Dynamic adaptive fuzzy neural-network identification and its application

In this paper, we propose a dynamic fuzzy neural-network structure, i.e., there are two classical fuzzy-neural network structures in dynamic fuzzy neural-network structure. In the practical identification processing, the function of the two classical fuzzy-neural networks is often changed. At the same time, one classical fuzzy-neural network can be used to estimate the model, and another classical fuzzy-neural network is used to learn. At the appropriate time, the role of the two classical fuzzy-neural networks is changed. The fuzzy-neural network that was used to estimate the model starts to learn, and the fuzzy-neural network that was learning is used to estimate the model, how to change is decided by a switching region. By using the method, the parameter adjustment of an adaptive fuzzy identification model and optimal parameters of the system can be obtained.