基于模糊神经网络的模糊辨识方法及其应用研究

利用模糊神经网络（FNN）的学习能力从控制操作的现场数据中获取模糊规则，并自动调节隶属函数，把建模的过程转化为FNN网络结构多数的生成与学习问题。用于一个非线性过程的模糊模型参数辨识问题，取得了满意的结果。

Research on a Method of Fuzzy Identification Based on Fuzzy Neural Network and Its Applications

Fuzzy model is indispensable in the design of fuzzy control system. But it is difficult to get an applicable fuzzy model, as the choice of structure and the tuning of parameters of a fuzzy model often depend on the human expert's experience. Fuzzy neural network (FNN ) is a new tool for getting fuzzy rules from operation data. In this paper, we present a fuzzy identification method based on a FNN as shown in Fig.1. This method converts the process of modeling to the learning problem of FNN. Here, the structure and parameters of the fuzzy model embed in the weight parameters of FNN. So the FNN can identify the fuzzy rules and modify the membership functions automatically through training with operation - data set.This paper presents the structure of the FNN and a hybrid learning aigorithm. Also a simulation example of fuzzy model identification problem about a reaction process of chemical production is given. It suggests that this method has merits of simple structure, and good modeling accuracy(Fig. 2).