外文翻译关于交流伺服驱动器的死区效应的分析与补偿Word下载.docx

外文翻译关于交流伺服驱动器的死区效应的分析与补偿Word下载.docx

《外文翻译关于交流伺服驱动器的死区效应的分析与补偿Word下载.docx》由会员分享，可在线阅读，更多相关《外文翻译关于交流伺服驱动器的死区效应的分析与补偿Word下载.docx（23页珍藏版）》请在冰点文库上搜索。

I.INTRODUCTION

Forthestorageeffectofpowerdevice,it’sverynecessarytosetadead-timeonbothswitchesinaninverterlegtoavoidtheshoot-throughinthedc-link.AccordingtotheInsulatedGateBipolarTransistor’s（IGBT）performance,weusuallysetthedeadtimeas2~5us,thisdistortsthevoltagesourceinverter’s（VSI）outputvoltage,resultinginphasecurrentdistortions,torqueripplesanddegradationsofcontrolperformance.Toovercomethedead-timeeffect,variousapproacheshavebeenstudiedandapplied.Conventionally,errorvoltageduetothedead-timeeffectcanbepre-calculatedandcompensatedbyaddingthemtothe3-phasereferencevoltageaccordingtothephasecurrentpolarity[1-2].Aon-linecompensationmethodispresentedin[3],computetheerrorvoltagebycomparethereferencevoltagewiththerealvoltage,andthenaddthistonextreferencevoltageasthenewreference.Inthisapproach,theerrorvoltagewhichcausedbymanyreasonssuchasdead-time,dc-linkvoltagedrops,etccanbeeliminated.However,estimationoftherealvoltageshouldusetheparameterofmotorwhicharenotfixedvaluebutvaryingwhileatworking.In[4],dead-timeeffectiscompensatedbymodifiedPWMgatesignalswithhardwarecircuits.In[6],amethodwhichchangethePWMgatesignalon-linebasedonthepolarityofphasecurrentispresented.Alltheapproachesmentionedabovecaneliminatethedeadtimeeffectgreatlyandhaveagoodperformance,however,withtheincreaseofcomplexityofthesoftwareandhardware,whichrestrictthepracticability.Thus,Inthepaper,amethodbasedonthe2-phasestationaryframeisproposed.Theproposedmethodproducesthecompensatingsignalbyusethephaseofthecurrentonly,doesnotneedanyadditionalhardwarecircuits.

II.ANALYSISOFDEAD-TIMEEFFECT

It’sconvenienttoanalyzethedead-timeeffectsfromonephaselegoftheinverterandextendtheresulttootherphaselegs.Figure.1showsthebasicconfigurationofone

phaselegoftheinverter.Definethedirectionofphasecurrentflowtoloadispositive.

Figure.1.onephaselegoftheinverter

A.whenphasecurrentiaispositive:

WhenupperswitchA+turnon,lowerswitchA-turnoff:

Duringthedead-timeandbeforetheA+turnoncompletely,thevoltageofpointaisclampedto–Udc/2fortheD-turnedon,thephasecurrentflowsthroughA+

untilA+turnedon,thus,theerrortimecausedbydeadtimeeffectisTd+ton,wheretonistheIGBT’sturnondelaytime.

WhenupperswitchA+turnoff,lowerswitchA-turnon:

phasecurrentflowsthroughA+beforeA+turnedoff,andflowsthroughD-whenturnedoff.Thus,theerrortimecausedbydead-timeeffectistoff,wheretoffistheIGBT’sturnoffdelaytime.

B.whenphasecurrentiaisnegtive:

Whenupperswitch（A+）-turnon,lowerswitch（A-）turnoff:

phasecurrentflowsthroughA-beforeA-turnedoff,andflowsthroughD+whenturnedoff.Thus,theerrortimecausedbydead-timeeffectistoff.Whenupperswitch（A+）turnoff,lowerswitch（A-）turnon:

Duringthedead-timeandbeforeA-turnoncompletely,thevoltageofpointaisclampedto+Udc/2forD+turnedon.Thus,theerrortimecausedbydeadtimeeffectisTd+ton.

Figure.2（a）showstheidealgatesignalpatterns,Figure.2（b）showsthepracticalsignalpatternsconsideringthedead-time,Figure.2（c）showstheactualoutputvoltageconsideringthedead-timeandswitchingtimeforia>

0andia<

0.

Figure.2.practicalswitchingpatternconsideringthedead-timeandturn

on/offtimeofswitchingdevices.（a）idealgatesignalpatterns.（b）

practicalsignalpatternsconsideringthedead-time.（c）actualoutput

voltageconsideringthedead-timeandswitchingtime

TheaverageerrorvoltagecanbecalculatedfromFigure.2asin

（1）:

（1）

Where

.

Accordingto

（1）,themostobviousimpactofdeadtimeeffectislowertheoutputvoltage,resultinginharmonicsinphasecurrent,torquerippleandspeedfluctuation,especiallyinlowspeed.

