FUZZY LOGIC BASED HPWM-MRAS SPEED OBSERVER FOR SENSORLESS CONTROL OF INDUCTION MOTOR DRIVE

Transcription

1 VOL., NO. 4, APRIL 21 ISSN ARPN Jounal of Engineeing an Applie Science Aian Reeach Publihing Netwok (ARPN). All ight eeve. FUZZY LOGIC BASED HPWM-MRAS SPEED OBSERVER FOR SENSORLESS CONROL OF INDUCION MOOR DRIVE M. Haeeb Khan 1 an J. Amanath 2 1 Royal Intitute of echnology an Science, Chevella, RR Dit. Anha Paeh, Inia 2 JNU College of Engineeing, JN Univeity, Hyeaba, Anha Paeh, Inia haeeb_m@yahoo.com, amanathjinak@yahoo.com ABSRAC he main aim of thi pape i to peent hybi pace vecto pule with moulation (HPWM) bae moel efeence aaptive ytem (MRAS) pee obeve fo enole contol of inuction moto ive. he popoe metho i eigne by uing the notion of tato flux ipple an the witching time ae calculate by utilizing the concept of imaginay witching time that ue the intantaneou value of the efeence voltage of the thee phae. By uing thi metho ipple peent in the toue can be minimize an alo the pee pefomance can be impove in the way that the eo between the aaptive moel pee an the efeence moel pee i mae a low a poible o that the etimate pee i mae a nealy eual a the pee of the efeence moel. Alo in thi pape the conventionally ue PI contolle in the aaptation mechanim of the MRAS obeve i eplace by a fuzzy logic contolle (FLC). In conventionally ue PI contolle when the efeence pee o the loa change the contol paamete ae to be change manually but when a FLC i ue in place of PI contolle thi i not neceaily euie. he fuzzy logic i alo helpful in keeping the pee elatively contant uing the loa vaiation which can be een fom the imulation eult. he imulation of the popoe metho ae pefome an the eult ae peente to valiate the popoe metho. Keywo: inuction moto, enole contol, MRAS, HPWM, fuzzy logic contol. Nomenclatue v = axi tato voltage (V) v = axi tato voltage (V) v = axi oto voltage (V) v = axi oto voltage (V) i = axi tato cuent (A) i = axi tato cuent (A) i = axi oto cuent (A) i = axi oto cuent (A) R = Stato eitance (Ω) R = Roto eitance (Ω) L = Stato inuctance (H) L = Roto inuctance (H) L = Mutual inuctance (H) m ω = Roto Spee (a/ec) INRODUCION In liteatue thee ae numeou enole contol techniue icue an thee ae ummaize in [1]. Vaiou techniue ae uggete uch a Luenbege an Kalman filte obeve, Moel Refeence Aaptive Sytem (MRAS), Atificial Intelligence (AI) echniue an o on. Out of the iffeent oto etimation techniue available in liteatue the MRAS techniue i foun to be mot common metho an fin lot of attention ue to it taight fowa tability appoach, low computational effot an implicity [2]. MRAS obeve bae on oto flux, back emf, eactive powe i ome of the MRAS obeve that have been intouce in liteatue [3-6]. Schaue wa the fit to popoe oto flux MRAS [], which i the mot populaly ue MRAS tategy. A lot of effot by the eeache ha been focue on thi tategy to futhe impove it pefomance. Space vecto pule with moulation (SVPWM) techniue i being fo a long time now. he main avantage of thi techniue i that the witching loe ae low; the hamonic pefomance i impove an pouce an impove output voltage fo the ame c bu voltage [7-1], when compae to a inuoial PWM (SPWM). SPWM i a techniue in which PWM output ae pouce bae on the pinciple that compae a tiangula caie ignal with a inuoial efeence wave [11]. SVPWM i an avance PWM metho an one of the bet metho among all the iffeent type of PWM metho available in liteatue fo vaiable pee ive an hence become the motly ue PWM techniue [12-]. hi cheme of SVPWM may look imple theoetically but the implementation of thi cheme become little ifficult with the ue of low ampling peio o high witching feuency [16]. In [17-2] iffeent type of hybi PWM algoithm have been evelope fo euce cuent ipple. In the popoe metho a HPWM MRAS pee obeve i evelope bae on notion of tato flux ipple an witching time ae calculate uing the logic of imaginay witching time which ue the intantaneou value of the efeence voltage of the thee phae. Alo thee ae ignificant uncetaintie in the inutial application o that the pefomance of the ive may 86

