基于改进模糊神经网络的PID参数自整定算法  

Self-Tuning of PID Parameter Algorithm Based on Improved Fuzzy Neural Networks

作　　者：欧国徽 刘春波 潘丰 

机构地区：[1]江南大学轻工过程先进控制教育部重点实验室,江苏无锡214122

出　　处：《江南大学学报（自然科学版）》2011年第2期145-149,共5页Joural of Jiangnan University (Natural Science Edition)　

基　　金：国家863计划项目(2009AA05Z203)

摘　　要：针对传统的PID控制算法参数整定困难,控制效果并不理想,将神经网络算法、模糊控制算法结合在一起,形成了模糊神经网络PID参数自整定算法,并且对模糊神经网络进行改进,将神经网络输入的状态变量进行模糊化和归一化处理,采用BP神经网络自整定PID控制器的参数,根据RBF神经网络得到受控对象的Jacobian信息。仿真结果表明,基于模糊神经网络的PID自整定控制效果较好,具有一定的应用前景。For traditional PID parameter tuning difficulties,control effect is not ideal.This paper addresses this issue.Fuzzy control algorithms and neural network algorithms are combined to form an FNN self-tuning of PID parameter algorithm.Besides,some improvement based on FNN normalization and obfuscation is made to deal with the input state variables of Neural Networks.BP Neural Network is used to adaptively adjusts PID parameters.Using RBF Neural Network get plant＇s Jacobian information.The experimental results show that the control results of self-tuning of PID parameter based on FNN has been enhanced in some practical application.

关 键 词：PID控制算法 神经网络 模糊控制 自整定 

分 类 号：TP312[自动化与计算机技术—计算机软件与理论]