II.COMPENSATIONOFDEAD-TIMEEFFECT

Themainprincipleofthecompensationisproduceasameamplitudeandoppositesignvoltagewiththeerrorvoltagetodecreaseoreliminatetheeffectofdead-time.From

（1）,amplitudeoftheerrorvoltagewhichdependondead-time,switchperiod,switchingtimeanddc-linkmvoltageisfixedforoff-linecompensation.Thekeyofcompensationistojudgepolarityofthephasecurrent.It’sdifficulttojudgethepolarityofphasecurrentaccuratelyintherealcontrolsystem,however,andtheambiguityatthezeropointofphasecurrentwilldeterioratethecompensationeffect.Thus,inthispaper,acompensationmethodwhichdependonthephaseofcurrentonlyinthe2-phasestationary（α/β）frameisproposed.Theerrorvoltagedependonthepolarityofphasecurrent,sothreephasecurrentwillhavesixdifferentcombinationsoftheerrorvoltagevector.Table.Ishows,theerrorvoltageandit’svectoratdifferentpolarityofthreephasecurrent,where

TABLEI

ERRORVOLTAGEVECTORACCORDINGTOTHEPOLARITYOFTHREE

PHASECURRENT

Figure.3showsthecompensationvoltagevector.The

compensationvoltagevectormovediscontinuously,and

itstrajectoryishexagon.

Figure.3.compensationvoltagevector

Thecurrentvectorin2-phaserotary（d/q）frameasshowninFigure.4.ThephaseofcurrentvectorcanbeobtainedfromFigure.4as:

Figure.4.Phaseofcurrentvector

TherotorfluxorientatedcontroltechnologybeusedtocontrolPMSMintheservosystem,andrealizedecouplingbycontrol

therefore,

.

CurrentvectorwithrespecttoerrorvoltagevectorasshowninTable.II.

TABLE.II

CURRENTVECTORWITHRESPECTTOERRORVOLTAGEVECTOR

Byconsideringtheperiodicityandparityofthecompensationvoltagevectorintheα/βframe,theFouriertransformcanbesummarizedasin

（2）:

（2）

Wherenacanbecalculatedasin（3）:

（3）

Inasimilarway,nbisrepresentedasin（4）:

（4）

Using（3）and（4）through

（2）,thecompensationvoltagein/framecanbeobtainedas:

（5）

From（5）,therearenozero-phasevoltagesuchas3-th,9-thharmonics,andhigheroddharmonicshavelittleeffecttotheoutputvoltagedistortion.Therefore,itssufficienttodecreasethedead-timeeffectbycompensatefundamental,5-thand7-thharmonicsonly.Theblockdiagramofvectorcontrolcontainingdead-timecompensationasshowninFigure.5.

Figure.5.BlockdiagramofFOCcontainingthedead-time

Compensation

IV.SIMULATIONOFDEAD-TIMEEFFECT

Apreliminarysimulationanalysishasbeenperformedtotestthecompensationvalidation.ThesimulationmoduleoftheservosystemhasbeensetupinaccordancewithFigure.5.ParametersofthetestedmotorislistedinTable.III.Asoperationcondition,a60r/minreferencespeedanda2Nmloadtorquehavebeenimposed.A10KHzmodulationfrequencyforthecontrolhasbeenimposed,whilea4.8usdelaytimehasbeenconsideredasthesystem’sdead-time.

TABLE.III

PARAMETERSOFTHETESTEDMOTOR

Figure.6showsthesimulationresultsofcurrentandvoltagebeforeandafterdeadtimecompensatebeenused.Wecanseethatthephasecurrentisdistortedseverelyduetothedead-timeeffects,andthedistortionisremarkablyreducedintheproposedscheme.

（a）withoutdeadtimeeffect

（b）withdeadtimeeffect

Figure.6.Simulationresultsofvoltageandcurrent

V.EXPERIMENTRESULT

TheproposedcompensationmethodisrealizedusingaDSP-basedcontrolsystemofPMsynchronousmotor.Theprocessoristhefixed-pointDSP（TMS320LF2407A）withaclockof40MHz.TheHCPL-7840isolationamplifierisusedtomeasurethephasecurrentandthedclinkvoltage,andthemeasuredsignalsareconvertedtodigitalvaluebyusingaanalog-to-digitalconverters（ADCs）witharesolutionof10-bits.Aincrementalencoderwitharesolutionof2500-pulse/revandapositionprocessorareemployedtoobtainpositionoftherotorfluxandmeasuretherotorspeed.Theresolutionofspeedmeasurementis2r/minconsideringtheresolutionofencoderandthesystemsamplingperiod.Thethree

phaseinverterisconstructedbyusingaintelligentpowermoduleincludingsixIGBTs,gatedrives,andprotectioncircuits.Theperformanceofthecontrolsystemwiththeproposedcompensationschemeiscomparedwiththenocompensationscheme.Figure.7（a）showstheexperimentresultsofphasecurrentwithoutdead-timecompensation.Withoutacompensationscheme,thedead-timecausestheundesiredcurrentpopulationinthepahsecurrents.However,intheproposedcompensationscheme,thesedistortionisremovedperfectlyasshowninFigure.7（b）.

（a）withoutcompensation

（b）withcompensation

Figure.7.Experimentresultsofa-phasecurrent

Figure.8describesthesteady-statespeedat60r/min.Consideringtheerrorsofmeasurementis2r/min,thespeederrorwithoutdeadtimecompensationis2r/min,andthisvaluereducedto0whentheproposedcompensationmethodisapplied.

Figure.8.Experimentresults