基于模糊对向神经网络的非线性动态系统辨识器

Identification of Nonlinear Dynamic Systems Using Fuzzy CP Network

模糊对向神经网络 (FCP)在功能上同模糊逻辑系统的TS模型是等价的 ,它具有神经网络和模糊逻辑系统各自的优点 ,因而适宜作辨识模型 .利用FCP网络为辨识模型 ,设计了一种新的非线性动态系统辨识器 .辨识模型的参数调整采用无监督聚类法和误差回传算法相结合的混合学习方法进行的 ,同时文中还给出选取初始参数的方法 .仿真实验表明提出的辨识器是十分有效的 .

The fuzzy counterpropagation (FCP) network is functionally equivalent to the TS model of fuzzy logic systems, and has the characteristics of both neural networks and fuzzy logic systems. Therefore, it's applicable to use FCP network for \{identification\} model of nonlinear dynamic systems. Training the parameters of the model is to use a hybrid learning algorithm based on combining an unsupervised clustering method and the error back propagation algorithm. A new identifier according to the above is developed. The simulation results show the identifier is efficient.