2 VOL., NO. 4, APRIL 21 ISSN ARPN Jounal of Engineeing an Applie Science Aian Reeach Publihing Netwok (ARPN). All ight eeve. eteioate if PI contolle ae ue. Due to thi it i woth uing contolle which have the capability of hanling uncetaintie caue by paamete vaiation. A fuzzy logic contolle offe a goo pefomance uing paamete vaiation an it can be ue in place of conventional PI contolle in the aaptation mechanim of MRAS obeve [21]. In the popoe metho the pefomance of the ive i alo enhance by eplacing PI contolle with fuzzy logic contolle. PRINCIPLE OF CONVENIONAL MRAS he conventional MRAS moel evelope by Schaue i hown in Figue-1. he main logic behin MRAS cheme i that thee i a efeence moel an an aaptive moel whee in the efeence moel i ue to etemine the euie tate an the aaptive moel alo known a ajutable moel povie the etimate value of the tate. he eo obtaine between the efeence an aaptive moel i given to an aaptation mechanim which ajut the aaptive moel by geneating the etimate value of oto pee. he poce of ajuting the aaptive moel i mae to continue till the eo obtaine between the two moel ten to zeo. Figue-1. Conventional oto flux MRAS pee obeve. he efeence oto flux component that ae obtaine fom the efeence moel ae given a [, 6, 22] ( Ri + L pi v m L pψ = σ ) (1) L ( Ri + L pi v m L pψ = σ ) (2) L he oto flux component that ae obtaine fom the aaptive moel ae given a [, 6, 22] pψˆ pψˆ ( Lmi ψˆ 1 = ˆ ωψˆ + ) (3) ( L i ψˆ ) 1 = ˆ ωψˆ + m (4) Whee σ i leakage coefficient an i given a L 2 m σ = 1 () L L Fom the above (1) an (2) the efeence moel i evelope an the aaptive moel i evelope by uing (3) an (4). Afte eveloping aaptive an efeence moel the aaptation mechanim i to be eigne which fom the vey impotant pat of the MRAS cheme. he aaptation mechanim i eigne in a way to geneate the value of etimate pee ue o a to minimize the eo between the etimate an efeence fluxe. In the commonly ue aaptive mechanim thi i one by moe geneally ue PI contolle. he pee tuning ignal i minimize by PI contolle which geneate the etimate pee an i fe back to the aaptive moel a hown in Figue-1. he pee tuning ignal i given a ex ψ ψˆ ψ ψˆ e x = (6) an the etimate pee i given a k p i ωˆ = k p + ex (7) 1 he electomagnetic toue euation of inuction moto in tationay fame i given a e 3 p Lm = ( i ψ i ψ ) (8) 2 2 L HYBRID PWM MEHODOLOGY he voltage vecto, pouce by a 3-phae, twolevel invete, ivie the pace vecto plane into ix ecto a hown in Figue-2. A the ecto ae ymmetic, the icuion in thi pape i limite to fit ecto only. In the pace vecto appoach, the eie e x 87

