The Study Of Fuzzy Chaotic Neural Network

The human brain is a highly intelligent complex system; it can deal with kinds of complex, imprecise and blur information, it also has chaotic recollection capability. From the information disposal point of view, to construct an intelligent computer, which can think as bright as human brain, is the goal of Artificial Intelligent all the while. Fuzzy logic, chaotic theorem and artificial neural network emerge as the times require. They all focus on imitate one aspect of human brain's functions or structures, and acquire certain achievements separately. While, the study of confuse fuzzy logic, chaotic theorem and artificial neural network together is just a start.The existing fuzzy neural networks, which are often static models, are not good at simulate nonlinear systems, especially chaotic systems. Therefore, the method of introduce chaotic mechanism to fuzzy chaotic neural network as to form a fuzzy chaotic neural network is a novel idea and direction. In this thesis, the combination of fuzzy, chaos and neural network is done on the basis of forward research. New conceive is also put forward in the end. The main works and results are as follows:1. Introduce the investigation development of chaotic neural network and fuzzy neural network by now. List relative theorems and analyze techniques of dynamic system subsequently. Different Lyapunov exponent calculate measures are present according to the form of dynamic systems. Less number method is emphasized when treat with chaotic systems.2. Analyze the dynamic behavior of time-delay neural network. Discuss the Hopf bifurcation and chaotic phenomenon of time-delay neuron system. Throughout simulation and computation of LLE, the existence of chaos is proved.3. Introduce the model of T-S fuzzy neural network and recurrent neural network, give their parameters learning method. Use the two neural networks to simulate regular nonlinear system and chaotic system separately. Simulation results show that T-S fuzzy chaotic neural network is prone to trap in local minimum point. And the recurrent fuzzy neural network owns limited approaching capacity to chaotic systems.4. Introduce chaotic mechanism to fuzzy neural network. Validate the performance of fuzzy chaotic neural network based on chaotic BP learning method.