform s mposed n phases of machne to attan a steady torque. Torque regulaton scheme s based on-lne nstant estmaton. The optmzaton of the current and mnmzaton of copper losses s done by current modulus mnmzaton and the results are compared. The optmzaton of current functon optmzes the mean torque, such that torque rpple s mnmzed [5]. A new nverter output flter topology s desgned to reduce the hgh frequency harmoncs whch reduces the breakdown strength of nsulaton and reduces nsulaton lfe and causes severe damage to the motor bearngs. These output flters reduce the phase shft n the current regulatng system and torque rpple respectvely and mproves the flter effcency [17]. An embedded phase doman model of PMSM, whch reduces nstabltes due to numercal. Both the conventonal and embedded phase doman models of PMSM shows ndstngushable results n the steady state and transent condtons, however the conventonal model of PMSM becomes unstable f the tme-step s ncreased [6]. Torque rpple dmnshng n a permanent magnet synchronous motor wth non deal back electromotve force. In order to acheve constant torque wth reduced torque rpples mproved current trackng n the presence of perodc reference sgnals and dsturbances s proposed by the applcaton of current repettve technques n a feld-orented PMSM drve [13]. To dmnsh the torque rpples and harmonc noses n PMSM an actve flter topology. The hysteress voltage control method s used n actve flter, whle the motor current uses the hysteress current control method. The smulaton results show that the total harmonc dstorton goes below 13% wth EMI nose dampng down to nearly to -10 db [7]. An applcaton of computatonal ntellgence technques lke fuzzy logc s used to reduce the torque rpples assocated wth drect torque control n PMSM [12]. 2. FIELD ORIENTED CONTROL The prmary lmtaton of snusodal commutaton s controllg motor currents whch are tme varant n nature. Ths breaks down as speeds and frequences ncreases due to the restrcted bandwdth of PI controllers. Ths problem can be solved by controllng the current space vector drectly n the d-q reference frame of the rotor whch s known as Vector Control or Feld Orented control or Decouplng control. Feld-orented control s an effcent method to control a PMSM n adjustable speed drve applcatons wth quckly changng load n a wde range of speeds ncludng hgh speeds where feld weakenng s requred. It demonstrates a synchronous motor to be controlled lke a separately excted dc motor by the orentaton of the stator mmf or current vector n relaton to the rotor flux. Feld orented control conssts of vectors to control the stator currents. The feld orented control s based on transformng a three phase coordnates systems whch are tme and speed dependent nto d and q co-ordnates (two co-ordnate system) tme nvarant system. The DC machne type control s acheved through these projectons of three phases to two phase conversons. Feld orented controlled machnes requre two constant nput references; they are the torque component whch s algned wth the q coordnate and the flux component whch s algned wth d coordnate. Ths makes the feld orented control system precse n all workng modes (transent and steady state) and unlmted bandwdth compared to lmted bandwdth n mathematcal model. The Feld Orented Control solves the followng steps: 1. Fndng the constant reference of flux component and torque component of the stator current. 2.Calculaton of torque from (d,q) reference frame by applyng drect torque control and the expresson of torque s gven by T α Ψ R sq (1) The lnear relatonshp of torque and torque component ( sq ) can be mantaned by controllng the ampltude of rotor flux (Ψ R ) at a fxed value. By controllng the torque components of stator current vectors the torque can be controlled. The three phase currents, fluxes and voltages of PMSM can be analyzed n stpulatons of composte space 370
