不确定广义系统的鲁棒预测控制研究

《不确定广义系统的鲁棒预测控制研究》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、鲁东大学硕士学位论文不确定广义系统的鲁棒预测控制研究姓名：王利杰申请学位级别：硕士专业：运筹学与控制论指导教师：刘晓华20090611鲁东大学硕士学位论文AbstractPredictivecontrol，whichisknown鹪allMvancedofcomputercontrolapproach,hasbeenappliedsuccessfulinindustry．However,themodel，whichisusedtodescribethedynamicsofcontroUedsystem，alwayshassomeuncertainty．In

2、ordertoguaranteethespecifiedperformance，modeluncertaintymustbetakenintoaccountproperly．Robustmodelpredictivecontrolisreferredtothepredictivecontrolmethodwhichcallmaketheperformanceindexwithintheacceptablerangeinthepresenceofmodeluncertainty．Itcombinesthemethodofrobustcontrolwithth

3、emovinghorizonprincipleofMPC，andhasthecapabilityofhandlingconstraintsandoptimizationoversomeperformanceindexinasystematicway．Thefeasibilityandrobuststabilityoftheclosed—loopsystemsareguaranteed．Basedontheexistingtheoreticalresultsonmodelpredictivecontrol，thisthesisisdevotedtothede

4、velopmentoftheframeworkofrobustmodelpredictivecontrolforuncertaillsingularsystems．Toachievethisgoal，therelevanttheoryandapproaches，suchaslinearmatrixinequalities(LMI)andLyapunovstabilitytheory,robustcontrol，sampled—datacontrolandcontrollableinvariantsetandSOon,areemployedintherese

5、archwork．Themaincontentsofthisthesisarestatedasfollows：1．Theproblemofrobustoutputfeedbackmodelpredictivecontrolisstudiedforuncertainsingularsystems．BasedontheideaofvariabletransformationandLMImethods，theinfinitetimedomain“min-max'’optimizationproblemsareconvertedintolinearprogramm

6、ingproblems，apiecewisecontinuousoutputfeedbackcontrollawisobtainedandthesufficientconditionsfortheexistenceofthiscontrollawaregiven．Itisprovedthattherobuststabilityoftheclosed—loopsingularsystemsisguaranteedbytheinitialfeasibleSolutionsoftheoptimizationproblems，andtheregularandthe

7、impulse—freeofsingularsystemsarealsoheld．2．Foraclassofuncertainsingularsystems诵thinputconstraints，anLMI—basedstate-backsolutionispresentedandconditionsforguaranteeingclosed—loopH—infinityperformanceandsatisfyingthetime—domainconstraintsarcgiven．Ateachsampletime，thepiecewisecontinu

8、ousstatefeedbackpredictivecontrol