CH2计算机控制技术(英文)_第1页
CH2计算机控制技术(英文)_第2页
CH2计算机控制技术(英文)_第3页
CH2计算机控制技术(英文)_第4页
CH2计算机控制技术(英文)_第5页
已阅读5页,还剩139页未读 继续免费阅读

下载本文档

版权说明:本文档由用户提供并上传,收益归属内容提供方,若内容存在侵权,请进行举报或认领

文档简介

Chapter2

Discrete-timeSystemsAnalysisDiscrete-timesystemsTheoryofthez-transformSignalsamplingandreconstructionPulsetransferfunctionofsampled-datasystemsStability,transientresponseandsteady-stateerrorWhatisadiscretetimesystem?Theyaresystemsinwhichtheinputsandoutputsaredescribedbydiscretesamplesintimedomain.DiscreteTimeSystemukykkykkdenotesthesamplinginstantattimet=kTsInputsandoutputsarenotcontinuousintimebutinsteadaresampledatt=kTswhereTsisthesamplinginterval.Samplingfrequency=1/TsHzor2p/Tsrad/s.ContinuoustimeDiscretesampleskuk123Discrete-TimeSystemsADiscrete-TimeSystemtransformsdiscrete-timeinputstodiscrete-timeoutputs.Theoutputataparticulartimeindexdependsonboththeinputatspecificindexvaluesandoutputvaluesatpreviousindices.Incontrasttoacontinuous-timesystemwhoseoperationisdescribed(ormodeled)byasetofdifferentialequation,adiscrete-timesystemcanbedescribedbyasetofdifferenceequations(差分方程).Howdoyoudescribetheinput-outputbehaviorofdiscretetimesystems?DosowithdifferenceequationsinsteadofdifferentialequationsExamplesofdifferenceequations(DE)1storderDE:2ndorderDE:3rdorderDE:Comparewithordinarydifferentialequations(ODE)1storderODE:2ndorderODE:ConvertingODEstodifferenceequationsApproximatebyHence,1storderODEwillleadto1storderdifferenceeqns2ndorderODEwillleadto2ndorderdifferenceeqnsThus,easytoseehowcontinuoustimesystemscanbeconvertedintoapproximatediscretetimemodels.TransformMethodsInlineartime-invariant(LTI)continuous-timesystems,theLaplacetransform

canbeusedinsystemanalysisanddesign.Inlineartime-invariantdiscrete-timesystems,thez-transform

isutilizedintheanalysisofthesystemdescribedbydifferenceequations.Whatisz-transform?SignalSampling

x(t)

载波器脉冲调制器x*(t)

x(t)tx*(t)tx(t)x*(t)SSamplingSwitchTheL-transformofx*(t):=1+e-Ts+e-2Ts+…Example:Unitstepsignalx(t)=1(t).Transcendentalfunction!X(z)iscalledthez-transformofdiscretesignalx*(t).Sincex*(t)

isasampledseriesfromthesignalx(t),wemayalsosayX(z)isthez-transformofx(t).Hencethefollowingnotation:LeteTs=z,thenInsomecases,x(kT)iswrittensimplyasx(k).

TheUnilateralZtransform

(单侧z变换)Incontrolsystemsanalysis,weusetheunilateralztransform.Justifiedbecauseincontrolsystems,weonlydealwithsignalsthatarecausal.2….k10Duetotheinfinitesum,convergenceisanimportantissue.Ideally,theregionofconvergence(ROC)shouldbestated.ROCreferstotheregiononthecomplexplaneonwhichthetransformexistsBilateralzTransforms

(双侧z变换)

Givenasamplesequence,{…x(-2),x(-1),x(0),x(1),x(2),…},wedefinethebilateralZ-transformasExample:UnitImpulseThediscreteversionofanunitimpulse(withdelay),d(t-t0),isdefinedtobeBydefinitionofthez-transform:Ifk0=0,D(z)=1!Example:impulseseries

