摘要： 文章基于单体模糊神经网络(MFNN)对多重模糊推理的Mamdani方法进行了推广，得到的广义方法(简称G-Mamdani法)克服了原有方法的若干不足。文章采用了求解模糊关系方程的方法来确定网络的权值，依此新方法，实现了一个模糊推理机，其推理效果较好。这一方法为模糊专家系统和模糊控制系统等提供了一种新的有力的推理工具。

关键词: 模糊推理, 单体模糊神经网络, 假言推理, 模糊关系方程

Abstract: This paper generalizes Mamdanis method of multidimensional fuzzy inference based on monolithic fuzzy neural network(MFNN). The generalized Mamdani's method (G-Mamdani method for short) overcomes a few serious shortcoming of conventional Mamdani's method. The weights of MFNN corresponding to the new method are adjusted by means of solution to fuzzy relation equations. The paper implements a fuzzy inference engine with G-Mamdani method. The new method proposed here is a powerful inference tool for fuzzy expert systems and fuzzy control systems etc.

中图分类号: 

• TP183