rentialgeometrymethod,whichisdiscussestheproblemsin"geometrydomain".Themainproblemistheacquisitionoftheinversemodelintheapplications.Sincenon-linearsystemisacomplexsystem,anddesiredstrictanalyticalinverseisveryobtain,evenimpossible.Theengineeringapplicationofinversesystemcontroldoesn’tmeettheexpectations.Asneuralnetworkhasnon-linearapproximateability,especiallyfornonlinearcomplexitysystefnormalinverterandinductionmotorisacomplicatednonlinearsystem,traditionalPIDcontrolstrategycouldnotmeettherequirementforfurthercontrol.Therefore,howtoenhancecontrolperformanceofthissystemisveryurgent.Theneuralnetworkinversesystemisanovelcontrolmethodinrecentyears.Thebasicideaisthat:foragivensystem,aninversesystemoftheoriginalsystemiscreatedbyadynamicneuralnetwork,andthecombinationsystemofinverseandobjectistransformedintoakindofdecouplingstandardizedsystemwithlinearrelationship.Subsequently,alinearclose-loopregulatorcanbedesignedtoachievehighcontrolperformance.Theadvantageofthismethodiseasilytoberealizedinengineering.Thelinearizationanddecouplingcontrolofnormalnonlinearsystemcanrealizeusingthismethod.CombiningtheneuralnetworkinverseintoPLCcaneasilymakeuptheinsufficiencyofsolvingtheproblemsofnonlinearandcouplinginPLCcontrolsystem.Thiscombinationcanpromotetheapplicationofneuralnetworkintopracticetoachieveitfulleconomicandsocialbenefits.Inthispaper,firstlytheneuralnetworkinversesystemmethodisintroduced,andmathematicmodelofthevariablefrequencyspeed-regulatingsysteminvectorcontrolmodeispresented.Thenareversibleanalysisofthesystemisperformed,andthemethodsandstepsaregiveninconstructingNN-inversesystemwithPLCcontrolsystem.Finally,themethodisverifiedinexperiments,andcomparedwithtraditionalPIcontrolandNN-inversecontrol..NeuralNetworkInverseSystemControlMethodThebasicideaofinversecontrolmethodisthat:foragivensystem,anα-thintegralinversesystemoftheoriginalsystemiscreatedbyfeedbackmethod,andcombiningtheinversesystemwithoriginalsystem,akindofdecouplingstandardizedsystemwithlinearrelationshipisobtained,whichisnamedasapseudolinearsystemasshowninFig..Subsequently,alinearclose-loopregulatorwillbedesignedtoachievehighcontrolmathematicmodelofthevariableperformance.Inversesystemcontrolmethodwiththefeaturesofdirect,simpleandeasytounderstanddoesnotlikedifferentialgeometrymethod,whichisdiscussestheproblemsin"geometrydomain".Themainproblemistheacquisitionoftheinversemodelintheapplications.Sincenon-linearsystemisacomplexsystem,anddesiredstrictanalyticalinverseisveryobtain,evenimpossible.Theengineeringapplicationofinversesystemcontroldoesn’tmeettheexpectations.Asneuralnetworkhasnon-linearapproximateability,especiallyfornonlinearcomplexitysyste气隙磁场测量或其他计算机相关量来计算同步速度。在间接磁场定向控制,必须得到同步参考系同步转速,对此值进行积分得到转变参考系函数W中角度。重些方程式(-)使eqr,得到edrererqrri(-)再次令eqr,重写(-)方程式emqseqrrmLiiLL(-)将式(-)代入式(-),e可表示为eermdrdrrerreemrqsmqsLLLriLi(-)此表达式提供了所须同步转速以及转子磁链,它们由控制器指定。转矩控制是通过调节q轴定子电流实现。滑差计算需要转子磁链时间常数rT,而且很多情况下需要在线估计,因为它取决于温度和其他因素。d轴定子电流产生给定转子磁通,可以使用公式(-)计算,参考系转变中使用角度可通过下式计算:tetd(-).网络反馈系统实现步骤.输入与输出运行样本采集采样对网络反馈系统建立是极其重要。