作　　者： ; ;

机构地区： 上海理工大学光电学院

出　　处： 《计算机仿真》 2010年第12期195-199,共5页

摘　　要： 研究工业过程控制,非线性系统难以建立其精确的数学模型,常规PID算法难以实现其控制。对此,将神经网络算法和PID算法结合起来,为提高性能和响应速度,设计了针对非线性系统的自适应PID控制器,PID控制器采用单神经元来实现自适应调整。分别采用了三种学习算法来实现单神经元PID控制器参数的调整,即基于二次型性能指标的学习算法、基于BP神经网络辨识的学习算法和基于RBF神经网络辨识的学习算法。三种算法的仿真结果表明,都实现了PID控制器的自适应调整的目的,实现对非线性系统的有效控制,证明三种方案的可行性。 It is difficult to establish the nonlinearity systems＇ mathematic model,and the general PID method is difficult to control the nonlinearity systems.This paper presents an adaptive PID controller which uses single-neuron to realize adaptive adjustment for nonlinearity systems.Three study methods for identifying the nonlinearity controlled object are introduced,which are the study method based on quadratic performance index,the study method based on BP neural network identify,and the study method based on RBF neural netwok identify.The simulations results indicate that all the three methods can ensure the PID controller＇s adaption,and control the objects effectly,and prove that the design is effective.At last,the paper analyzes the virtues and flaws of the three method by comparing the simulations results.

关 键 词： 单神经元 控制器 二次型性能指标 神经网络 径向基函数 辨识

领　　域： [自动化与计算机技术] [自动化与计算机技术]