递归T-S模糊模型的神经网络  

Neural Network Based on Recurrent T-S Fuzzy Model

作　　者：宋春宁[1] 刘少东[1] 

机构地区：[1]广西大学电气工程学院,南宁530004

出　　处：《化工自动化及仪表》2013年第5期578-581,共4页Control and Instruments in Chemical Industry

基　　金：国家自然科学基金资助项目(60924002)

摘　　要：在常规T-S模糊神经网络的基础上加入动态递归元件,提出了递归T-S模糊模型的神经网络。在系统辨识中采用无监督聚类算法和动态反向传播算法训练该递归神经网络的参数,给出了该递归网络的逼近性证明。辨识效果与常规T-S模糊模型作比较,说明递归T-S模糊模型的神经网络在非线性系统辨识中表现出更好的性能。The dynamic recursive elements were added to the general T-S fuzzy neural network to propose a recurrent T-S fuzzy neural network.In the system identification,the unsupervised clustering algorithm and dynamic back-propagation algorithm were applied to the parameter training of this recurrent neural network and the approximation of the fuzzy neural network was proved.Comparing the identification results of the two fuzzy neural networks shows that the recurrent T-S fuzzy neural network can perform well in nonlinear system identification.

关 键 词：递归神经网络 T-S模糊模型 非线性系统辨识建摸 模糊基函数 无监督聚类算法 动态BP算法 

分 类 号：TH865[机械工程—仪器科学与技术]