模糊逻辑推理的神经网络方法及其应用

The Neural Networks Method of Fuzzy Logical Inference and It's Application

应用可能性度量理论提供了一种模糊逻辑推理的神经网络实现方法．在该方法中，模糊规则前件（ａｎｔｅｃｅｎｄｅｎｔ）和后件（ｃｏｎｓｅｑｕｅｎｔ）的语言变量被表示为神经网络中的权重，采用Ｂａｃｋ－Ｐｒｏｐａｇａｔｉｏｎ学习算法来学习相关语言变量的隶属度水平。介绍了控制器的构造和推理方法以及在移动机器人路径跟踪中的应用。

A neural networks implementation method of fuzzy logical inference based on the theroy of possibility mersure is provided in the paper. The linguistic variable associated with the antecedent and consequent of fuzzy rulues are represented as weights in this neural network structure, and the structure allows learning of the membership grades of the associated linguistic variables by means of BackPropagation algorithm. The design, inference methods of this controller and it's application on the road following of mobile robot are also discussed.