3 VOL., NO. 4, APRIL 21 ISSN ARPN Jounal of Engineeing an Applie Science Aian Reeach Publihing Netwok (ARPN). All ight eeve. efeence vecto i geneate by time aveaging the uitable icete voltage vecto in ampling peio. Fo a given efeence voltage V ef an angle in ecto I, the volt-time balance i maintaine by applying the active vecto 1, 2 an zeo tate togethe fo uation 1, 2 an Z epectively, a given in (9) to (11) Figue-2. Switching tate an coeponing voltage vecto. o ( in( α ) M i = 6 (9) π = 2 3 ( in α ) M 2 i π (1) z = ( ) (11) In the popoe algoithm, the witching time can be calculate by uing the concept of imaginay witching time which ue intantaneou value of the efeence voltage of a, b an c phae. hi metho oe not epen on the magnitue of the efeence voltage pace vecto an it elative angle with epect to the efeence axi. he imaginay witching time peio popotional to the intantaneou value of the efeence phae voltage ae efine a an V = V an c bn V = V bn c V = V cn cn (12) c Fuzzy logic contolle Fuzzy logic i epecially avantageou fo poblem that cannot be eaily epeente by mathematical moeling becaue ata i eithe unavailable, incomplete o the poce i too complex. Such ytem can be eaily upgae by aing new ule to impove pefomance o a new featue. In many cae, fuzzy contol can be ue to impove exiting taitional contolle ytem by aing an exta laye of intelligence to the cuent contol metho. he iffeent type of fuzzy contolle that ae icue in liteatue ae baically multi input ingle output (MISO) type contolle [21, 23-]. he main pefeence of the fuzzy logic i that i eay to implement contol an it ha the ability of genealization. In the conventional metho of MRAS pee obeve PI contolle wa commonly ue in the aaptation mechanim which i geneating etimate pee which in tun i eucing the pee tuning ignal e x o eo between the aaptive an efeence moel. In the popoe metho the PI contolle that i ue in the aaptation mechanim i eplace by a fuzzy logic contolle. he geneal block iagam of fuzzy contolle i epeente a hown in Figue-3. Figue-3. Block iagam of fuzzy logic contolle. Spee tuning ignal e x an it ate of change ae the input to the popoe FLC which ae multiplie by two caling facto K 1 an K 2 epectively. hen the output of the contolle i multiplie by a thi caling facto K 3 an etimate pee i obtaine. he tial an eo techniue i ue to fin the value of caling facto fo the bet pefomance [21, 26]. he value hence obtaine fo K 1, K 2 an K 3 ae., 2 an 3 epectively. he fit tep in eigning the fuzzy contolle i to geneate the fuzzy ule bae on the knowlege of the expet. Accoing to the expet, thee ituation can be itinguihe fo the moto pee, namely, above, aoun an below the eie efeence pee. By efining the ytem eo between the meaue pee an the eie pee, the popoition, highe, aoun an beneath the eie efeence pee ae othewie expee a Poitive, Zeo an Negative eo. Fo 2 input an N numbe of linguitic vaiable the numbe of ule ae given a N 2. In thi popoe metho linguitic value of i choen which give 88

4 VOL., NO. 4, APRIL 21 ISSN ARPN Jounal of Engineeing an Applie Science Aian Reeach Publihing Netwok (ARPN). All ight eeve. ule. Fo fuzzyfication tiangula fuzzyfication i ue an fo effuzzyfication the centoi effuzzyfication metho i ue in the popoe metho. he fuzzy et ue in the popoe metho ae NB: Negative Big, NS: Negative Small, ZE: Zeo Eual, PS: Poitive Small, PB: Poitive Big. he look-up table fo the popoe metho i hown in able-1 an FLC implementation uing thi i hown in Figue-3. able-1. Fuzzy ule fo popoe ytem. e x e x PB PS ZE NS NB PB PB PB PB PS ZE PS PB PB PB ZE NS ZE PB PS ZE NS NB NS PS ZE NS NB NB NB ZE NS NB NB NB Fuzzy logic bae MRAS pee obeve i hown in Figue-4. Conventionally PI contolle wee ue in aaptation mechanim which i eplace by a fuzzy logic contolle o that the pefomance of the ive can be enhance an alo uing loa vaiation the pee can be kept contant. Figue-4. FLC bae MRAS pee obeve. PROPOSED FUZZY LOGIC BASED HPWM MRAS SPEED OBSERVER he popoe fuzzy logic bae HPWM MRAS pee obeve fo enole contol of inuction moto ive i hown in Figue. Figue. Block iagam of popoe FLC bae HPWM MRAS pee obeve. In the popoe metho, the poition of the * efeence tato flux vecto ( ψ ) i eive by the aing lip pee an actual oto pee. Afte each an evey ampling uation actual tato flux vecto ( ψ ) i ajute by the eo an it tie to attain the efeence flux pace vecto. hu the flux eo i minimize in each ampling inteval. Refeence value of the an axe tato fluxe an actual value of the an axe tato fluxe ae compae in the V ef calculato block. he eo in the an axe tato flux vecto ae obtaine a ψ = ψ * ψ ψ = ψ * ψ, (13) We can etemine the appopiate V ef pace vecto by uing flux eo an tato ohmic op an it i given a * ψ * ψ V = Ri +, V = Ri + (14) Whee, i ampling peio alo known a the uation of ub-cycle SIMULAION RESULS AND DISCUSSIONS o valiate the popoe metho imulation of conventional MRAS cheme an popoe cheme ae pefome fo 3- phae 2 pole inuction moto with paamete a R = 3. Ω, R = 4. Ω, L = L =.76 mh, L m =.646 mh an moment of inetia of.33 kg/m 2. Figue-6 an 7 how the pee epone an toue evelope fo the conventional MRAS ive fo a pee of 1 pm an in Figue-8 an 9 the pee epone an toue evelope ae hown fo the popoe ive. A can be een fom imulation eult the ipple peent in the toue ae euce an the teay tate of the ive i eache fate than the conventional ive. Figue-1, 11 how the pee epone an toue evelope at 13 pm with loa applie at 1 ec fo conventional MRAS. Figue-12, 13 how the epone of the popoe ive with a loa of 1 N-m applie at 1 ec. In Figue-14, an Figue-16, 17 the pee epone an toue evelope when the fo both conventional an popoe ive when a loa of 1 N-m applie at.8 ec an emove at 1. ec at 13 pm. Fom the wavefom it can be obeve that when the loa i applie the pee i emaining moe table fo the popoe ive when compae to the conventional MRAS ive an the ipple peent in the toue ae alo euce. 89

