一种复杂模糊系统生成方法

生成模糊系统传统方法的工作量往往随输入变量数的增长而爆炸性也增加，用于抽取模糊规则的神经网络的规模迅速地增加且能量的极小值点也迅速地增多。针对这一问题，本文发展了一种新的模糊系统生成方法，将复杂系统的模糊输入，输出关系分解成简单的模糊输入，输出关系叠加，采用了一种新的网络优化的方法－基于浮点编码的遗传算法来生成该系统。

A New Approach to Generate Fuzzy System

In traditional method, the amount of work of extracting and adjusting membership functions and rules expands startlingly with the increasing of the number of input variables. The scale of the neural network used to extract fuzzy rules becomes too large for training, and its number of local minimums will also increase rapidly. To solve this problem, a new method is developed, in which the complicated MISO system can be obtained from the combination of several simple SIMO systems;A new network optimization method, float coding based genetic algorithm(FGA), is used to generate the membership functions and rules of the complicated system. Because the decomposing complicated problem into combination of simple ones and the employment of FGA, this method can be used to generate MISO fuzzy systems.