简介:Basedonhighorderdynamicneuralnetwork,thispaperpresentsthetrackingproblemforuncertainnonlinearcompositesystem,whichcontainsexternaldisturbance,whosenonlinearitiesareassumedtobeunknown.Asmoothcontrollerisdesignedtoguaranteeauniformultimateboundednesspropertyforthetrackingerrorandallothersignalsintheclosedloop.Certainmeasuresareutilizedtotestitsperformance.Noaprioriknowledgeofanupperboundonthe'optimal'weightandmodelingerrorisrequired;theweightsofneuralnetworksareupdatedon-line.Numericalsimulationsperformedonasimpleexampleillustrateandclarifytheapproach.