基于对角递归神经网络的在线自整定解耦控制算法.pdf

《基于对角递归神经网络的在线自整定解耦控制算法.pdf》由会员上传分享，免费在线阅读，更多相关内容在行业资料-天天文库。

1、机床与液压Jun．2013HydromechatronicsEngineeringV0l_41No．12DOI：10．3969／j．issn．1001—3881．2013．12．026DecouplingControlAlgorithmofOnlineSelf-tuningBasedonDRNNTAOPing，XIAOChao．DPCenter，ChongqingSecondIntermediatePeople’sC0“，Chongqing404020，China；2．CollegeofAutomation，Chon

2、gqingUnivemity，Chongqing400044，ChinaAbstract：Inordertosolvethepuzzlethatthechangeofaloopcircuitparameterresultsinopera-tionparameterschangeofotherloopcircuitinthecontrolsystem，thepaperproposedasortofdecouplingcontrolalgorithmofonlineself—tuningbasedonDRNN．Inthe

3、paper，ittookthetern‘peratureandhumiditycentreIofacertaincontroledobjectasanexample．constructedthemath-ematicmodel，analyzedthecouplingrelationshipamongthesystemvariable，designedthede。couplingnetwork．Ittransformsthemulti-variablecontrolsystemwithcouplingrelations

4、hipastheindependentsingle-variablecontrolsystemsoastoeliminatetheefectamongrelatedcontrolchannels．BasedondecouplingalgorithmofDRNNproposedinthisPaper，itmadetheresearchonsystemsimulationexperiment，andtheresponseofsystemsimulationdemonstratedthatitisverysmalltoth

5、eefectoftwochannelsoftemperatureandhumiditycontrolafterthroughdecou·ping，andrealizedthedecouplingamongcouplingvariables．Theresultsofsimulationresearchshowthattheproposeddecouplingcontrolalgorithmisfeasibleandreasonable．Keywords：self-tuningdecouplingPIDcontrolle

6、r，diagonalrecurrentneuralnetwork，parameteradjustingstrategy，temperatureandhumiditydecoupling(RNN)hasinternalfeedback，andthereforeitcan1．Introductionreflectthedynamiccharacteristic．Inwhich，thedi．agonalrecurentneuralnetwork(DRNN)possessesInthecontrolengineering，t

7、hebasicobjectiveofstrongerprocessingandpresentationskill，bywayofdecouplingsystemistoseekanappropriatecontrolasimplificationofRNN，itcanbemoreconvenientlylaw，andwhichcanmakethemulti-variablesystemappliedtocontrolsystemtorealizethedecouplingofinputandoutputcorrela

8、tionrealizethateachout-controlofmulti．variablesystemf4—6]．putisonlycontrolledbyacorrespondinginput，anddifferentoutputisalsocontrolledbydifferentinput．2．Structureandlearninga