1、湖南大學碩士學位論文神經網絡技術在預測控制中的應用研究姓名：禹柳飛申請學位級別：碩士專業(yè)：控制理論與控制工程指導教師：王輝20030101AbstractAlongwithadvancementofindustrialcontroldemand，developmentofcontroltheoryandcomputertechnology，apredictivecontrolalgorithmisproducedwitheffectiv

2、econtrolandstrongrobustness，Itisapplicabletocomplexindustrialprocessesandthecontrolsystemthatisnoteasilytoestablishtheaccuratemathematicsmodel，andissuccessfulapplledinpetroleum，chemicalindustry，metallurgyandmechanism，and

3、haveagoodprospectinapplicationInthiSpaper，itmakesadiscussiononthebasicstructureandtheoryofdynamicmatrixpredictivecontrol，makesadetailedanalysisincludingitspredictivemodel，itsmethodsofrevisingfeedbackandrecedinghorizonopt

4、imization，itsstructureofinsidemodelcontrol(IMC)anditsstability＆rebustnessSimulationresultsconfirmadvancedofthedynamicmatrixcontrolalgorithmOnthebasiSofpointingouttheproblemofthepresentdifficultyandactuality，weproposethei

5、deathatDMCCOmbinedbyneuralnetworksActuallyitusestheneuralnetworksastheidentifledmodelofcontrolplanttoproducepredictivesignal，thecontrollawiSsolvedbyoptimizedalgorithmAccordinglywerealizethepredictivecontrolofthenonlinear

6、andtime—varietysystemWechooseBPandRBFneuralnetworksasidentifiedmodelfortheycanapproachthefunctionverywellFirstweidentifythecontrolledplantoffline，whentheprecisionreachesacertainextent，wewillachieverecursivepredictivemode

7、lbyon一1ineidentificationFinallyweacquiretheoptimizedcontrollawbyminimizingthefunctionofperformanceguidelineThiSalgorithmnotonlysolvestheproblemthatnonlinearandtime—varietyplantisdifficultytobuildmodel，butalsodecreasesthe

8、controllerealculativeburdenItbenefitsitsapplicationtoreal—timesystemThereforeitsappiicationscopeofpredictivecontrolisfurtherbroadenedIntheend，onthebasisofintroducingtechnicalflowinhydrocrackingunits，twopredictivemodelsof