vectors. Assumng that a, b, c are the nstantaneous stator phase (a, b, c) currents respectvely. a + b + c =0 (2) These stator currents space vectors depct the three phase snusodal system. It stll wants to be altered nto a tme nvarant co-ordnate system. Ths converson can be dvded nto two steps: 1. Clarke transformaton ((a, b, c) to (α, β)) = a 1 2 sβ = a + 3 3 b (3) The nverse Clarke transformaton transforms from a 2-phase (α,β) to a 3-phase ( sa, sb, sc ) system. sa = 1 3 sb =- + 2 2 sβ 1 3 sc = - - 2 2 sβ (4) 2. Park transformaton ((α,β) to (d,q)) sd = cosθ + sβ snθ = - snθ + cosθ sq sβ where θ s the rotor flux poston. (5) These components depend on the current vector (α,β) components and on the rotor flux poston. Inverse Park transformaton modfes the voltages n d,q rotatng reference frame n a two phase orthogonal system. V = V cosθ - V snθ ref sdref sqref V = V snθ + V cosθ sβref sdref sqref (6) The proposed block dagram of PMSM drven by FOC usng Fuzzy logc controller wth SVPWM s descrbed n the Fgure 1. 3 Ф AC supply Rectfer V dc Tref Tm Fuzzy logc controller sdref = 0 Sqref PI- q PI- d V sqref V sdref Park Inverse Transformato n V α V β SVPWM 3 Phase Inverter Poston (θ) Poston and Speed Sensng Sq Sd Park Transformato n α β Clark Transformaton a b c PMSM Sensor nformatons from motor LOAD Shaft Fgure 1 Proposed Block Dagram Of PMSM drven by FOC Usng FLC Wth SVPWM 371
3. SPACE VECTOR PULSE WIDTH MODULATION The space vector PWM method s an advanced, computaton-ntensve PWM method whch s the best among all the PWM technques for varable-frequency drve applcatons. There are eght possble combnatons for the swtch commands whch determne eght phase voltage confguratons. Ths PWM technque approxmates the reference voltage V ref by a combnaton of the eght swtchng patterns (V 0 to V 7 ) are descrbed n the Fgure 2 wth the Swtchng patterns and output voltages of a three-phase power nverter n Table 1. To mplement the space vector PWM, the voltage equatons n the abc reference frame can be transformed nto the statonary αβ reference frame that conssts of the horzontal (α) and vertcal (β) axes, as a result, sx non-zero vectors and two zero vectors are possble. Fgure 2 Swtchng Vectors And Sectors Table 1: Swtchng Patterns And Output Voltages Of A Three-Phase Power Inverter Vector Va Vb Vc Vab Vbc Vca V 0 ={000} 0 0 0 0 0 0 V 1 ={100} 2/3 1/3 1/3 1 0-1 V 2 ={110} 1/3 1/3 2/3 0 1-1 V 3 ={010} 1/3 2/3 1/3-1 1 0 V 4 ={011} 2/3 1/3 1/3-1 0 1 V 5 ={001} 1/3 1/3 2/3 0-1 1 V 6 ={101} 1/3 2/3 1/3 1-1 0 V 7 ={111} 0 0 0 0 0 0 4. FUZZY LOGIC CONTROLLER (FLC) [1] Fuzzy logc (FL) s defned as mult-valued logc whch deals wth problems that have fuzzness or vagueness. Fuzzy logc s a methodology used for problem solvng n any type of control system and t can be mplemented n mcroprocessors, embedded mcrocontrollers, FPGA chps and PC based data acquston and control systems. FL provdes a trouble-free