基于一种改进BP神经网络的一类非线性系统辨识方法

Identification of a Class of Nonlinear Systems Using Neural Networks

对一类非线性系统的辨识方法进行了讨论。辨识模型为一个改进的神经网络,该神经网络包括:线性网络和BP神经网络。对线性网络的模型选择和多层前馈网络的关系做了细致的研究, 并分别进行了仿真。仿真结果表明,该系统辨识方法对于一类非线性系统具有很好的辨识作用。

In this paper,we present a method of identification for a class of nonlinear system. The identification model is a modified neural network, which is composed of two parts: one is a linear neural network, and the other is a multi-layered feed-forward neural network. A delicate research about the relationship of the selection of model of the linear network part and the multi-layered feed-forward network part and simulation are presented. It shows that this method is effective for a class of nonlinear systems.