Example: confidence

3.1 State Space Models - Rutgers University

,obtainedviadualitywiththecontinuous-tim emodels, continuous-timedynamicsystemcanbederived eitherfromthesystemmodelgiveninthetimedo mainbya thephasevariableform(controllerform),the observerform,themodalform,andtheJordanfo rm whichareoftenusedinmoderncontroltheoryan dpractice, generalth-ordermodelofa dynamicsystemrepre-sentedbyanth-orderdif ferentialequation ( )Atthispointweassumethatallinitialcondit ionsfortheabovedifferentialequation, , systematicprocedurethattransformsadiffer entialequationofordertoa statespaceformrepresentinga systemoffirst-orderdifferentialequations ,wefirststartwitha simplifiedversionof( ),namelywestudythecasewhenno9596 STATESPACEAPPROACH derivativeswithrespecttotheinputareprese nt ( )Introducethefollowing(easytoremember)ch angeofvariables .. ( )whichaftertakingderivativesleadsto.

The block diagram for this decomposition is given in Figure 3.1. U(s) V(s) V(s)/U(s) Y(s)/V(s) Y(s) Figure 3.1: Block diagram representation for (3.17) Equation (3.17a) has the same structure as (3.6), after the Laplace transformation is applied, which directly produces the state space system equation identical to (3.9). It remains to find ...

Tags:

  Structure, Diagrams

Information

Domain:

Source:

Link to this page:

Please notify us if you found a problem with this document:

Other abuse

Transcription of 3.1 State Space Models - Rutgers University

1 ,obtainedviadualitywiththecontinuous-tim emodels, continuous-timedynamicsystemcanbederived eitherfromthesystemmodelgiveninthetimedo mainbya thephasevariableform(controllerform),the observerform,themodalform,andtheJordanfo rm whichareoftenusedinmoderncontroltheoryan dpractice, generalth-ordermodelofa dynamicsystemrepre-sentedbyanth-orderdif ferentialequation ( )Atthispointweassumethatallinitialcondit ionsfortheabovedifferentialequation, , systematicprocedurethattransformsadiffer entialequationofordertoa statespaceformrepresentinga systemoffirst-orderdifferentialequations ,wefirststartwitha simplifiedversionof( ),namelywestudythecasewhenno9596 STATESPACEAPPROACH derivativeswithrespecttotheinputareprese nt ( )Introducethefollowing(easytoremember)ch angeofvariables .. ( )whichaftertakingderivativesleadsto.

2 ( )STATESPACEAPPROACH97 Thestatespaceformof( )isgivenby ..( )withthecorrespondingoutputequationobtai nedfrom( )as .. ( )Thestatespaceform( )and( )is ( ),whichincludesderivativeswithrespecttot heinput,weformanauxiliarydifferentialequ ationof( )havingtheformof( )as ( )98 STATESPACEAPPROACH forwhichthechangeofvariables( )isapplicable .. ( )andthenapplythesuperpositionprincipleto ( )and( ).Sinceistheresponseof( ),thenbythesuperpositionpropertytherespo nseof( )isgivenby ( )Equations( )producethestatespaceequationsintheforma lreadygivenby( ).Theoutputequationcanbeobtainedbyelimin ating from( ),byusing( ),thatis Thisleadstotheoutputequation .. ( )It is interestingtopointoutthatfor , whichis almostalwaysthecase,theoutputequationals ohasaneasy-to-rememberformSTATESPACEAPPR OACH99givenby .. ( )Thus,insummary,fora givendynamicsystemmodeledbydif-ferential equation( ),oneisabletowriteimmediatelyitsstatespa ceform,givenby( )and( ),justbyidentifyingcoeffi-cients and , :Considera dynamicalsystemrepresentedbythefollowing differentialequation!

3 #"%$!#&%$!(')$! $! $!#*%$! $where! $standsforthethderivative, ! $ .Accordingto( )and( ), ,wepresenttwomethods,knownasdirectandpar allelprogrammingtechniques, ,likeintheprevioussubsection, ,wherebytransferfunctionmodeswemeanpoles oftheoriginaltransferfunction(beforezero -polecancellation,ifany,takesplace).Ifso mezerosandpolesinthetransferfunctionarec ancelled, ,wefirstusedirectprogrammingtechniquesto derivethestatespaceformsknownasthecontro llercanonicalformandtheobservercanonical form;then,bythemethodofparallelprogramin g, convenientinthecasewhentheplanttransferf unctionisgivenina nonfactorizedpolynomialform+++ , -+ , --.++ , -+ , --.( )Forthissystemanauxiliaryvariableisintro ducedsuchthatSTATESPACEAPPROACH101thetra nsferfunctionissplitas// 0 1/ 0 112( )/// 0 1/ 0 112( )Theblockdiagramforthisdecompositionis (s)V(s)V(s)/U(s)Y(s)/V(s)Y(s) :Blockdiagramrepresentationfor( )Equation( )hasthesamestructureas( ),aftertheLaplacetransformationis applied,whichdirectlyproducesthestatespa cesystemequationidenticalto( ).