way to get at a defnte concluson based upon unclear, vague, naccurate, nosy, or absent nput nformaton. Fuzzy Logc approach towards a control system problem mmcs how a person would make decsons. The nputs to the FLC are error of torque and change n torque error and the output s torque lmt are descrbed n the Fgure 3, 4 and 5. The nputs and outputs contans seven lngustcs membershp functons, they are as follows PB- Postve Bg, PM- Postve Medum, PS- Postve Small, Z- Zero, NS- Negatve Small, NM- negatve medum, NB- Negatve Bg, are descrbed n Table 2. 372
NB NM NS Z PS PM PB -2.2-1.32-0.88-0.44 0 0.44 0.88 1.32 2.2 Fgure 3 Inputs To FLC- Torque Error NB NM NS Z PS PM PB -0.125-0.075-0.05-0.025 0 0.025 0.05 0.075 0.125 Fgure 4 Inputs to FLC- Change In Torque Error NB NM NS Z PS PM PB processng the speed s very hgh. The FPGAs gves lmtless flexblty to the desgners due to reconfgurable opton. It dstrbutes memory throughout the devce, so the dedcated memory needed by each task s permanently allocated. Ths provdes a hgh degree of solaton between tasks. The desgner of an FPGA controller has complete flexblty to select any combnaton of perpherals and controllers. 6. RESULTS AND DISCUSSION The smulaton parameters of PMSM used n the smulnk matlab and hardware are shown n table 3 wth the Hardware block dagram and Hardware setup n Fgure 6 and Fgure 7. The Fgure 8 and Fgure 9 shows Inverter Lne to lne Voltage waveform and current waveform at 75% loadng. The Fgure 11 and Fgure 12 shows Inverter Lne to lne Voltage waveform and current waveform at 100% loadng. The Fgure 10 and Fgure 13 shows the torque waveform for 75% and 100% loadng. R Y B PC 5 V Power Supply 3 Ф Rectfer -2.2-1.32-0.88-0.44 0 0.44 0.88 1.32 2.2 Fgure 5 Torque Lmt Output Of FLC RS-232 Seral Port Communcaton V dc e e Table 2: Rules for FLC B NM NS Z PS PM PB Gate pulse Intellgent Power Module PEC16DSMO1 (3 phase nverter) NB NB NB NB NM NM NS Z NM NB NB NB NM NS Z PS NS NB NM NS NS Z PS PM Z NM NM NS Z PS PM PM PS NM NS Z PS PS PM PB PM NS Z PS PM PM PB PB PB Z PS PM PM PB PB PB A/D Converter I a I b I c T m T ref Current sensor PMSM 5. FIELD PROGRAMMABLE GATE ARRAY FPGA CONTROLLER SPARTAN 3A DSP Feld Programmable Gate Array (FPGA) s low cost, hgh-performance DSP soluton for hghvolume, cost-conscous applcatons. Ther effcency n concurrent applcatons s acheved by usng multple parallel processng blocks. Due to parallel processng paths, dfferent tasks do not want to compete for the same resources. In a sngle FPGA devce multple control loops can be executed at dfferent rates. Because of parallel LOAD SHAFT Fgure 6 Hardware Block Dagram of Feld Orented Control Of PMSM Usng FPGA Controller 373
Table 3: Parameters Of PMSM Rated Power Rated Torque Rated Voltage 1.1HP 2.2Nm 220 V Rated Current 3.69A Rated Speed 4600 rpm Torque Constant 0.6 Nm/Arms Termnal To Termnal Resstance 3.07Ω Termnal To Termnal Inductance 6.57 mh Moment of Inerta 1.8 kgcm 2 6.1 Waveforms Of Inverter Lne To Lne Voltage, Current And Torque (75% Loadng) Fgure 8 Inverter Lne To Lne Voltage Waveform For 75% Loadng (X-Axs: Tme In Ms: 1 Dv= 10 Ms, Y- Axs (Vab, Vbc, Vca): Voltage In Volts: 1 Dv= 325 V) Fgure 7 Snapshot Of Hardware Setup Fgure 9 Current Waveform For 75% Loadng (X-Axs: Tme In Ms: 1 Dv= 10 Ms, Y-Axs (Iab, Ibc, Ica): Current In Amps: 1 Dv= 2 A) 374