它不仅需要获得原系统动态数据,还需要获得了静态数据。参考信号应该包括原始系统所有工作范围,并确保近似。信号欲处理第一阶段是从每HZ到HZ中得到HZ,并得到开环响应。第二阶段是混乱信号输入,当每秒钟出现预处理信号时,随机信号输入,并得到闭环响应。基于这些输入,将得到组得到运行样本。.网络建设静态神经网络和动态神经网络完美组合将能构建一个反馈系统。静态神经网络结构是由个输入层神经元,个输出层神经元和个隐蔽层神经元组成。隐藏神经元激励函数是单调平滑双曲正切函数。输出层是由线性临界激励函数神经元组成。运行数据是这些速度开环,闭环相对应速度和设置参考速度。次运行之后,神经网络运行错误达到.。神经网络负荷和临界值被保存下来。并得到原系统反馈模型。.实验和结果.系统硬件硬件系统包括上层监督计算机安装,控制结构软件WinCC.,西门子S-PLC,变频器,异步电动机和光电编码器。选择S--DPPLC控制器,它有一个PROFIBUS-DP接口和一个MPI接口。高速采集模块是FM-。WinCC用MPI协议被CP贯穿到S-。这个逆变器类型是西门子MMV。西门子PLC能兼容美国协议。在这个系统上ACB模块被增加在逆变器上。.软件编程..通信介绍MPI(多点接口)是一种简单、便宜通讯策略,运用在运行慢,非大型数据转换场合。在WinCC与PLC之间数据转换不是很大,所以选择MPI协议。MMV变频器作为从动装置连接到PROFIBUS网络,并安装到CBPROFIBUS模块上。PPO或PPO数据类型可供选择。它允许控制信号直接发送到变频地址,或者使用STEPV.SFC/系统功能模块。OPC能有效提供完整数据和通信能力。不同类型服务器和客户机可以存取彼此数据来源。比较传统软件模式和硬件发展,设备生产商只需要培养一个操作员。这样可以缩短开发周期,节省人力资源,并简化了整个控制系统结构。矩阵实验室神经网络运行需要系统各种各样数据时候,这些数据不能从PLC或WinCC直接读取。所以OPC技术可以用来获得在WinCC和Exce之中所需数据。设置WinCC作为OPCDA服务器,一个OPC客户将被很好建立关于VBA。系统实时数据被WinCC读取并写到Excel上,然后Excel上数据被转换到矩阵实验室为在离线运行时获得原系统反馈系统。..控制程序通常用STEPV.标准模板库来对通讯,数据采集和控制算法进行编程,速度采样程序和存储程序被编程为有规律中断程序A,中断周期为毫秒。为了阻止程序A运行时间超过毫秒,减小程序运行周期和系统错误,控制步骤和神经网络算法被编程为主程序B。神经网络算法标准化对运行采样来说是必要以便加快信号收集速度,在最终运行之前输入和输出信号乘以一个放大系数。.实验结果当速度参照是秒每周期方波信号时,逆变器运行是矢量模式。结果表明,神经网络控制跟踪性能均优于传统常规PI控制。当速度参照保持恒定时,经过秒时间,负荷降低到没有负荷,经过秒时间,负荷增加到满负荷,所以在传统控制下速度响应曲线和网络反馈控制下速度响应曲线如下图所示。很明显,在稳定性能上,网络反馈控制负载扰动优于传统PI控制负载扰动。图-PI控制下速度响应图-网络反馈控制下速度响应.结论为了改善PLC变频调速系统控制性能,因而神经网络反馈系统被使用。并给出了一个变频调速系统数学模型,且其可逆转性得到了检验。反馈系统和原系统被组合并构建成伪线性系统,并设计了线性控制方法进行控制。通过实验,PLC神经网络反馈系统在工业应用中具有有效性和可行性到了验证。fnormalinverterandinductionmotorisacomplicatednonlinearsystem,traditionalPIDcontrolstrategycouldnotmeettherequirementforfurthercontrol.Therefore,howtoenhancecontrolperformanceofthissystemisveryurgent.Theneuralnetworkinversesystemisanovelcontrolmethodinrecentyears.Thebasicideaisthat:foragivensystem,aninversesystemoftheoriginalsystemiscreatedbyadynamicneuralnetwork,andthecombinationsystemofinversean毕业设计(外文翻译)英文题目RealizationofNeuralNetworkInverseSystemwithPLCinVariableFrequencySpeed-RegulatingSystem中文题目PLC变频调速网络反馈系统实现系(院)自动化系专业电气工程与自动化学生姓名学号指导教师职称讲师二〇一三年六月RealizationofNeuralNetworkInverseSystemwithPLCinVariableFrequencySpeed-RegulatingSystemThevariablefrequencyspeed-regulatingsystemwhichconsistsofaninductionmotorandageneralinverter,andcontrolledbyPLCiswidelyusedinindustrialfield.However,forthemultivariable,nonlinearandstronglycoupledinductionmotor,thecontrolperformanceisnotgoodenoughtomeettheneedsofspeed-regulating.Themathematicmodelofthevariablefrequencyspeed-regulatingsysteminvectorcontrolmodeispresentedanditsreversibilityhasbeenproved.Byconstructinganeuralnetworkinversesystemandcombiningitwiththevariablefrequencyspeed-regulatingsystem,apseudo-linearsystemiscompleted,andthenalinearclose-loopadjustorisdesignedtogethighperformance.UsingPLC,aneuralnetworkinversesystemcanberealizedinacturalsystem.Theresultsofexperimentshaveshownthattheperformancesofvariablefrequencyspeed-regulatingsystemcanbeimprovedgreatlyandthepracticabilityofneuralnetworkinversecontrolwastestified..IntroductionInrecentyears,withpowerelectronictechnology,microelectronictechnologyandmoderncontroltheoryinfiltratingintoACelectricdrivingsystem,invertershavebeenwidelyusedinspeed-regulatingofACmotor.Thevariablefrequencyspeed-regulatingsystemwhichconsistsofaninductionmotorandageneralinverterisusedtotaketheplaceofDCspeed-regulatingsystem.Becauseofterribleenvironmentandseveredisturbanceinindustrialfield,thechoiceofcontrollerisanimportantproblem.Neuralnetworkinversecontrolwasrealizedbyusingindustrialcontrolcomputerandseveraldataacquisitioncards.Theadvantagesofindustrialcontrolcomputerarehighcomputationspeed,greatmemorycapacityandgoodcompatibilitywithothersoftwareetc.Butindustrialcontrolcomputeralsohassomedisadvantagesinindustrialapplicationsuchasinstabilityandfallibilityandworsecommunicationability.PLCcontrolsystemisspecialdesignedforindustrialenvironmentapplication,anditsstabilityandreliabilityaregood.PLCcontrolsystemcanbeeasilyintegratedintofieldbuscontrolsystemwiththehighabilityofcommunicationconfiguration,soitiswildlyusedinrecentyears,anddeeplywelcomed.Sincethesystemcomposedofnormalinverterandinductionmotorisacomplicatednonlinearsystem,traditionalPIDcontrolstrategycouldnotmeettherequirementforfurthercontrol.Therefore,howtoenhancecontrolperformanceofthissystemisveryurgent.Theneuralnetworkinversesystemisanovelcontrolmethodinrecentyears.Thebasicideaisthat:foragivensystem,aninversesystemoftheoriginalsystemiscreatedbyadynamicneuralnetwork,andthecombinationsystemofinverseandobjectistransformedintoakindofdecouplingstandardizedsystemwithlinearrelationship.Subsequently,alinearclose-loopregulatorcanbedesignedtoachievehighcontrolperformance.Theadvantageofthismethodiseasilytoberealizedinengineering.Thelinearizationanddecouplingcontrolofnormalnonlinearsystemcanrealizeusingthismethod.CombiningtheneuralnetworkinverseintoPLCcaneasilymakeuptheinsufficiencyofsolvingtheproblemsofnonlinearandcouplinginPLCcontrolsystem.Thiscombinationcanpromotetheapplicationofneuralnetworkintopracticetoachieveitfulleconomicandsocialbenefits.Inthispaper,firstlytheneuralnetworkinversesystemmethodisintroduced,andmathematicmodelofthevariablefrequencyspeed-regulatingsysteminvectorcontrolmodeispresented.Thenareversibleanalysisofthesystemisperformed,andthemethodsandstepsaregiveninconstructingNN-inversesystemwithPLCcontrolsystem.Finally,themethodisverifiedinexperiments,andcomparedwithtraditionalPIcontrolandNN-inversecontrol..NeuralNetworkInverseSystemControlMethodThebasicideaofinversecontrolmethodisthat:foragivensystem,anα-thintegralinversesystemoftheoriginalsystemiscreatedbyfeedbackmethod,andcombiningtheinversesystemwithoriginalsystem,akindofdecouplingstandardizedsystemwithl 毕业设计(外文翻译)英文题目RealizationofNeuralNetworkInverseSystemwithPLCinVariableFrequencySpeed-RegulatingSystem中文题目PLC变频调速的网络反馈系统的实现系(院)自动化系专业电气工程与自动化学生姓名学号指导教师职称讲师二〇一三年六月1RealizationofNeuralNetworkInverseSystemwithPLCinVariableFrequencySpeed-RegulatingSystemThevariablefrequencyspeed-regulatingsystemwhichconsistsofaninductionmotorandageneralinverter,andcontrol