基于扩张终端约束集的非线性双模预测控制

？Aimingataclassofnonlinearsystemswithmultipleequilibriumpoints,wepresentadual-modemodelpredictivecontrolalgorithmwithextendedterminalconstraintsetcombinedwithcontrolinvariantsetandgainschedule.LocalLQRcontrollawsandthecorrespondingmaximumcontrolinvariantsetscanbedesignedforfiniteequilibriumpoints.Itisguaranteedthatcontrolinvariantsetsareoverlappedeachother.Theunionofthecontrolinvariantsetsistreatedastheterminalconstraintsetofpredictivecontrol.Thefeasibilityandstabilityofthenoveldual-modemodelpredictivecontrolareinvestigatedwithbothvariableandfixedhorizon.Becauseoftheintroductionofextendedterminalconstrainedset,thefeasibilityofoptimizationcanbeguaranteedwithshortpredictionhorizon.Inthisway,thesizeoftheoptimizationproblemisreducedsoitiscomputationallyefficient.Finally,asimulationexampleillustratingthealgorithmispresented.