A stochastic, fuzzy, neural network for nonlinear dynamic systems

摘要

The stochastic, fuzzy, neural network (SFNN) is developed for the modeling of nonlinear dynamic systems. In this network, a non-singleton fuzzifier with Gaussian membership functions instead of the singleton fuzzifier in the fuzzy logic isintroduced. An online supervised parameter learning algorithm of the SFNN based on the membership functions is proposed for avoiding the local minimum phenomenon in the learning process of the established neural networks. An off-line algorithm for thestructure learning of the SFNN is developed for reduction of the computational time of the SFNN. This SFNN is used for the modeling of maneuvering target motion with high non-linearity as an example, and the multi-sensor information fusion is completedthrough the SFNN. This new network gives a good performance compared to the existing network, and it is also applicable to stochastic control and decision systems and the identification of chaos in nonlinear systems.