5 VOL., NO. 4, APRIL 21 ISSN ARPN Jounal of Engineeing an Applie Science Aian Reeach Publihing Netwok (ARPN). All ight eeve Spee (pm) Etimate Refeence oue (N-m) Figue-6. Spee epone of conventional MRAS without loa Figue-9. oue evelope fo popoe ive. oue (N-m) 2 1 Spee (pm) Etimate Spee Refeence Spee Figue-7. oue evelope fo conventional MRAS Figue-1. Spee epone fo conventional MRAS, loa of 1 N-m applie at 1 ec. 1 8 Etimate Refeence 2 Spee (pm) 6 4 oue (N-m ) Figue-8. Spee epone of popoe ive without loa Figue-11. oue evelope fo the above cae. 9

6 VOL., NO. 4, APRIL 21 ISSN ARPN Jounal of Engineeing an Applie Science Aian Reeach Publihing Netwok (ARPN). All ight eeve Spee (pm) Etimate Spee Refeence Spee oue (N -m ) Figue-12. Spee epone fo popoe MRAS, loa of 1 N-m applie at 1 ec Figue-. oue evelope fo the above cae. oue (N-m ) 2 1 Spee (pm) Etimate Spee Refeence Spee Figue-13. oue evelope fo the above cae Figue-16. Spee epone of Popoe MRAS at 13 pm with tep change in loa, Loa of 1 N-m applie at.8 ec an emove at 1. ec Etimate Spee Refeence Spee 2 Spee (pm ) Figue-14. Spee epone of conventional MRAS at 13 pm with tep change in loa, Loa of 1 N-m applie at.8 ec an emove at 1. ec. oue (N-m ) Figue-17. oue evelope fo the above cae. 91