Example:UnitStep2….k10AstepsequenceRegionofconvergenceis|z|>1.Poleatz=12….k10a<12….k10a>1Regionofconvergenceis|z|>aPoleatz=aWhatdoesthistellsusabouttherelationshipbetweenstabilityandpoles?Powerseries2….k10arampsequenceHowtoshowthis?Forunit-stepsignal:Multiplybothsideswith

(-Tz),andobtainthez-transformofunitrampfunction:Takederivativewithrespecttoz:Proof:Example:Exponentialfunction(指数函数)

x(t)=e-at(a:constantparameter〕

Thisisageometricserieswithacommonratioof(e-aTz-1)

,When|e-aTz-1

|<1,thisseriesisconvergentandcanbewritteninclosedformasfollows:Example:Sinusoidalsignal(正弦信号)

x(t)=sintPropertiesofz-transformLinearity:IfX(z)=Z[x(t)],Then

RealTranslation(TimeShift,实数位移定理)实数位移定理若

X(z)=Z[x(t)],则Proof:假定k<0时x(kT)=0ykk0123456...yk-2k0123456...shiftsequence{yk}sequence{yk-2}AppearslikeadelayedsequenceIfZ{yk}=Y(z),thenZ{yk-2}=z-2Y(z)Example:x(t)=t2,solveforX(z).Solution:x(t)=t2,x(0)=0。

x(t+T)=(t+T)2=t2+2Tt+T2

x(t+T)-x(t)=T(2t+T)Takingz-transform:Bytimeshift:

ComplexTranslation(复平移定理)Ifx(t)

X(z),then

Example:,X(z)=?

DifferentiationLaw(z域微分定理

)Example:x(t)=t3,findX(z).Proof:Ifx(t)

X(z),thenExample:,X(z)=?Solution:Bytimeshift:Consider{x(k)}asasampledseriesfromsignalx(t)=tat-1

withT=1andApplythedifferentiationlaw:Ifx

(n)

X(z),then

Z[anx(n)]=X(z/a),a:constant.

ScalingTheorem

(z域尺度定理)Solution:

Example:

Findthez-transformof

InitialValueTheorem

(初值定理)Proof:Bythedefinitionofz-transform:FinalValueTheorem(终值定理)Proof:Bythedefinitionofz-transform:andBytimeshift:FinalValueTheoremGiventheLaplaceTransform,X(s)ofsignalx(t),thefinalvalueisgivenbyGiventhedefinitionoftheztransform,X(z)ofasequence,Inbothcases,weassumethatthesignalisstable.Thediscreteconvolutionoftwosampledsequencesx(nT)andy(nT)

isdefinedas

,then:

Z[x(nT)*y(nT)]=X(z)Y(z)

ConvolutionofTimeSequence(离散卷积定理

)Proof:Letn-k=m,thenm=-kwhenn=0.Substituten=m+kintoEqn(1)

:Usefulz-transformpairsBypartialfractionexpansion:TakinginverseL-transform:Takingz-transform:Example:thecasewhenthesignalisgivenbyitsL-transform.InverseztransformGivenX(z),whatisthesequencex(k)?Powerseriesexpansion(幂级数法)longdivisiontoobtainsequencePartialfractionexpansion(部分分式展开法,又称查表法)makesuseofknownz-transformpairstofindtheclosedforminversesolutionInversetransformintegral(反演积分法、留数法)makesuseofResiduecalculus

Powerseriesexpansion(幂级数法)

X(z)canbewrittenasarationalfunction:

Usinglongdivision(综合除法)toobtainasequence

:kckTx=)(Thenbydefinition:

)=?()(kxazzzX,+=Solution:Example:Example:Given,Findthecorrespondingsignalx*(t).Solution:(2)Partialfractionexpansion

(部分分式展开法,又称查表法)DecomposeX(z)intoPartialfractionexpansionandthenmakeuseofknownz-transformpairstofindtheclosedforminversesolution.Example:Given,x(kT)=?Solution:Example:

Given,finditsinverseztransform.