4 It remainstofindmatricesfortheoutputequatio n( ).Equation( )canberewrittenas333 4 53 4 556( )indicatingthatis justa ( )maybeconsideredasa differentialequationintheoperatorformfor zeroinitialconditions,where. Inthatcase,,, andaresimplyreplacedwith,,and, , ,adders,subtracters,andmultipliers, ,functiongeneratorsareusedto simulationdiagramis is relativelyeasytodraw(design)a simulationdiagramfora givendynamicsystem, ( ) ,respectively,by7#8:97#8 Useformula( )toconstruct, ( ,whererepresentstheintegratorblock).From ( )wehavethat7#8:98 ; <7#8 ; uHyIbJ0bnbJ2bJ1-a0-a1-an-1vK(n-1)vK(1)v1 :Simulationdiagramforthedirectprogrammin gtechnique(controllercanonicalform)Asyst ematicproceduretoobtainthestatespaceform froma simulationdiagramistochoosetheoutputsofi ntegratorsasstatevariables.

5 Usingthisconvention, , P ( )104 STATESPACEAPPROACHandQQRSSRR T SR T SRR( )Thisformofthesystemmodeliscalledthecont rollercanon-icalform. Itisidenticaltotheoneobtainedintheprevio ussec-tion equations( )and( ).Controllercanonicalformplaysanimportan troleincontroltheorysinceit is importanttopointoutthattherearemanystate spaceformsfora givendynamicalsystem, ,togetherwiththedevelopmentofotherimport antstatespacecanonicalforms,canbefoundin Kailath(1980; ).NotethattheMATLAB functiontf2ssproducesthestatespaceformfo ra giventransferfunction,infact,it ( )andSTATESPACEAPPROACH105( ),thestatespacecontrollercanonicalformis givenbyandDirectProgrammingTechniqueandO bserverCanonicalFormInadditiontocontroll ercanonicalform, ( )iswrittenintheformZZ [ \Z [ \\]ZZZ [ \Z [ \\]( )andexpressedasZZ [ \Z [ \\]ZZZZ [ \Z [ \\]( )106 STATESPACEAPPROACH leadingto^ _=`a^ _a^ _ ``^b^^ _ `a^ _a^ _ ``^b( )Thisrelationshipcanbeimplementedbyusing a simulationdi-agramcomposedofintegratorsi na cascade, ,termscontainingshouldpassthroughonlyone integrator,signals^ _aand^ _ashouldpassthroughtwointegrators, ,signalsbandbshouldbeintegrated-times, (t)dy(t)exf2-a0+1/sxf2bc1-a1+1/sxfn-1xfn 1/sxfn-1xfnbcn-1-an-1+1/sbcn+.]]]]

6 Simulationdiagramforobservercanonicalfor mDefiningthestatevariablesastheoutputsof integrators,andrecordingrelationshipsamo ngstatevariablesandthesystemout-put,wege tfromtheabovefiguregg( )hiiigiigjhhhhhghhgkjjjjjgjjggg lhg lhg lhg lhg lhgg lhg lhg( )Thematrixformofobservercanonicalformise asilyobtainedfrom( )and( ) qop qnp qnp( )andp( ) veryusefulforcomputersimulationoflineard ynamicalsystemssinceit ,thisformrepresentsanobservablesystem,in thesensetobedefinedinChapter5,whichmeans thatallstatevariableshaveanimpactonthesy stemoutput,andviceversa,thatfromthesyste moutputandstateequationsoneisabletorecon structthestatevariablesSTATESPACEAPPROAC H109atanytimeinstant,andofcourseatzero,a ndthus,determinerstintermsoftheoriginali nitialconditionsuuturutur. ,withoutlossofgenerality,thatthepolynomi alinthenumeratorhasdegreeof, thenvrstrrsstt( )Hererstaredistinctrealroots(poles) formis (t)wy(t)x yxz2xz2-p2+k{21/sxz1xz1-p1+k{11/sxznxzn- pn+k{n1 :Thesimulationdiagramfortheparallelprogr ammingtechnique(modalcanonicalform)Thest atespacemodelderivedfromthissimulationdi agramisgivenby|}.}}

