资源描述:
《bp算法改进及其在pid优化控制中应用的研究》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、Subject:ImprovementofBPAlgorithmandResearchonItsApplicationinthePIDOptimizationControlSpecialty:AppliedMathematicsName:LiHu(Signature)Instructor:LiJunmin(Signature)ABSTRACTProportional-Integral-Derivativecontrol,abbreviatedasPIDcontrol,isakindofmechanismo
2、ffeedbackcontrolthatistoreducetheerrorbyproportionalfunction,toeliminatestaticdeviationbytheintegralcalculusfunctionandtopredictfuturebythedifferentialcalculusfunction.Becauseofitsclearprinciple,simplestructureandeasyimplementation,ithasbeenwidelyusedinth
3、econtrolfield.ThekeytotheapplicationofthePIDcontrolleristoselectappropriateparametersinordertoadapttodifferentobjects.Butforanobjectwithtime-varyingandlargetimedelaycharacteristics,thePIDcontroller’seffectisnotveryideal,evenhardtoachievecontrolrequirement
4、s.ThemainreasonisthatcontrolparametersofthePIDcontrollerdonothavethecapabilityofjustifyingtheparametersonline.Asanintelligentalgorithm,theBPneuralnetworkalgorithmisanapplicationmoreextensivelyusedinthefieldofartificialintelligence.Itisakindoferrorbackprop
5、agationalgorithmaccordingtothetrainingofthemultilayernetwork,hastheexcellentcapacityoflearningandadaption.Inthispaper,akindofimprovedBPalgorithmisputforward,andthisBPalgorithmisusedinthechoiceofPIDparameters,torealizetheparameterssettingandonlineadaption.
6、BecausetheinitialweightsoftheBPnetworkisselectedatrandom,thiswillgreatlyaffectthestabilityofthenetwork,thus,inthispaper,theoptimizationfunctionoftheadaptivegeneticalgorithmisusedby,puttingforwardanalgorithmbasedonadaptivegeneticalgorithmtooptimizetheweigh
7、tsoftheBPneuralnetwork,inordertoimprovetheparameterself-tuningandtheadaptiveabilityofthePIDcontroller.ThispaperistostudythebasicprincipleandthestructureoftheBPneuralnetwork,toanalyzethelimitationsoftheBPalgorithmandtomakeproperimprovementstothebasicBPalgo
8、rithm.Accordingtotheoptimizationcharacteristicofthegeneticalgorithm,thispaperistotrainthebestweightsoftheBPnetwork,andtocombinetheimprovedalgorithmwiththePIDcontrol,torealizetheon-linetuningofthePIDcontroller,sothat