摘要
Adissipative-basedadaptiveneuralcontrolschemewasdevelopedforaclassofnonlinearuncertainsystemswithunknownnonlinearitiesthatmightnotbelinearlyparameterized.Themajoradvantageofthepresentworkwastorelaxtherequirementofmatchingcondition,I.e.,theunknownnonlinearitiesappearonthesameequationasthecontrolinputinastate-spacerepresentation,whichwasrequiredinmostoftheavailableneuralnetworkcontrollers.Bysynthesizingastate-feedbackneuralcontrollertonaketheclosed-loopsystemdissipativewithrespecttoaquadraticsupplyrate,thedevelopedcontrolschemeguaranteesthattheL2-gainofcontrolledsystemwaslessthanorequaltoaprescribedlevel.Andthen,itisshownthattheoutputtrackingerrorisuniformlyultimatebounded.Thedesignschemeisillustratedusinganumericalsimulation.
出版日期
2004年02月12日(中国期刊网平台首次上网日期,不代表论文的发表时间)