Solution:FirstdecomposeX(z)/zintopartfractionexpansionx(nT)=(-1+2n)10

x*(t)=x(0)(t)+x(T)(t-T)+x(2T)(t-2T)+…

=0+10(t-T)+30(t-2T)+70(t-3T)+…

Fromthelook-uptable:=x(0)+x(T)z-1+x(2T)z-2+…

Multiplybothsideswithzk-1:X(z)zk-1=x(0)zk-1+x(T)zk-2

+…+x(kT)z-1+…

TheaboveisaLaurentseries,andx(kT)isthecoefficientofz-1.Applyingtherelevanttheoremintheoryofcomplexvariables:

(3)Inversetransformintegral(反演积分法、留数法)WhereCisaclosedcontourintheROCofX(z).

ByCauchyintegraltheorem:

wherezi

isthepoleofX(z)zk-1insidecontourC.ResiduecalculusIfhasapolewithmultiplicityr,andanother(m-r)polesatdifferentlocations,thenSolution:Example:Given,Findthecorrespondingsignalx*(t).SolutiontoDifferenceEquation

—Numericalsolution:

SequentialProcedure(迭代法):computetheD.E.recursivelyfromsomeinitialvalues.usedincomputersolutionofD.E.—Analyticalsolution:1.assumingsomeformforthesolutionwithunknownconstantsandsolvefortheconstantstomatchtheinitialconditions.2.usingz-transformThez-transformApproach(z变换法)

Determinethez-transformoftheD.E.usingtherealtranslationproperty;SolvethealgebraicequationforY(z);Obtainthesolutionofy(k)bytakingtheinversez-transform.Example:FindthesolutiontothefollowingD.E.y(0)=-1Solution:Takingthez-transformonbothsidesofD.E.Fromthelook-uptable,wehaveDiscreteTransferFunctionConsiderafirstorderdifferenceequationisgivenby:TakingZ-transformsonbothsides:a,bareconstantsTransferfunctioninz-domain!Poleatz=aFora2ndorderdifferenceequation:TakingZ-transformsonbothsides:Polesatz=-1,-2

ukykWhatistheunitstepresponseofthisdiscretetimesystem?Z.T.NotetheneedtosetzasideTakinginversez-transform:Thez-transformoftheoutput:fork=0,1,2,3,4,…Outputsequence:Outputsamplesgettinglargerforlargek.Thereforesystemisunstable!DoesnotgetlargerwithkDoesnotgetlargerwithk

GrowslargerwithkRecallthatthepolesareatz=-1and-2.Henceweconcludethatpoleswithmagnitudes|z|>1leadtounstablesystems!Poleswithmagnitudes|z|<1arestable.SignalSamplingandReconstructionSignalSampling

x(t)x*(t)SSamplingSwitch(a)

tx(t)(b)tx*(t)Obviously,isaperiodicfunction,hencecanbeexpandedintoFourierseries:whereisthesamplingfrequency.

HenceTakingLaplacetransformandusingcomplextranslationtheorem:Itsspectrum(频谱)canbegivenby--X(j)00--(a)

Spectrumofx(t).(b)Spectrumof

x*(t)(>2)

Ideallow-passfilterSpectrumpreserved,Signalx(t)canberecoveredAliasing(混叠):ωs

<2ωmaxSpectrumoverlap,Signaldistorted,Cannotberecovered.NyquistSamplingTheorem(采样定理)NyquistSamplingTheorem:Onecanrecoverasignalfromitssamplesifthesamplingfrequency(ωs

=2π/T)isatleast

twicethehighestfrequency(ωmax

)inthesignal,i.e.,Putinanotherway:Foragivensamplingfrequencyωs,onlywhenthehighestfrequency(ωmax

)ofthesignalisnolargerthanhalfofsamplingfrequency(ωs)canwerecoverthesignalwithoutanydistortion,i.e.,NyquistfrequencyIdeallow-passfilter-IdealReconstructionofSignalAfterfiltering:Impulseresponse:Noncausal!Cannotbeimplementedphysically.t/T123-1-2-3Signal

Reconstruction:

apolynomialextrapolationapproach.