Fgure 10 Torque Waveform For 75% Loadng(X-Axs: Tme In S, 1 Dv= 10s, Y-Axs: Torque In Nm, 1 Dv= 2.2Nm) T Torque Rpple Factor(TRF%)= p e a k -p e a k T a v e 1 0 0 (7) From the Fgure 10 the motor torque(tm) follows the reference motor torque (Tref) wth less torque rpples and t s shown n the crcled porton of waveform whch s zoomed and shown nsde the fgure 10. The torque rpple factor for the proposed scheme as per equaton (7) s gven below for fgure 10. Fgure 12 Current Waveform For 100% Loadng(X-Axs: Tme In Ms: 1 Dv= 10 Ms, Y-Axs: Current In Amps(Iab, Ibc, Ica): 1 Dv= 3 A) From the Fgure 13 the motor torque(tm) follows the reference motor torque (Tref) wth less torque rpples and t s shown n the crcled porton of waveform whch s zoomed and shown nsde the fgure 13. 1.65 1.62 Torque Rpple Factor (TRF %) = 1.65 =1.81% 6.2 Waveforms Of Inverter Lne To Lne Voltage, Current And Torque (100% Loadng) Fgure 13 Torque Waveform For 100% Loadng(X- Axs: Tme In S, 1 Dv= 10s, Y-Axs: Torque In Nm, 1 Dv= 1.46 Nm) The torque rpple factor for the proposed scheme as per equaton (7) s gven below for fgure 13. Fgure 11 Inverter Lne To Lne Voltage Waveform For 100% Loadng(X-Axs: Tme In 1 Dv= 10 Ms, Y-Axs: Voltage In Volts: 1 Dv= 325 V) 2.24 2.2 Torque Rpple Factor (TRF %) = 2.2 =1.81% 375
Table 4: Comparson Of Control Strateges In PMSM CONTROL STRATEGIES TORQUE RIPPLE (%) Proposed FLC wth SVPWM 1.81 P. Mattavell, L.Tubana, and M. 3.8 Zglotto,2005 [13] W. Qan and K. Panda, 2004 [14] 3.9 M.Tarnk and J.Murgas, 2011 [20] 4 H. Hasanen, 2010 [19] 12 It s clear that varaton n Torque shown n table 4 s less n case of Fuzzy logc controllers and they can acheve a mnmum torque rpple than other control technques. It has been vewed that the proposed control strategy has helped n reducng the torque rpples to 1.81%. Thus by usng FLC based controller, rpples are reduced completely. 7. CONCLUSION Fuzzy logc controller based Torque controller model of PMSM motor drve have been modeled and mplemented usng FPGA and the results have been presented to demonstrate the proposed FLC based control. The mplemented hardware result of Fuzzy Logc Controller has shown that t s better over the PI Torque controller n reducng the torque rpples to 1.81%. And therefore t can be successfully used n poston of PI Torque controller. In future mplementaton, Hybrd Neuro-Fuzzy controllers can be used to replace the PI controller. ACKNOWLEDGEMENTS We would lke to thank the Management, Prncpal and Head of the EEE-PG Department of Sr Ramakrshna Engneerng College for provdng facltes and valuable support for carryng out ths work. REFERENCES [1] B.Adhavan, M.S.Brundha, C.S.Ravchandran, and V.Jagannathan, Torque rpple Reducton n Permanent Magnet Synchronous motor usng Fuzzy logc control, Australan Journal of Basc and Appled Scences, Vol.7,No.7 May 2013, pp. 61-68. [2] B.Adhavan, A.Kuppuswamy,G.Jayabaskaran, and V.Jagannathan, Feld orented control of Permanent Magnet Synchronous Motor (PMSM) usng Fuzzy logc controller,proceedngs of Internatonal Conference on Recent Advances n Intellgent Computatonal Systems, Trvandrum, Kerala(Inda), September 22-24, 2011, pp.587-592. [3] J.C.Baslo and S.R.Matos, Desgn of PI and PID controllers wth transent performance specfcaton, IEEE Transactons on Educaton, Vol. 45, No. 4, August 2002, pp.364-370. [4] Bmal.K.Bose, Modern Power Electroncs and AC Drves, PHI Learnng Prvate Lmted, Inc., 2005. [5]F.Colamartno,C.Marchand, and A.Razek, Torque rpple