基于人工神經(jīng)網(wǎng)絡改進型參數(shù)整定的智能pid控制研究

1、中國石油大學（華東）碩士學位論文基于人工神經(jīng)網(wǎng)絡改進型參數(shù)整定的智能PID控制研究姓名：劉燕燕申請學位級別：碩士專業(yè)：計算機應用技術(shù)指導教師：史永宏20090501ResearchonIntelligentPIDBasedonANNandImprovedParameterModificationLiuYanyan(ComputerApplicationTechnology)DirectedbyAssociateProfShiYongho

2、ngAbstractPIDcontroliswidelyadoptedinmanyfieldsbecauseofitssimplestructure，highreliabilityandeasilyimplementation。ThetraditionalPIDhasgoodcontroleffectiftheparametersofsystemmodelhavenotbigvariationHowever，晰nlthedevelopm

3、entofthemodemindustry，theprocessofindustrialproductionissometimesnonlinear，uncertainandisdifficulttOestablishaccuratemathematicmodel，SOusingtraditionalPIDcontrollerisimpossibletOachieveperfectcontrollingeffectArtificialn

4、euralnetwork(aNN)isallimportantbranchofintelligentcontrol，ithastheabilityofselfstudy，serfadaptingandselforganizingSocombinlngtheANNandthePID，formthecompoundANNPIDcontrollerwhichneednotsettheaccuratemathematicmodelItident

5、ifiesandsetsthecontrolledprocessparameterautomatically，andCansolvetheproblemofconventionalPIDcontrollerwithdifficultparametersetting，unforcementrobustnessandnotreal—timeparametertuningTheworkofthispapermainlyconsistsoftw

6、opartsAtfir觀analysisthetraditionalPIDalgorithmdeeplyForthesystem’srequestoftheintegral：whenthea唧isbiggishtheintegralshouldbereduced，contrarily，theintegrateshouldbestrengthenedBasedonthisideal，putforwardtheelasticintegral

7、PIDalgorithmbasedontheshiftedintegralThesimulationshowsthatthealgorithmCallmakethecontrolsystemmoresteadyInthenextplace，BPneuralnetworkPIDcontrollerisresearchedprimarilyinthepaperWiththeBPalgorithm，thechoiceofinitialweig

8、htinfluencesontheappearanceoflocalminimumpo訪tandimprovementofnetworkeonstringencyIfthechoiceofscopeisnotrightatthebeginningofthelearningprocess，mayarisethe“pseudosaturation’phenomenonevenenterthelocalminimumpointandthene

9、tworkisnotconvergedTherefore，usetheglobal—searchabilityofparticleswarmoptimizationifso)tooptimizetheinitialweightsofBPNN，avoidthepossibilityofslowconstringency，theexistenceoflocalminimuill，andSOonFortheprecociousphenomen