高性能全自动电动机控制试验中英文翻译、外文文献翻译、外文翻译

附录附录1英文原文High-PerformanceAutomotiveEngineControlinEngineTesterMichitakaHori(MemberIEEE)MasahikoSuzuki(MemberIEEJ)MasakatsuNomurawemberIEEE)MasayukiTerashimamemberIEEE)MeidenshaCorporationAbstractThispaperpresentsanoveldecouplingcontrolmethodontheenginetorquecontrolfortheautomotiveenginetester.Theenginetesterismainlycomposedofadynamometercontrolsystemandanenginecontrolsystem.Theconventionalenginetesterhastheproblemthattheperformanceoftheenginetorquecontrolsystemisdeterioratedbytheinfluencesoftheinterferencebetweenthedynamometerspeedcontrolsystemandtheenginetorquecontrolsystem.Theauthorsproposedthepracticalenginetorquecontrolsystembasedonanobserverandanidentificationsystemtoeliminatetheinferenceofdynamometerspeedcontrolsystem.Theeffectofobserver'sparameterserrorontheenginetorqueestimationresponsewasanalyzed.Accordingtotheresultofthisanalysis,apracticalmethodisproposedtoidentifytheengineinertiamomentandtheshaftspringcoefficientthatareparametersoftheobserver.Theauthorsconfirmedthattheproposeddecouplingenginetorquecontrolsystemrealizedarobustcontrolsystemfromtheinterferencewiththedynamometerspeedcontrolsystemthroughsimulationandexperiments.I.IntroductionRecently,environmentalprotectionisoneofthemostimportantproblemsintheworld,andtheexhaustgasfromautomobilesisalsostrictlyregulatedbylaw.Undersuchcircumstances,theperformanceofautomotiveenginesisimprovingyearbyyear,andenginetesters,whichareusedtomeasureenginecharacteristics,arerequiredtohavehighcontrolability.However,intheconventionaltorquecontrolofengines,dynamometertorqueorshafttorqueisappliedasafeedbackvariableinsteadofenginetorquebecausetheenginetorquecannotbedetecteddirectlybecauseofitsstructure.Asaresult,thereisacouplingbetweentorquecontrolandspeedcontrol.Forthereasongivenabove,ithasbeendifficulttoattainhighcontrolabilityintheconventionalsystems.Theauthorshaveproposedanapproachtoeliminateaninterferencecomponentfromthedynamometerspeedcontrolsystem,throughestimationofenginetorqueusinganobserverandidentificationsystem.Theeffectofobserver'sparameterserroronthetorqueestimationresponsewasanalyzed.Accordingtotheresultofthisanalysis,apracticalmethodisproposedtoidentifytheengineinertiamomentandtheshaftspringcoefficientwhichareparametersoftheobserver.Weconfirmedthattheproposedenginetorquecontrolsystemrealizedarobustcontrolsystemfromtheinterferencewiththedynamometercontrolsystem.Theeffectivenessofthiscontrolsystemhasbeenconfirmedthroughsimulationsandexperiments.II.ConfigurationofEngineTesterandTestmethodA.ConfigurationofConventionalEngineTesterFig.1showsaconfigurationofaconventionalenginetester.Theenginetesterconsistsofanenginetobetested,adynamometer,andanactuatorthatregulatesthethrottlevalve.Inthissystem,theenginetorqueiscontrolledbyregulatingthepositionofthethrottlevalvethatisconnectedtotheactuatorbyawire.Detectedvariablesarethedynamometertorqueandspeed,andtheactuatorposition.Theenginecontrolleriscomposedof12-bitAID,D/AconvertersandaDSP(TMS320C25).B.EngineTestMethodTheengineistestedforitsperformanceoneachdrivingmodeshowninTable1.TheEnginespeedortorqueismaintainedtothepredeterminedpattern,whiletheexhaustgasandfuelcostaremeasured.Theengineperformanceisevaluatedbasedontheresultofmeasurements.Fig.2showstheblockdiagramofthecontrolsystemofthedrivingmode1showninTable1.Theenginetorquecanbedetecteddirectlyusingbytheindicatedmeaneffectivepressureintheory.However,inpractice,itisdifficulttodetecttheenginetorquedirectlybecauseofitsstructure.Intheconventionalenginetesting,thedynamometertorqueisappliedasafeedbackvariableinsteadoftheenginetorqueasshowninFig.2.Theenginetorquecontrolsystemisaffectedbytheaccelerationtorqueofthedynamometerinatransientcondition.Asaresult,Theperformanceoftheenginetorquecontrolsystemisdeteriorated.Weapplyanenginetorqueobservertoeliminatefrominterferenceofdynamometerspeedcontrolsystemintheenginetorquecontrol.Ill.DecouplingControlMethodA.EngineTorqueEstimationMethodTheenginetesterisequivalenttoatwo-massmodelofadynamometerandanengine.Weattemptedtoestimatetheenginetorquebytheuseofareducedorderobserver.Thestate-spacerepresentationforatwo-mass-modelshowninFig.3isasfollows,wheretheviscositytermisneglected.Theestimatedvariablesaretheenginetorque,theenginespeed,andtheshafttorque.Thereducedorderobserverequationsaregivenbelow.where,isfortheestimatedstatevectors.ValueListhegainmatrixofthereducedorderobserver.Welocatedthetriplepolesoftheobserverats=-wg[rad/s]andrefertothedynamicandstaticcharacteristicsofenginetorqueestimatedbythereducedorderobserver.B.EffectofModelingErrorinEngineTorqueEstimationAnalysiswascarriedouttoinvestigatetheeffectofaparametererrorinatwo-massmodelsystemuponobserver'sestimationresponse.Atransferfunctionfromtheenginetorquetotheestimatedenginetorqueisdefinedtoexaminetheeffectofparametererroronconditionthatthedynamometerspeedcommandiskeptconstant.Equation(5)expressestheestimatedtorquebyindicatingmodelParametersoftwo-masssystemwith,assumingthetruthvaluestobe.Theeffectofparametererrorupontheestimationresponseisgiventhesecondtermintherightsideofequation(5).Aneffectupontheestimationresponsewasinvestigatedthroughsimulations,whenerrorsarepresentedintruthvaluesandmodelvaluesofengineinertiamomentandshaftspringcoefficient.Fig.4(a)and(b)showthestepresponseofestimatedtorqueforrealenginetorquewhenparametersmj,mshowntherelationshipbetweenamodelvalueandatruthvaluewerearbitrarilychanged.Fig5(a)and(b)showfrequencycharacteristicofequation(5).Table2showsvaluesofparametersusedforsimulations.Whenthetwo-masssystemmodelisidenticalwiththetruthvalue,theresponseofestimatedenginetorquebecomesathreeordersystem,havingtriplerootsofresponsefrequencywg.Itisalsopossibletoconfirmthatengineestimationisaffectedbyresponseinthedynamometerspeedcontrolsystemortwo-masssystem,whenerrorsarepresentedinmodelparametersandtruthvalues.Accordingtotheresultsofthisanalysis,apracticalidentificationmethodhasbeenproposedfortheengineinertiamomentandtheshaftspringcoefficient.C.ParameterIdentificationinTwo-MassSystemThetransferfunctionintwo-masssystemfromthedynamometertorquetothedynamometerspeedisgivenequation(6).Fig.6showsthefrequencycharacteristicoftransferfunctiongivenbyequation(6).Thetwo-masssystemcanberegardedasaone-masssysteminalow-frequency,andasaresonancesysteminahigh-frequencydomain.Utilizingthisdistinctivenature,theengineinertiamomentisidentifiedinarangeoffrequenciesfarlowerthantheantiresonancefrequency(w1)andtheshaftspringcoefficientisidentifiedinahigh-frequencydomain.C-I.IdentificationMethodI)EngineInertiaMomentIdentificationFig.7showstheidentificationmethodofinertiamomentofone-masssysteminarangeoffrequenciesfarlowerthantheantiresonancefrequency.Enginetorqueisalwayskeptatzeroduringidentification.Identificationofinertiamomentofone-masssystemiseffectedbyinputtingalowfrequencysinewave(identificationsignal)indynamometertorquecommandandbyadjustingthemodelinertiamomentofone-masssystem,sothatdserencebecomeszerobetweendynamometerspeedandthespeedofone-masssystemmodel.Equation(7)givesaDCcomponentmj(t)inspeeddifferencebasedondifferencebetweenthemodelinertiamomentandthetruthinertiamomentbytheuseofthesignalthathasa/2shiftinphaseagainsttheidentificationsignal.Themodelinertiaisregulatedbyequation(8)sothattheDCcomponentmj(t)becomestozero.Inthesimulation,aDCcomponentisdetectedthroughathree-orderlow-passfilterinsteadoftheintegrationshowninequation(7).Thedynamometerinertiamomentisgenerallyknownandsotheengineinertiamomentcanbeidentified.2).ShaftSpringCoefficientIdentificationIdentificationofshaftspringcoefficientcanbeeffectedbyadjustingthemodelshaftspringcoefficient(Km)sothatthedeviationbecomeszerobetweentheshafthelixangle（）detectedbytheuseofadisturbanceobserverandtheshafthelixangle()estimatedbyuseofmodelshaftspringcoefficient.Fig3showstheblockdiagramoftheidentificationmethodoftheshaftspringcoefficient.Theshafthelixangle()isgivenbyequation(9)basedontheengineinertiamomentidentificationvalue()andtheshafttorque()estimatedbythedisturbanceobserver.Theestimatedshafthelixangle()is-givenbyequation(10),usingtheestimatedshafttorque(TP)andthemodelshaftspringcoefficient(Km).Whenasinewaveasanidentificationsignalisenteredindynamometertorquecommand,deviationq(t)betweentheshafthelixangle()andtheestimatedshafthelixangle()derivesfromthedifferencebetweenthemodelshaftspringcoefficientandthetruthvalue.Equation(11)givesaDCcomponentV(t)inshafthelixangledifferencebasedondifferencebetweenthemodelshaftspringcoefficientandatruthshaftspringcoefficientbytheuseofthesignalthathasashiftinphaseagainsttheidentificationsignal.Frequencyoftheidentificationsignalinvolvesthefollowingrestrictions:TheshaftspringcoefficientisidentifiedbytheregulationmethodbelowsothattheDCcomponentV(t)becomestozero.Inthesimulation,theDCcomponentisdetectedthroughathree-orderlow-passfilter.C-2.SimulationResultsofIdentificationMethodTheidentificationmethodfortheengineinertiamomentandtheshaftspringcoefficientwereverifiedandconfirmedbysimulation.Identificationsimulationwascurriedoutundertheconditionthattheinitialvalueofaninertiamomentmodelofenginewassetat1.5timesofthetruthvalue.Fig.9showstheresultofsimulationfortheidentificationofaninertiamomentofone-masssystem.Fig.10showstheresultofsimulationforshaftspringcoefficientidentificationwhentheinitialvalueofanshaftspringcoefficientmodelwassetat1.5timesofthetruthvalue.Bymakinginternalparametertuningfortheobserverbasedontheresultofidentification,itispossibletoestablisharobustenginetorqueestimation.Table2showsvaluesofparametersusedforsimulations.D.ExperimentResultsofEngineTorqueEstimationFirstly,theevaluationmethodoftheenginetorqueobserverisexplainedasfollows.Manyliteraturepresentedthatautomotiveengineshavecomplicateddynamiccharacteristics.Equation(13)showstherelationshipbetweentheenginetorque(TE)andtheboostpressure().where,:deviationfromequilibriumvaluesubscript0:equilibriumvalue,superscript*:valuenormalizedbyequilibriumvalueTheenginetorqueisdirectlyproportionaltotheboostpressurewithatimedelayasshowninequation(13),butaproportionalcoefficientvarieswiththeenginespeed.Wecurriedoutexperimentsusingbythe1800[cc],4-cylinderand4-cycleYgasolineengineinthesamecondition.Therelationshipbetweentheenginetorqueandtheboostpressurewasconfirmedbyexperiments.Thedeviationofdynamometertorquewasmeasuredinthestaticconditionthattheboostwaschangedfrom40%to70%ataconstantspeed.Thedynamometertorqueisequaltotheenginetorqueinthestaticstate.Fig.llshowstherelationshipbetweenandtheenginespeed.Thisexperimentalresultprovedthattheenginetorqueisdirectlyproportionaltotheboostpressureinthefollowingestimationexperimentalcondition.Wepracticallycannotusetheboostpressureinacaseofagasolineengineprovidedwithnosensororadieselone.However,theboostpressurewasmeasuredforthepurposeofevaluatingtheenginetorqueestimatedbytheobserver.Intheexperiment,weestimatedtheenginetorquebytheuseoftheobserver.TheobserverwasoperatedbyaDSP.Theresponsefrequencyoftheobserverisdesignedtobewg=50[rad/s].ValuesofparametersusedforexperimentsareshowninTable2.Fig.l2(a)showstheresponseresultundertheconditionthattheenginespeedischangedfrom2000[rpm]to1500[rpm]ataconstantaccelerationbythespeedcontrolofthedynamometerwhiletheenginetorqueisnotcontrolledandthethrottlevalveangleismaintainedataconstant20[%].Thedeviationoftheenginetorqueisdirectlyproportionaltothatoftheboostpressureapproximatelyasmentionedabove.Thedynamometertorquecontainsanaccelerationtorqueof2[%]duringthespeedcontrolofthedynamometerinadditiontheenginetorque.Fig.12(b)showsaresultantresponseundertheconditionthatthepositionreferenceoftheactuatorstepis20[%]ataconstantspeedof2000[rpm]whentheenginetorqueisnotcontrolled.Theestimatedenginetorqueisproportionaltotheengineboostinthisexperiment.Thedynamometertorqueisnotequaltotheestimatedenginetorqueinatransientstatebecausetheaccelerationtorqueisaddedtotheenginetorque.Weconfirmedthattheenginetorquecouldbeestimatedbythereducedorderobserverwithoutanyinfluencefromdynamometerspeedcontrolsystem.IV.DecouplingTorqueControlandExperimentalResultA.DecouplingTorqueControlMethodWeproposethedecouplingenginetorquecontrolsystemshowninFig.13.Theproposedcontrolsystemisthepracticalmethodthattheestimatedenginetorqueisfedbacktothecontrolsystemwithidentificationsystem.B.ExperimentalResultsofDecouplingTorqueControlMethodTheeffectofthedecouplingenginetorquecontrolisconfirmedinexperiments.Wecomparedtheproposedcontrolmethodwiththeconventionalone.Fig.14showstheexperimentalresultsinthecasewiththedynamometertorqueisappliedathefeedbackvariableandtheestimatedenginetorqueisapplied.Thedynamometerspeedchangesfrom2000[rpm]to1800[rpm]witharampfunctionunderaconstanttorquereference(10%),andthePIcontrollerofanenginetorqueisdesignedtohaveatheresponsefrequencyofabout2[rad/s].Forthedynamometertorque,itfluctuatesby2%oftheratedtorquebecausetheaccelerationtorquechanges.Fortheestimatedtorque,itsfluctuationvalueislessthan1%.Weconfirmedthedecouplingenginetorquecontrolusingtheobserver.ValuesofparametersusedforexperimentsareshowninTable2.V.ConclusionThispaperpresentsanoveldecouplingcontrolmethodontheenginetorquecontrolfortheautomotiveenginetester.Theconventionalenginetesterhastheproblemthattheperformanceofenginetorquecontrolsystemisdeterioratedbytheinfluencesoftheinterferencebetweenthedynamometerspeedcontrolsystemandtheenginetorquecontrolsystem.Theauthorsproposedthepracticalenginetorquecontrolsystembasedonanobserverandanidentificationsystemtoeliminatetheinferenceofdynamometerspeedcontrolsystem.Theeffectofobserver’sparameterserrorontheenginetorqueestimationresponseisanalyzed.Accordingtotheresultofthisanalysis,apracticalmethodisproposedtoidentifyengineinertiamomentandshaftspringcoefficientthatareparametersoftheobserver.Theauthorsconfirmedthattheproposeddecouplingenginetorquecontrolsystemrealizedarobustcontrolsystemfromtheinterferencewiththedynamometerspeedcontrolsystemthroughsimulationandexperiments.中文翻译高性能全自动电动机控制试验MichitakaHori(MemberIEEE)MasahikoSuzuki(MemberIEEJ)MasakatsuNomurawemberIEEE)MasayukiTerashimamemberIEEE)MeidenshaCorporation摘要这篇论文介绍了一种全自动电机测试实验的新奇控制方法.这个实验主要有一个功率控制系统和电机控制系统组成。由于功率速度控制系统和电机转矩控制系统相互干扰将会影响这个便利的电机测试系统的测试。所以作者建议实际扭矩控制系统操作时，由观察器来识别来删除功率速度控制系统的干扰。我们通过分析观察到的电机扭矩估测响应值参数误差，通过这个结果的分析，建议一个实用性的方法来确定电机的惯性转矩和轴的弹性系数这些需要观察的变量。作者肯定的是那个提出的解偶电机转矩控制系统和功率测速控制系统之间干扰的方法通过实验和仿真实现了一个合理的控制系统。I.介绍目前环境保护是一个全世界都关注的问题，连从汽车排出的尾气都有严格的法律规定。在这种情形下，汽车发动机一年一年的在改进，随之电机性能的测试也需要进一步提高。但是在传统的扭矩控制中，因为电机的构造使得电机扭矩不能被直接检测，所以只能用功率转矩法和轴的转矩来代替。结果就是在转矩控制和转速控制间存在一个耦合的关系。因为这个原因，所以对于传统的系统就很难获得一个高性能的控制。作者提议通过检测人员和识别系统来测定电机转矩，从而排除功率转速控制系统的干扰。我们通过分析观察到的电机扭矩估测响应值参数误差，通过这个结果的分析，建议用一个实用性的方法来确定电机的惯性矩和轴的弹性系数这些需要观察的变量。我们肯定的是解偶电机转矩控制系统和功率测速控制系统之间干扰的方法通过实验和仿真实现了一个合理的控制系统。II.电机构造和测试方法A.传统的电机测试结构图1描述了传统电机测试系统的结构。电机测试由一个待测电机和一个能调节气阀活门的传动装置。在这个系统中，气阀活门和传动装置通过电缆相连，通过调整活门的位置来控制电机的转矩。需要测量的变量是转矩、转速和传动装置的位置。电机控制装置是由12位的D/A，A/D和DSP(TMS320C25)组成。B.电机测试方法在表1描述了在每个驱动模式下的测试中的表现。电机的转速和转矩保持在先前已经设定模式，而排气和油料花费是测量的。电机的测量表现是在测量结果的基础上进行评价的。图2描述了在表1模式1中控制系统的框架程序图。在理论上通过指示平均压力可以直接测试电机的转矩。但是实际上因为电机的构造很难直接测试电机的转矩。在传统的电机测试中，用测功转矩作为回馈变来替代电机转矩，像图2中显示的那样。电机转矩控制系统在瞬时状态下受到测功计加速转矩的影响。结果就使得电机转矩控制系统的性能受到恶化。在电机转矩控制过程中，我们应用一个观测者来排除功率转速控制系统带来的干扰。Ill.解偶控制方法电机转矩的判断方法电极测试器与功率计和电机这二个集合模型是等效的。我们尝试着减少命令观察器判断电机扭矩。Fig.3描述了状态空间表示法的二个集合模型，其中摩擦项是被忽略的。需要判断测量的变量是电机的转矩，电机转速和轴转矩。在减少观测器的条件下等式方程如下：这里，目的是为了判定状态矢量。值L是在减少观察器而获得的矩阵。我们在时找到了观察的三电极和在减少命令观察器的前提下涉及到静态动态特征的扭矩。在电机转矩判定中模型误差的影响分析被用来核查观察器在两集合模型参数误差判定反映的结果。从电机转矩到判定电机转矩的传递函数，在功率计转速指令保持恒定的条件下，被定义成检查参量误差错误的结果。通过在双集合系统模型参数，等式(5)表达估计的扭矩值。从而计算出真实的。 参数误差的结果逼近了在等式(5)右边给出的判定响应值。当误差错误出现在真实值和电机惯性时刻，轴弹性系数时，结果通过仿真的核实将逼近判定响应值。(a)and(b)描述了估计转矩值对于真实电机转矩的阶越响应，其中参数描述了模型值与真实值之间的关系，它们是可以任意改变的。图5（a）（b）描述了等式（5）的频率特征。表格（2）描述了进行仿真后的参数值。当两级和系统模型与真实值相同时，估计电机转矩的响应值将成为一个三规则系统，一个具有相应频率是WG三重根的系统。当模型参数和真实值出现误差错误时，这个系统也能通过在测功转速控制系统和两集合系统的相应值确定电机估测值是否被影响。根据这个分析的结果，一个实际确定的方法被用来测试电机瞬时力矩和轴弹性系数。在两集合系统的变量参数识别等式（6）给出了从测功转矩到测功转速的传递矩阵。图（6）描述了等式（6）给出的传递矩阵的频率特征。两集合系统在低频时可以看作一个单系统，在高频范围中可以一个共振系统。利用这个优越的特性，在远低于返共振频率w1的一定范围内，电机瞬时力矩将可以被确定，在高频范围内，轴转矩弹性系数也可以被确定。C-I.确定方法1)电机瞬时力矩确定图（7）描述了在远低于返共振频率范围内，单个系统瞬时力矩的确定方法。在确定期间，电机转矩经常被保持在0。如果在测功转矩命令输入一个低频正弦波信号（已确定的信号）和在单系统中调整模型瞬时力矩将会影响单系统的瞬时力矩的确定方法，所以在测功转速和单系统模型转速中的差别会是0。等式（7）给出了dc参数在转速上的差别，而这个差别是基于模型瞬时力矩和真实瞬时力矩两者之间的差别，通过使用信号和相移/2相对于确定的信号。通过等式（8）来调整模型的瞬时力矩，以得到DC参数变量成为零。在仿真中，通过三次序低通道滤波来代替在等式（7）中的积分就可以确定DC的参数变量。通常的电机瞬时力矩是已知的所以电机瞬时力矩能被确定。2).轴弹性系数确定 调整模型轴弹性系数（km）能影响轴弹性系数的确定，所以导致了通过干扰观察器