7 ~..|}~( )Thisformisknownintheliteratureasthemoda lcanonicalform(alsoknownasuncoupledform) . :Findthestatespacemodelofa andthestatespaceformobtainedbyusing( )and( )ofthedirectprogrammingtechniqueisNoteth attheMATLAB functiontf2ssproduceswhichonlyindicatesa permutationinthestatespacevariables,that isEmployingthepartialfractionexpansion(w hichcanbeobtainedbytheMATLAB functionresidue),thetransferfunctioniswr ittenas112 STATESPACEAPPROACHT hestatespacemodel,directlywrittenusing( ),isNotethattheparallelprogrammingtechni quepresentedis , :Leta transferfunctioncontaininga pairofcomplexconjugaterootsbegivenbyWe firstgroupthecomplexconjugatepolesina second-ordertransferfunction,thatis Then, ,correspondingtothepairofcomplexconjugat epoles,isimplementedusingdirectprogrammi ng, ,wherethecontrollercanonicalformis usedtorepresenta sothattherequiredstatespaceformisu(t)wy( t)x yxz2xz2-10+3 1/sxz1xz1-5+ 21/sxz4xz3xz3xz4-2-2+8 8 1/s1.

8 Simulationdiagramforasystemwithcomplexco njugatepoles114 STATESPACEAPPROACHM ultipleRealRootsWhenthetransferfunctionh asmultiplerealpoles, realpole ofthetransferfunctionhasmultiplicityandt hattheotherpolesarerealanddistinct,thati s > Thepartialfractionformoftheaboveexpressi onis > @ Thesimulationdiagramforsucha systemis (t) k1r-1k11u(t)x 2-p1+ 1/sx rx r-p1+ 1/sx r+1-pr+1+ k r+11/sx n-pn+1/sk nx r+1x nx 2x 1x 1-p1+ k 1r1 :ThesimulationdiagramfortheJordancanonic alformTakingforthestatevariablestheoutpu tsofintegrators,thestateSTATESPACEAPPROA CH115spacemodelisobtainedasfollows .. % .. > .. ( ) % > > > Thisformofthesystemmodelis knownastheJordancanonicalform. ThecompleteanalysisoftheJordancanonicalf ormrequiresa ,understandingtheJordanformis crucialforcorrectinterpretationofsystems tability,henceinthefollowingchapter.

9 Findthestatespacemodelfromthetransferfun ctionusingtheJordancanonicalform Thistransferfunctioncanbeexpandedas (ordinaryth-orderdifferentialequation,st atespaceortransferfunction), differentialth-ordersystemmodelintheoper atorform wheretheoperatorisdefinedas and, thenthecharacteristicequation, accordingtothemathematicaltheoryoflinear differentialequations(BoyceandDiPrima,19 92),isdefinedby ( )Notethattheoperatorisreplacedbythecompl exvariableplayingtheroleofa ( )that Thecharacteristicequationhereisdefinedby ( ) single-inputsingle-outputsystemis ( ) ,thecharacteristicpolynomial(obtainedfro mthecorrespondingcharacteristicequation) , ( )and( ),butisnotsoclearfrom( ).It isleftasanexercisetothereadertoshowthat( )and( )areidentical( ).Theeigenvaluesaredefinedinlinearalgebr aasscalars,,satisfying(FraleighandBeaure gard,1990)( )118 STATESPACEAPPROACH wherethevectorsarecalledtheeigenvectors.

10 Thissystemoflinearalgebraicequations(isf ixed)hasa solutionifandonlyif( )Obviously,( )and( ) ( )=( ),it followsthatthelastequationisthecharacter isticequa-tion, , thenumberofeigenvaluesisequalto. Thus,therootsofthecharacteristicequation inthestatespacecontextaretheeigenvalueso fthema-trix. Theserootsinthetransferfunctioncontextar ecalledthesystempoles, accordingtothemathematicaltoolsforanalys isofthesesystems havepointedoutbeforethata ,weestablisha relationshipamongthosestatespaceformsbyu singa givensystem wecanintroducea newstatevectorbya ( )STATESPACEAPPROACH119where = ( ) ,thatis, ( )Notethatinthisproofthefollowingproperti esofthematrixdeterminanthavebeenused thatthetransferfunctionremainsthesamefor bothmodels,whichcan120 STATESPACEAPPROACH beshownasfollows ( )Notethatwehaveusedin( )thematrixinversionproperty(AppendixC) Theaboveresultisquitelogical thesystempreservesitsin-put outputbehaviornomatterhowit is , where thenhave ( )


Related search queries