使用在线参数辨识的永磁同步电机无位置传感器控制

SensorlessControlofPermanent-MagnetSynchronousMotorsUsingOnlineParameterIdenti cationBasedonSystemIdenti cationTheory

ShinjiIchikawa,Member,IEEE,MutuwoTomita,Member,IEEE,ShinjiDoki,Member,IEEE,

andShigeruOkuma,Member,IEEE

Abstract—Anonlineparameteridenti cationmethodandsen-sorlesscontrolusingidenti edparameterswererealizedinsurfaceandinteriorpermanent-magnetsynchronousmotors(SPMSMsandIPMSMs,respectively).Asthismethoddoesnotuserotorpositionorvelocitytoidentifymotorparameters,theidenti edpa-rametersarenotaffectedbypositionestimationerrorundersen-sorlesscontrol.Theproposedmethodcanbeappliedtoallkindsofsynchronousmotors.Theeffectivenessoftheproposedmethodwasveri edbyexperimentsinbothSPMSMsandIPMSMs.IndexTerms—Extendedelectromotiveforce(EEMF),onlineparameteridenti cation,permanent-magnetsynchronousmotor(PMSM),sensorlesscontrol.

I.INTRODUCTION

YNCHRONOUSmotorsarewidelyusedinindustrialapplicationsbecauseoftheirhighef ciency.Therearemainlythreekindsofsynchronousmotors,namely:1)surfacepermanent-magnetsynchronousmotors(SPMSMs),2)interiorpermanent-magnetsynchronousmotors(IPMSMs),and3)syn-chronousreluctancemotors(SynRMs).PMSMshavebecomeespeciallywidespread,andmanystudiesonthePMSMshavebeenreported.Althoughrotorpositionandvelocitycanbeusedtoachieveprecisecontrolofthesemotors,positionsensorshaveseveralproblemssuchascostanddurability.Therefore,manysensorlesscontrolmethodshavebeenproposed[1],[2].

Thesesensorlesscontrolmethodscanbemainlypidedintotwotypes,i.e.,1)thoseusinghigh-frequencyvoltageorcurrentsignals[3]–[15]and2)thoseusingthefundamentalcomponentsofvoltageandcurrentsignals[3],[10],[16]–[30].Theformermethodsuserelationsamongthree-phasecurrents[3],injectionofhigh-frequencysignalsofvoltagesorcurrents[4]–[6],[10],[11],[13],[14],specialinverterpulsewidthmodu-lation(PWM)patterns[7],currentresponseofstepvoltages[8],[9],informationonharmonicreactivepower[12],andasystemidenti cationmethod[15]todetectrotorpositioninformation,

ManuscriptreceivedJune30,2004;revisedDecember16,2004.AbstractpublishedontheInternetJanuary25,2006.

S.IchikawawaswiththeOkumaLaboratory,DepartmentofElectricalEngineeringandComputerScience,NagoyaUniversity,Nagoya464-8603,Japan.HeisnowwithToyotaMotorCorporation,Toyota-city471-8572,Japan(e-mail:shinji@ichikawa.tec.toyota.co.jp).

M.TomitaiswiththeDepartmentofElectricalandComputerEngineering,GifuNationalCollegeofTechnology,Gifu501-0495,Japan.

S.DokiandS.OkumaarewiththeOkumaLaboratory,DepartmentofElectricalEngineeringandComputerScience,NagoyaUniversity,Nagoya464-8603,Japan.

DigitalObjectIdenti er10.1109/TIE.2006.870875

S

i.e.,magneticsaturationorrotorsaliency.Thesemethodsareeffectiveatstandstillandinlow-speedrangesbecausetheam-plitudeofhigh-frequencysignalsusedforpositionestimationdoesnotdependonrotatingvelocity.Inasmuchassomemeth-odsdonotneedmotorparameterstoestimaterotorposition,positionestimationerrorisnotcausedbyparametervariations.Onthecontrary,thelattermethodsbasedonfundamentalcomponentswereproposedstartingabout20yearsago[16],andmanymethodshavebeenproposed.Earlymethodsusedetectedterminalinformationonelectromotiveforce(EMF)[17],informationonphaseof ux[19],[23],differenceofcurrentsorvoltages[20],[22],andaslidingobserverfor uxestimation[21]toestimaterotorpositioninformation,i.e.,backEMForrotorsaliency.Thesemethodsareusefulinmiddle-andhigh-speedrangesbecausetheyusethefunda-mentalcomponentsofcontrolsignalsforpositionestimationanddonotgeneratetorquerippleornoises.However,thesemethodsusemotorparameterstoestimaterotorposition,andpositionestimationerroriscausedbyparametervariations.Inaddition,themathematicalmodelofsalient-polePMSMsiscomplicated,andpositionestimationusinginformationofbothbackEMFandsaliencyrequirescomplicatedcalculationsandapproximation.AnextendedEMF(EEMF)model[25],whichisamathematicalmodelofsynchronousmotors,hasbeenproposedforpositionestimationinIPMSMs.InasmuchasboththeEMFgeneratedbypermanent-magnetsandtheEMFgeneratedbyrotorsaliencyareincludedintheEEMFterm,positionestimationusingtheEEMFcanbeeasilyrealizedforallkindsofsynchronousmotors.Severalreports[26]–[30]havedescribedtheusefulnessoftheEEMFmodel.

SensorlesscontrolmethodsbasedonEEMFrequiremotorparameterstoestimaterotorpositionjustasothermethods.Inasmuchasmotorparametersarechangedbymagneticsatura-tionandtemperature,apositionestimationerrorisgeneratedwhentherearedifferencesbetweenactualmotorparametersandonesusedintheestimationsystem.Therefore,thesepa-rametersshouldbemeasuredinalldrivingareas,andatableofparametersshouldbemadetomaintainaccuracy[30],[31].However,parametermeasurementsarecumbersomeanddif -cult.Itishopedthattheseparameterscanbemeasuredonlineundersensorlesscontrol.

Tosolvethisproblem,severalparameteridenti cationmethodsundersensorlesscontrolhavebeenproposed[32]–[36].Somemethodsidentifymotorparametersusingspecial

0278-0046/$20.00©2006IEEE

signalsatastandstillstate[33]orundercertainloadcondi-tions[34].Inthesecases,itisdif cultbothtoidentifymotorparametersundermotorcontrolandtorespondtochangesintheseparameters.Othermethodscanidentifymotorpara-metersonline[35],[36].Inonemethod[35],theaccuracyofidenti edparametersdependsontheestimationaccuracybecauserotorpositionandvelocityareusedtoidentifymotorparameters.Inanothermethod[36],thestatorresistanceandthebackEMFconstantareidenti ed,butinductancescannotbeidenti ed.

Inthispaper,anonlineparameteridenti cationmethodandasensorlesscontrolmethodusingidenti edparametersareproposedtomaintainpositionestimationaccuracy,andsensorlesscontrolisrealizedinbothSPMSMsandIPMSMs.Inasmuchastheproposedidenti cationmethoddoesnotusepositionandvelocitytoidentifymotorparameters,identi edparametersarenotaffectedbytheaccuracyofpositionesti-mationundersensorlesscontrol.Priorparametermeasurementsarenotnecessaryusingtheproposedmethod.Althoughtheproposedmethodrequiressignalinjectiontoidentifymotorparameters,themethodcanuseanysignalthatsatis estheconditionofpersistentexcitation[37].Therefor …… 此处隐藏：23194字，全部文档内容请下载后查看。喜欢就下载吧 ……