UsingaTaylor’sseriesexpansionaboutt=nT,Wedefineasthereconstructedversionofx(t).Suchamechanismiscalleddatahold,andxh(t)istheoutputofthedatahold.

IfonlythefirsttermoftheTaylor’sseriesisused,thedataholdiscalleda

zero-orderhold(零阶保持器),i.e.,IfthefirsttwotermsoftheTaylor’sseriesareused,itisthefirst-orderhold(一阶保持器),i.e.,Weapproximatethederivativesbybackwarddifference.Zero-OrderHold(ZOH,零阶保持器)ZOHisthemostcommonlyuseddatahold,itmaintainsthesampledvalueforthewholesamplingperiod,andoutputastaircasesignal.xh(t)x*(t)x*(t)t

ZOHxh(t)tTakingLaplacetransform:

HencethetransferfunctionofZOHisgivenbyThenSamplingandHoldxh(t)Gh(s)x*(t)x(t)Sampler

DataHoldForZOH:FrequencyresponseofZOHAmplitude:Phase:FrequencyresponseofZOHTHighfrequencycomponentsareattenuated,butcannotbetotallyerased;PhasedelayrelatedtoT.First-OrderHold(FOH,

一阶保持器)0T2T3T…..Itsfrequencyresponse:ThetransferfunctionofFOHisgivenbywhereFrequencyresponseofFOHConclusion:FOHisnotbetterthanZOH.FOHZOHPulsetransferfunction

(脉冲传递函数)1.

OpenloopPulsetransferfunction

G(s)r*(t)

r(t)y*(t)

y(t)Pulsetransferfunction(ztransferfunction,discretetransferfunction)isdefinedastheratioofthez-transformofoutputy*(t),

orY(z),tothatofinputr*(t),orR(z),i.e.,H(z)=Y(z)/R(z).Anycontinuous-timesignalr(t)sampledbyanidealsamplerwithperiodTwillproduceatrainofpulsesignalas:If

isinputintoG(s),Iftheinputis,Assumingthatthecontinuousoutputc(t)isalsosampledbyanidealsamplerasthatofinput,thentheoutputsampleatt=nT

isBythetheoremofdiscreteconvolution:G(z)=Z[G(s)]GenerallyG(z)

canbewrittenas:Caution:G(z)isdeterminedbythestructureandparametersofthediscretesystem,andisindependentofthereferenceinput.Example:findtheztransferfunctionforthesystemwiththefollowingstransferfunction:Solution:Example:determinethepulsetransferfunctionforthefollowingopen-loopsampled-datasystem:r*(t)

r(t)y*(t)y(t)Solution:PulsetransferfunctionofcascadedsystemsCase1:NosamplerbetweentwocascadedsubsystemsG1(s)G2(s)

r*(t)

r(t)y*(t)y(t)TheblockdiagramcanbereducedtoG1(s)G2(s)

r*(t)

r(t)y*(t)y(t)ThenLetCase2:Thereisasamplerbetweentwocascadedsubsystems,andsamplersaresynchronized.

y*(t)y(t)G1(s)G2(s)

r*(t)

r(t)y1*(t)Case3:OpenloopsystemprecededbyaZOH.Gp(s)

r*(t)

r(t)y*(t)y(t)3.Pulsetransferfunctionofclosed-loopdiscretesystemsy*(t)G1(s)G2(s)H(s)r(t)

e(t)e*(t)

d(t)

b(t)

y(t)-++Figure:LineardiscretesystemwithdisturbanceByassumingd(t)=0,thediagramcanbereducedto:

