基于PID神经网络的非线性动态系统控制  

Nonlinear Dynamic System Control with PID Neural Networks

作　　者：曹海云[1] 李守巨[1] 刘迎曦[1] 

机构地区：[1]大连理工大学工业装备结构分析国家重点实验室,辽宁大连116023

出　　处：《控制工程》2007年第B05期38-40,共3页Control Engineering of China

基　　金：国家自然科学基金(10472025)

摘　　要：基于PID神经网络的控制器可以完成变量的单输入-单输出非线性系统的任务。该控制器采用BP(误差反向传播)算法来修正连接权重值,通过在线训练和学习,使目标函数到达最优值。充分利用了BP神经网络算法逼近任意连续有界非线性函数的能力,显示了神经网络在解决非线性系统方面的潜能。为了达到控制的目的,和其他非线性建模技术相比较,PID神经网络有几个明确的优点和它独特的用法相一致。仿真结果表明,在时非线性动态系统控制时,基于PID神经网络的控制系统具有很强的灵活和高效性,能取得良好的控制效果。An neural network-based controller is designed and analyzed for a class of single-input nonlinear system. In order to optimize the object function through training on-line and learning, neural network＇s weights are adjusted by BP （Error Back Propagation）algorithm. The controller makes the best of BP neural network algorithm＇s ability of approaching all continuous, nonlinear functions, which shows the potential of neural network in solving the nonlinear system. Compared with other nonlinear modeling techniques for control purposes, it has several specific advantages that make it suited to particular applications. The simulation results show that the proposed control system with PID neural network is flexible and efficient, and it can fetch the favorable control results.

关 键 词：非线性系统控制 PID神经网络 非线性最优化 BP算法 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]