基于神经网络的运动控制系统算法研究  

Study on Algorithm for Motion Control System Based on Neural Network

作　　者：周治平[1] 夏娟 

机构地区：[1]华东理工大学,上海200237 [2]江南大学,江苏无锡214122

出　　处：《计算机工程与应用》2006年第3期205-208,211,共5页Computer Engineering and Applications

基　　金：教育部重点科技项目(编号:03085)

摘　　要：文章研究了神经网络控制算法在运动控制系统中的应用,以典型的运动控制系统为例,基于神经网络内模自适应控制方法设计速度环控制器。采用递归神经网络的ANNI用于充分逼近被控对象的动态模型,前向反馈神经网络ANNC用作控制器,在网络拓扑结构的设计及训练算法的选择上,进行探讨,提供了解决方法并获得了一些有意义的结论。实时仿真结果表明了基于神经网络的控制系统有很强的自适应能力和抗负载扰动能力,在系统的性能上优于传统的控制算法,为类似运动控制系统的设计与研究提供了新的方法。In recent years,there are some new hot topics in the field of motion control,more and more researchers become caring about how to use advanced intelligent control strategy to satisfy the requirement of control system.In this paper,adaptive neural network internal control algorithm is adopted,the controller includes two neural networks,one is ANNC,the other is ANNI.At the same time,how to design well topological structure and how to choice effective trained algorithm are discussed and some meaningful conclusions are acquired.The real-time simulation results show the control algorithm based on ANN has good properties in terms of robustness and tracking ability,and it is superior to the traditional strategy.The method offers a new throughway for designing and studying motion control system.

关 键 词：神经网络内模自适应 运动控制 实时仿真 控制器 网络拓扑结构 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置] TP183[自动化与计算机技术—控制科学与工程]