基于自回归动态神经网络的逆模型辨识及在线控制

INVERSE MODELING AND ON LINE CONTROL BY USING RECURRENT NEURAL NETWORKS

研究了自回归动态神经网络及其学习算法，提出应用于动态逆模型辨识的结构，并与ＰＩＤ控制相结合形成了非线性动态对象的在线自适应控制系统．仿真结果表明此方案简单可行，克服了静态网络的一些局限性．

A recurrent dynamic neural network and its learning algorithm are studied. The structure of inverse modeling by using the neural network is presented. Furthermore, on line adaptive control system for nonlinear dynamic plant, which combines the inverse model with PID, is formed. Simulation results show that this strategy is available and simple. It overcomes some disadvantages of the static neural networks when the strategy is applied for identification and control.