一类基于RBF神经网络的动态系统在线自适应辨识方法

AN ADAPTIVE APPROACH TO ON-LINE IDENTIFICATION OF DYNAMIC SYSTEM BASED ON RBF NEURAL NETWORKS

研究了基于神经网络的动态系统在线自适应辨识模型的基本结构 ,依据 RBF(Radial Basis Func-tion)网络线性输出的特点 ,给出了辨识模型参数的在线自适应校正的方法 ,并进行了仿真实验 ,结果表明 。

In this paper, a neural network based on line adaptive identification model of dynamic system is developed. In order to improve the performance of model, an on line adaptive correcting method of model parameters is proposed based on RBF network's linear outputs. A great amount of simulations in this investigation demonstrates that this identification model and on line adaptive correcting method are of generalization and applicability.