资源描述:
《the separation principle》由会员上传分享,免费在线阅读,更多相关内容在学术论文-天天文库。
1、Experiment7TheSeparationPrinciple7.1IntroductionWhenaplantismodeledbyastateequation=+(7.1)onepopularcontrolstrategyistoclosethefeedbackloopbyapplyingtheinput=−+(7.2)whereisamatrixofgainsandistheexternalinputtothefeedbacksystem.Thisstructureiscalledstatefeedback.Comb
2、ining(7.1)and(7.2),weseethattheclosed-loopsystemisgovernedby=(−)+(7.3)Thedynamicsoftheclosed-loopsystemcanbeadjustedbychoosing.Forexample,canbechosensothat+hassomespecifiedeigenvalues;infact,thisproblemisessentiallythesameastheproblemofplacingobservereigenvalues,enco
3、unteredinExperiment6.Toseethis,notethat(−)=−(7.4)Sinceformingthetransposeofamatrixdoesnotchangeitseigenvalues,thetwoproblemscoincide,with,,andplayingtherolesof,,and,respectively.Themaindrawbacktothestatefeedbackapproachisthat,inmostapplications,someofthestates
4、aresimplynotmeasureddirectly.Forexample,thespring-masssystemofExperiment6hasfourstates,onlytwoofwhicharemeasuredandavailableasoutputs.7.2TheSeparationPrincipleInthoseapplicationswherenotallstatesareaccessible,anobviousapproachtocompensatordesignistobuildanobserverfortheplant,aswe
5、didinExperiment6,andfeedbacktheestimatedstatesasiftheywereexact.Sincestateestimatesoftensufferfromconsiderableestimationerror,onewouldbejustifiedinmaintainingadegreeofskepticismthatsuchasimpleapproachwouldactuallywork.Asitturnsout,however,thesefearsareputtorestbyoneofthecentraltheo
6、remsofmoderncontroltheory,theseparationprinciple.Supposethattheplantismodeledbyastateandoutputequation=+(7.5)=(7.6)5556EXPERIMENT7.THESEPARATIONPRINCIPLEAsinExperiment6,wemayconstructastateestimatorb=(−)b++(7.7)Toclosetheloop,weapply(estimated)statefeedback=
7、−b(7.8)(Forthisexperimentwillnotneedtheexternalinputpresentin(7.2).)Combining(7.7)and(7.8),thecompensatorisgivenbyb=(−−)b+(7.9)withoutputequation(7.8).Thecompensatorhasthe(1×2)transferfunctionmatrix()=−(−++)−1(7.10)Fromequations(7.5)-(7.8),weobtainthecl
8、osed-loopsystem∙¸∙¸∙¸−=(7.11)b