7 VOL., NO. 4, APRIL 21 ISSN ARPN Jounal of Engineeing an Applie Science Aian Reeach Publihing Netwok (ARPN). All ight eeve. CONCLUSIONS Uually the contol of inuction moto ive uing low pee ange i vey ifficult an the pee pefomance uing low pee egion eteioate. Alo in the conventional MRAS contol ipple ae peent in the toue an the pee get affecte with change in loa. In thi pape a novel fuzzy logic contolle bae HPWM- MRAS metho fo enole contol of inuction moto ive i popoe. he conventionally ue PI contolle in the aaptation mechanim of the MRAS obeve i eplace by a fuzzy logic contolle. A can be een fom the imulation eult the pefomance of the ive uing low pee egion i impove an alo when the loa i applie the pee pefomance i bette when the popoe ive i ue. he fuzzy logic contolle i mainly helpful when the loa i vaie the pee i le effecte by the loa vaiation. REFERENCES [1] J. W. Finch an D. Giaoui. 28. Contolle AC Electical Dive. IEEE anaction on Inutial Electonic. : [2] M. Rahe an A. F. Stonach. 24. A table back- EMF MRAS-bae enole low pee inuction moto ive inenitive to tato eitance vaiation. IEE Poceeing Electic Powe Application. 1: [3] S. amai, H. Sugimoto, M. Yano Speeenole vecto contol of inuction moto with moel efeence aaptive ytem. Conf. Reco of the 198 IEEE-IAS Annual Meeting. pp [4] F. Peng an. Fukao Robut pee ientification fo pee-enole vecto contol of inuction moto. IEEE anaction on Inuty Application. 3: [] C. Schaue Aaptive pee ientification fo vecto contol of inuction moto without otational tanuce. IEEE anaction on Inuty Application. 28: [6] P. Va Senole Vecto an Diect toue contol. New Yok: Oxfo Univeity Pe. [7] M. A. Jabba, A.M. Khambakone an Y. Zhang. 24. Space-vecto moulation in a two-phae inuction moto ive fo contant-powe opeation. IEEE an. In. Electon. 1(): [8] V. S. Kuma an P. S. Kannan. 24. Hamonic tuie in pace vecto PWM invete ive ytem. In: Poc. Int. Conf. Powe Syt. echnol. 1: [9] M. zynalowki, K. Boiov, L. Yuan an Q. Ling.. A novel anom PWM techniue with low computational ovehea an contant ampling feuency fo high-volume, low-cot application. IEEE an. Powe Electon. 2(1): [1] P. McGath, D. G. Holme an. Meyna. 26. Reuce pwm hamonic itotion fo multilevel invete opeating ove a wie moulation ange. IEEE an. Powe Electon. 21(4): [11] H. Bai, Z. Zhao, S. Meng, J. Liu an X. Sun. 23. Compaion of thee PWM tategie-spwm, SVPWM an one-cycle contol. In: Poc. th Int. Conf. Powe Electon. Dive Syt. 2: [12] Ibal. 28. Analyi of pace vecto pule with moulation fo a five phae voltage ouce invete. IE (I) jounal-el. 89(3): 8-. [13] Ahmet M. Hava an Eme UN. 29. Pefomance Analyi of Reuce Common-Moe Voltage PWM Metho an Compaion with Stana PWM Metho fo hee-phae Voltage-Souce Invete. IEEE anaction on Powe Electonic. 24(1): [14] G. Naayanan, H. K. Kihnamuthy, Z. Di an R. Ayyana. 26. Avance bu-clamping PWM techniue bae on pace vecto appoach. IEEE an. Powe Electon. 21(4): [] R. Joon Hyoung, L. Kwang Won an L. Ja Sung. 26. A unifie flux an toue contol metho fo tc-bae inuction-moto ive. IEEE an. Powe Electon. 21(1): [16] N. Celanovic an D. Booyevich A fat pace vecto moulation algoithm fo multilevel theephae convete. In: Poc. In. Appl. Conf. 2: [17] Vlaimi Blako Analyi of a hybi PWM bae on moifie pace-vecto an tianglecompaion metho. IEEE an. In. Applicat. 33(3): [18] H. Kihnamuthy, G Naayanan, V.. Ranganathan, R. Ayya. 23. Deign of pace vecto-bae hybi PWM techniue fo euce cuent ipple. IEEE- APEC. 1: [19] G. Naayanan, Di Zhao, H. Kihnamuthy an Rajapanian Ayyana. 28. Space vecto bae hybi techniue fo euce cuent ipple. IEEE an. In. Applic. (4): [2] Uaya Senthil, B.G. Fenane. 23. Hybi pace vecto pule with moulation bae iect toue contolle inuction moto ive. IEEE PESC. pp

8 VOL., NO. 4, APRIL 21 ISSN ARPN Jounal of Engineeing an Applie Science Aian Reeach Publihing Netwok (ARPN). All ight eeve. [21] M. Haeeb Khan, Ahia Azam, J. Amanath. Fuzzy Logic Bae MRAS pee obeve fo contol of Inuction moto ive. Accepte fo publication in Wol Jounal of Moeling an Simulation. [22] Mohamma Haeeb Khan, J. Amanath. Contol of Inuction Moto Dive without Shaft Encoe Uing Moel Refeencing Aaptive Sytem an o Avoi oue Jek in anition at Stating. Intenational Jounal of Applie Engineeing Reeach. 4(6): [23] Zaeh L. A Fuzzy et. Infomation an Contol. 8: [24] P. Va Atificial-Intelligence-Bae Electical Machine an Dive-Application of Fuzzy. Neual, Fuzzy-Neual an Genetic Algoithm Bae echniue. New Yok: Oxfo Univeity Pe. [] Lim. an Bien Z FLC Deign fo Multiobjective Sytem. Intenational Jounal of Applie Mathematic an Compute Science. 6(3): 68. [26] Y. Milou an A. Daou. 22. Fuzzy logic bae oto eitance etimato of an iniect vecto contolle inuction moto ive. In: Poc. IEEE 28 th Annual Confeence of the Inutial Electonic Society IECON 2. 93