基于改进型RBF神经网络辨识的PID控制  

PID Control Based on Ameliorative RBF Neural Network Identification

作　　者：李绍铭[1] 刘寅虎[1] 

机构地区：[1]安徽工业大学电气信息学院,马鞍山243002

出　　处：《自动化与仪表》2006年第5期40-43,57,共5页Automation & Instrumentation

基　　金：安徽省教育厅自然科学基金资助项目(2006KJ032B)

摘　　要：针对工业控制领域复杂非线性时变系统,提出了基于改进型RBF神经网络的PID参数在线自整定方法。采用改进型RBF神经网络辨识器在线辨识系统模型,自动调整PID控制器参数,实现系统的智能控制。仿真结果表明,与常规RBF神经网络PID控制方法相比,该方法具有控制精度高、响应速度快的优点,并且具备较强的自适应性和鲁棒性。To complex systems which are of characteristics of nonlinearity and time-variation in the industrial control fields,this paper presents a serf-adaptive PID control method based on improved RBF neural network ,which identifies system model online by means of neural network identifier,adjusts automatically parameters of PID controller and achieves intelligence control of system. The simulation result indicates that the system,compared with PID control method based on the conventional RBF neural network, possesses the advantages of high precision, high response speed, great adaptability and robustness.

关 键 词：径向基函数 改进型RBF神经网络 PID控制 最近邻聚类算法 在线自整定 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]