Figure:LineardiscretesystemBythedefinitionofpulsetransferfunction:G1(s)G2(s)H(s)r(t)

e*(t)

y*(t)y(t)b(t)DefinetheerrorpulsetransferfunctionGe(z)

as:Hencetheclosed-looppulsetransferfunctionGB(z)isgivenbyNowassumer(t)=0,andobtainthefollowingdiagramwithdisturbanceasanequivalentinput:G2(s)G1(s)H(s)r(t)=0

e*(t)

y*(t)-y(t)

d(t)++Figure:Lineardiscretesystemwithdisturbanceasinput.Example:Considerthefollowingsampled-datasystem:G(s)H(s)

r(t)b*(t)y*(t)y(t)-

AnalysisofDiscreteSystemsTransientresponseStabilitySteady-stateerror1.

Transientresponse

Closed-looptransferfunctionofatypicaldiscretesystem:N(z)andD(z)aremonicpolynomialofz.TheunitstepresponseisgivenbyBypartialfractionexpansion:where(1)pkisreal:Casea:pk=1,yk(n)isaconstantsequence.Theoutputseries:Caseb:0<pk<1,decayinggeometricsequenceCasec:pk>1,expandinggeometricsequence.Casee:-1<pk<0,decayinggeometricsequencewithalternatingsigns.

Cased:pk=-1,alternatingsequence.

Casef:pk<-1,expandinggeometricsequencewithalternatingsigns.

Summary:transientresponsewith

asinglerealpolepkImRe[Z]

f

f

d

daa

c

c

b

bee(2)

pkisconjugatecomplex(inpairs)Then,ckandck+1formaconjugatepair:

Themagnitudeofpole,|pk|,willdeterminewhethertheresponseisconvergentordivergent.Thetransientresponse:Casea:|pk|<1,dampedsinusoidalsequenceCaseb:|pk|=1,sinusoidalsequenceCasec:|pk|>1,exponentiallyexpandingsinusoidalsequenceAlargermeansfasteroscillationinthetransientresponse.Let

θk=ωdT,then

istheoscillatingfrequencyoftheresponse,andtheperiodofoscillation

isgivenbyTheimpactoftheargument(3).Deadbeatsystem(有限时间响应系统)

Whenalltheclosed-looppolesareattheorigin,thetransientresponsewillsettledownwithinlimitedperiods.Suchasystemiscalleddeadbeatsystem.Theunitimpulseresponse:Thetransientprocesswilldieoutafternperiods.Thispropertyisneverfoundinacontinuous-timesystem.Averyimportantqualitativepropertyofadynamicsystemisstability.Internalstabilityisconcernedwiththeresponsesatalltheinternalvariables.Externalstabilityisconcernedwiththeinput-outputrelation.ThemostcommondefinitionofappropriateresponseisthatforeveryBoundedInput,weshouldhaveaBoundedOutput.i.e.,wecallthesystemBIBOstable.2.StabilityAnalysis

LinearDiscreteSystem:G(s)

r(t)

y*(t)y(t)_

Ifallclosed-looppolesofasystemisinsidetheunitcircle,thesystemisstable.Ifatleastonepoleisonoroutsidetheunitcircle,thecorrespondingsystemisnotBIBOstable.1+G(z)=0Thestabilityboundaryofdiscrete-timesystems(inthe

z-plane)isdifferentfromthatofcontinuoussystems(inthes-plane).Howdoesthishappen?Considerthefollowingmapping(fromstoz):

z

=eTs

Foranypointinthes-plane:s=σ+jω,thenaftermapping,thepointinz-planeis:

case1:σ=0,theimaginaryaxisins-planeismappedintotheunitcircleinz-plane–stabilityboundary.

case2:σ<0,theLHPofs-planeismappedintotheinterioroftheunitcircleinz-plane–stabilityregion.