mnmzaton n permanent magnet synchronous servo drve, IEEE Transactons on Energy Converson, Vol. 14, No. 3, September 1999, pp. 616-621. [6]A.B.Dehkord,A.M.Gole, and T.L.Magure, Permanent Magnet Synchronous Machne Model for Real- Tme Smulaton, Proceedngs of Internatonal Conference on Power System Transents (IPST 05), Montreal, Canada, June 19-23, 2005,pp.1-6. [7] K.Gulez, A.A.Adam, and H.Pastac, Torque rpple and EMI nose mnmzaton n PMSM usng actve flter topology and feld orented Control, IEEE Transactons on Industral Electroncs, Vol. 55, No. 1, January 2008, pp. 251-257. [8]F.Heydar, A.Shekholeslam, K.Gorgan Frouzjah, and G.Ardeshr, Predctve Feld Orented Control of PMSM Usng Fuzzy Logc, Proceedngs of 24 th Internatonal Conference on power system, Tehran, Iran, November 16-18. 2009, pp.1-10. [9]T.M.Jahns, and W.L.Soong, Pulsatng torque mnmzaton technques for permanent magnet AC motor drves-a revew, IEEE Transactons on Industral Electroncs, Vol. 43, No. 2, Aprl 1996, pp.321-330. [10] M.Kadjoudj, N.Golea, and M.E.H. Benbouzd, Fuzzy Rule Based Model Reference Adaptve Control for PMSM Drves, Serban Journal Electrcal Engneerng, Vol. 4, No. 1, June 2007, pp.13-22. 376
[11] R. Krshnan, Permanent Magnet Synchronous and Brushless DC Motor Drves, CRC Press, Inc., 2009. [12] N.L., X.We, and X.Feng, An Improved DTC Algorthm for Reducng Torque Rpples of PMSM Based on Fuzzy Logc and SVM, Proceedngs of Internatonal Conference Artfcal Intellgence and Educaton (ICAIE), Hangzhou,Chna, October 29-30, 2010, pp. 401-405. [13] P.Mattavell, L.Tubana, and M.Zglotto, Torque-rpple reducton n PM synchronous motor drves usng repettve current control, IEEE Transactons Power Electroncs, Vol. 20, No. 6, November 2005, pp. 1423 1431. [14] W.Qan, S.K. Panda and J.X.Xu, Torque rpple mnmzaton n PM synchronous motors usng Iteratve learnng control, IEEE Transactons on Power Electroncs, Vol. 19, No. 2, March 2004, pp 272-279. [15] E.S.Sergak, P.S.Georglaks, A.G.Kladas, and G.S.Stavrakaks, Fuzzy Logc Based Onlne Electromagnetc Loss Mnmzaton of Permanent Magnet Synchronous Motor Drves, Proceedngs of 18 th Internatonal Conference Electrcal Machnes,Vlamoura(Algarve), Portugal,September 6-9, 2008, pp.1-7. [16] H.F.E. Solman and M.E.Elbuluk, Improvng the Torque Rpple n DTC of PMSM usng Fuzzy Logc, Proceedngs of Industral Applcatons socety annual meetng (IAS), Edmonton, AB, October 5-9, 2008,pp.1-8. [17] Y. Sozer, D.A.Torrey, and S. Reva, New nverter output flter topology for PWM motor drves, IEEE Transactons on Power Electroncs, Vol. 15, No. 6, November 2000, pp. 1007 1017. [18] M.N.Uddn, An Adaptve-Flter-Based Torque-Rpple Mnmzaton of a Fuzzy- Logc Controller for Speed Control of IPM Motor Drves, IEEE Transactons on Industral Applcatons, Vol. 47, No. 1, February 2011, pp.350-358. [19] H. Hasanen, Torque rpple mnmzaton of permanent magnet synchronous motor usng dgtal observer controller, Energy Converson Management (Elsever), Vol. 51, No. 1, January 2010, pp. 98-104. [20] M.Tarnk and J.Murgas, Addtonal Adaptve Controller for Mutual Torque Rpple Mnmzaton n PMSM Drve Systems, Proceedngs of 18 th IFAC World Congress, Mlano, Italy,Vol. 18, No. 1, January 2011, pp. 4119-4124. [21] M. Saafan, Y. Hakal, F. Saraya, Fayez F.G. Areed, Adaptve Control for Torque Rpple Mnmzaton n PM Synchronous Motors, Internatonal Journal Computer Applcatons, Vol. 37, No. 5, January 2012, pp. 1-8. 377