论文部分内容阅读
一种基于Ｔａｋａｇｉ－Ｓｕｇｅｎｏ模型的模糊神经网络由前件网络和后件网络两部分组成。前件网络用来匹配模糊规则的前件，它相当于每条规则的适用度。后件网络用来实现模糊规则的后件。总的输出为各模糊规则后件的加权和，加权系数为各条规则的适用度。所提出的模糊神经网络具有局部逼近功能，且具有神经网络和模糊逻辑两者的优点。它既可以容易地表示模糊和定性的知识，又具有较好的学习能力。给出了调整规则后件参数及前件隶属度函数参数的学习算法，举例说明了它的逼近性能。
A kind of fuzzy neural network based on Takagi-Sugeno model consists of two parts: the pre-network and the post-network. The antecedent network is used to match the antecedent of fuzzy rules, which is equivalent to the applicability of each rule. The post-part network is used to implement the fuzzy rules of the aftermath. The total output is the weighted sum of the pieces of the fuzzy rules, and the weighting coefficient is the fitness of each rule. The proposed fuzzy neural network has the local approximation function and has the advantages of both neural network and fuzzy logic. It can not only express fuzzy and qualitative knowledge, but also has good learning ability. The learning algorithm for adjusting the parameters of the rules after the rules and the parameters of the membership function is given, and the approximation performance of the algorithm is illustrated.