基于聚类有效性神经网络的模糊规则提取方法

模糊控制以其自适应性、鲁棒性和易于实现等优点得到广泛应用．然而模糊控制规则的获取通常由专家根据经验给出，这就存在诸如规则不够客观，专家经验难以获取等问题．为此，本文给出一种其于聚类有效性神经网络的模糊规则提取的新方法．该方法采取了对训练样本预划分子集聚类，模糊语言量的自动确定，模糊隶属度函数自适应调整等策略，克服了以往规则提取法在训练样本不充分时，规则提取不足及规则数目难以确定等缺点，并结合神经网络技术使所提取的控制规则的质量得到提高，改善了模糊控制器的性能，最后，以倒车系统为例证明了该方法的有效性．

A Method of Extracting Fuzzy Rules Based on Neural Networks with Cluster Validity

Fuzzy control has been widely used due to its self-adaptability,robust and easy-to-be-realized.However fuzzy control rules are usually given by experts according to their experiences,so many problems are existed such as the rules are not objective the expert's experiences are difficult to get,and etc.A new method of extracting fuzzy rules based on neural networks with cluster validity is presented in th is paper.This method overcomes the disadvantages of the available mothods with which the rules extracted are not perfect and it is hard to determine the rule number when the training samples are not enough.Combined with neural network techniques,the quality of the control rules is largely advanced and the performance of the fuzzy controller is improved.Finally,the validity of this method is demonstrated with a truck controller system.