基于动态递归网络PID自适应控制器设计  

The design of an adaptive neural PID controller based on Dynamic Recurrent Neural Network

作　　者：刘学军[1] 段慧达[1] 

机构地区：[1]北华大学电气信息学院,吉林132021

出　　处：《微计算机信息》2008年第7期-,共3页Control & Automation

基　　金：吉林省科技厅项目(200505263)

摘　　要：本文分析了改进的ELMAN网络的结构,并讨论了神经网络的学习算法,针对BP算法的缺陷,提出了用遗传算法修正网络权值的学习算法。本文不仅将采用遗传算法进行训练的改进ELMAN网络应用于电阻炉加热系统的建模,而且针对该系统的特点提出了一种带预测模型的神经网络PID自适应控制器,并最后将该控制器应用于电阻炉温度控制,取得了良好的控制效果。A modified Elman network and its algorithm are studied in this paper.To overcome the slow convergence of the BP algorithm,genetic algorithm(GA) is proposed,which can train the weight of the network..In addition,in this paper,an adaptive neural PID controller with the predictive model is proposedan adaptive neural PID controller based on Dynamic Recurrent Neural Network.At last we use this adaptive neural PID controller in the control of temperature.It can be seen that the effect of control is quite well,ach...

关 键 词：改进的ELMAN网络 遗传算法 系统辨识 PID自适应控制器 

分 类 号：TP273.2[自动化与计算机技术—检测技术与自动化装置]