BP神经网络PID控制的研究

沈阳理工大学学士学位论文I摘要目前，由于PID具有结构简单，可通过调节比例积分和微分取得基本满意的控制性能，广泛应用在电厂的各种控制过程中。电厂主汽温的被控对象是一个大惯性大迟延非线性且对象变化的系统。常规汽温控制系统为串级PID控制或导前微分控制，当机组稳定运行时，一般能将主汽温控制在允许的范围内。但当运行工况发生较大变化时，却很难保证控制品质。因此本文研究BP神经网络的PID控制，利用神经网络的自学习、非线性和不依赖模型等特性实现PID参数的在线自整定，充分利用PID和神经网络的优点。本处用一个多层前向神经网络，采用反向传播算法依据控制要求实时输出Kp、Ki、Kd，依次作为PID控制器的实时参数，代替传统PID参数靠经验的人工整定和工程整定，以达到对大迟延主气温系统的良好控制。对这样一个系统在MATLAB平台上进行仿真研究，仿真结果表明基于BP神经网络的自整定PID控制具有良好的自适应能力和自学习能力，对大迟延和变对象的系统可取得良好的控制效果。关键词：主汽温;PID;BP神经网络;MATLAB仿真沈阳理工大学学士学位论文IIAbstractAtpresent,becausePIDhasasimplestructureandcanbeadjustedproportionalintegralanddifferentialtosatisfactorycontrolperformance，,itiswidelyusedinpowerplantsofvariouscontrolprocess.ThesystemofPowerplantmainsteamtemperatureisanlargeinertia、bigtime-delayedandnonlineardynamicsystem.ConventionalsteamtemperaturecontrolsystemadoptedcascadePIDcontrolorthedifferentialcontrolofleadbefore,whentheunitisstable,generalwillallowthesteamtemperaturecontrolintherange,butwhenoperatingconditionschangedgreatly,itisdifficulttoensurethequalityofcontrol.ThisarticlestudiesPIDcontrolbasedBPneuralnetwork.Usingsuchcharacteristicsofneuralnetworkself-learning,nonlinearanddon'trelyonmodelrealizePIDparametersonlineauto-tuning.ItcanmakefulluseoftheadvantagesofPIDandneuralnetwork.Here,weuseamultilayerfeedforwardneuralnetworkusingbackpropagationalgorithmandbasedoncontrolrequirements.Thisnetcanreal-timeoutputKp,Ki,KdasthePIDcontrollerparameters,insteadingofthetraditionalPIDparametersdeterminedbyexperience.Soitcanobtaingoodcontrolperformance.Forsuchasystem,wecansimulateinMATLABsimulationplatform.ThesimulationresultsshowthatthePIDcontrolbasedBPneuralnetworkhasgoodadaptiveabilityandself-learningability.Forthesystemoflargedelayandfree-modelcanobtaingoodcontroleffect.KEYWORDS:Mainsteamtemperature;PID;BPneuralnetwork;MATLABsimulation沈阳理工大学学士学位论文III目录摘要.....................................................................................................................................IABSTRACT..............................................................................................................................II第一章绪论..............................................................................................................................11.1选题背景和意义..............................................................................................................11.2国内外研究现状..............................................................................................................11.3立论依据..........................................................................................................................61.4本文所做的主要工作......................................................................................................61.5本论文的章节安排..........................................................................................................7第二章神经网络原理和应用..................................................................................................72.1MP模型............................................................................................................................82.2神经网络的学习方式和学习规则..................................................................................92.2.1神经网络的学习方式..............................................................................................92.2.2神经网络的学习规则..............................................................................................92.3神经网络的特点及应用................................................................................................112.4BP神经网络..................................................................................................................112.4.1BP神经网络的结构............................................................................................112.4.2BP神经网络算法................................................................................................122.4.3BP神经网络的前向传播算法............................................................................122.4.4BP神经网络的反向传播计算............................................................................13第三章基于BP神经网络的PID控制...............................................................................163.1PID控制器的离散差分方程.......................................................................................163.2基于BP神经网络的PID整定原理............................................................................173.3基于BP网络的PID控制器控制的算法流程............................................................21第四章基于BP神经网络的PID控制在主汽温控制系统中的应用.................................234.1锅炉主汽温的特点......................................................................................................234.1.1主汽温的控制任务..............................................................................................234.1.2主汽温控制对象的动态特性..............................................................................23沈阳理工大学学士学位论文IV4.2主汽温的数学模型........................................................................................................254.2.1主汽温控制系统调节信号的选择..........................................................