基于改进的PSO在并联机构神经网络控制系统中的应用.pdf

《基于改进的PSO在并联机构神经网络控制系统中的应用.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、Jun．2015HydromechatronicsEngineeringVo1．43No．12HydromechatronicsEngineeringhttp：／／jdy．qks．cqut．edu．onE-mail：jdygcyw@126．tomApplicationofimprovedPS0一basedtoneuralnetworkcontrolsystemofparallelmechanismChang-jianWANG，PengWANG(SchoolofMechanicalEngi~ering，YangtzeUniversity，fin~hou4340

2、00，China)Abstract：AsthetraditionalPIDneuralnetworkcouldnoteffectivelycontrolthereal—timenonlinearmultivariablesystem．thispaperproposedanewtypeofmuhivariableadaptivePIDneuralnetworkcontroller．rrhiscontrolsys-terncouldputoutfeedbackandactivationfeedback，withthefunctionofproportion，in

3、tegrationanddifferentia．tion．WeusedtheParticleSwarmAlgorithmwhichisbasedonthesolutionspacedivisiontooptimizetheparame．tersofthecontroller．Italsocouldeliminateefectofinitialvsluesontheaccuracyofthecontrollerandcanbeappliedtotheparallelmechanismcontrolsystem．Asthesimulationresultssho

4、wn，controllerhadhigherpreci-sion．betterrobustnessandadaptability．Thisresearchprovidedatheoreticalbasisfortheoptimizationdesignandperformanceanalysisoftheparallelmechanism．Keywords：PIDneuralnetwork，Parallelmechanism，Improvedparticleswarmalgorithmdoi：10．3969／j．issn．1001-3881．2015．12．

5、010Documentcode：ACLCnumber：TH165+．2nonlinearmuhivariablesystemsanduncertaintiesandotherfactomoutsideinterference．SoPIDcontro1anditscombinationwithothercontroltheorycanbeusedComparingtotheseriesrobot，theparallelrobothastosolvesuchproblemsandfavoredbythemajorityofhighstiffness，strong

6、bearingcapacity，highprecisionresearchers．Neuralnetworkhaspowerfulcomputing，andcompactstructure．Itcouldbesuitableforsomestrongrobust，highfaulttoleranceandself—learningapplicationsinmachining，aircraftmanufacturing，andthatcanbeapproachedtocontinuous1inearfunction，healthcarewhichhavesm

7、allworkspaceandlargebutthepresenceofslowlearningspeed，manyinputloadstrength．parametersandpoordynamicperformancemakeitnotWiththerapiddevelopmentofcomputertechnologyeasytoachieveinreality．Tosolvethisproblem，theandartificialintelligence，peopleintegratemathemati-advantagesofPIDcontroll

8、erandneuralnetworkwerecalm