case3:σ>0,theRHPofs-planeismappedintotheexterioroftheunitcircleinz-plane–instabilityregion.s=σ+jωReReImImMappingthes-planeintoz-planes-planez-planes=σ+jωWaystocheckstabilityDirectcalculation:forsimplecases;Bilineartransform+Routhtest;Jury’stest:similartoHurwitztestincontinuous-timecase.Otherways:rootlocus,Nyquiststabilitycriterion,Lyapunovtheorem,etc.Solution:

TheopenlooppulsetranserfunctionisExample:Checkthestabilityofthefollowingsampled-datasystemwithT=1s.

r(t)y*(t)

y(t)1+G(z)=0z2+4.952z+0.368=0z1=-0.076z2=-4.876Thereisonepoleoutsidetheunitcircle,hencethe

systemisunstable.Theclosed-loopC.E.isgivenby

Define(1)Thetwocomplexvariableszandwcanbewrittenasz=x+jyw=u+jv(2)(3)Substitute(2)and(3)into(1):Bilineartransform+RouthtestthenBilineartransform,w-transformCase1:x2+y2=1,theunitcircleinz-plane,

u=0theimaginaryaxisinw-plane.Case2:x2+y2<1,theinteriorofunitcircleinz-plane,

u<0thelefthalfofw-plane.Case3:x2+y2>1,theexteriorofunitcircleinz-plane,

u>0therighthalfofw-plane.z=x+jyw=u+jvForDiscrete-timesystems:polesareinsideunitcircle(zplane)?Stability?ForContinuous-timesystems:polesareonthelefthalfplane(wdomain)?BilineartransformRouthtest

Givenasampled-datasystemwithT=1s.

Checkitsstabilityforthecasewhen

K=10,andfindthecriticalgainK.Example:Solution:⑴Theclosed-loopCEis:

z2+2.31z+3=0Bymanualcalculation:

1=-1.156+j1.292=-1.156-j1.29Bothpolesareoutsidetheunitcircle,hencethesystemisunstable.

C(s)R(s)

—Open-looppulseTF:when

K=10,theclosed-loopTF:CE:1+G(z)=0

z2-(1.368-0.368K)z+(0.368+0.264K)=0⑵Openlooppulsetransferfunction:Thecriticalvalueofgain

Kis:

Kc=2.4Routharray:

w20.632K2.736-0.104K

w11.264-0.528K0

w02.736-0.104K

Forstability,weneedAfterw-transform:

0.632Kw2+(1.264-0.528K)w+(2.736-0.104K)=00<K<2.4StabilityConditions

for2ndordermonicC.E.CE:f(z)=z2+az+b=0Sufficient&NecessaryConditionsforStability:f(1)>0f(-1)>0|f(0)|<13.Steady-stateerrorindiscrete-timesystemsConsideradiscrete-timesystemwithunitfeedback:G(s)

r(t)

温馨提示

  • 1. 本站所有资源如无特殊说明,都需要本地电脑安装OFFICE2007和PDF阅读器。图纸软件为CAD,CAXA,PROE,UG,SolidWorks等.压缩文件请下载最新的WinRAR软件解压。
  • 2. 本站的文档不包含任何第三方提供的附件图纸等,如果需要附件,请联系上传者。文件的所有权益归上传用户所有。
  • 3. 本站RAR压缩包中若带图纸,网页内容里面会有图纸预览,若没有图纸预览就没有图纸。
  • 4. 未经权益所有人同意不得将文件中的内容挪作商业或盈利用途。
  • 5. 人人文库网仅提供信息存储空间,仅对用户上传内容的表现方式做保护处理,对用户上传分享的文档内容本身不做任何修改或编辑,并不能对任何下载内容负责。
  • 6. 下载文件中如有侵权或不适当内容,请与我们联系,我们立即纠正。
  • 7. 本站不保证下载资源的准确性、安全性和完整性, 同时也不承担用户因使用这些下载资源对自己和他人造成任何形式的伤害或损失。

评论

0/150

提交评论