System Identification Method Study Based On Dynamic Fuzzy Neural Networks

Fuzzy neural networks with the neural network self-learning,adaptive,parallel processing capacity and strong function approximation ability, but also integrates fuzzy logical rules extraction ability. Thus in the system identification, pattern recognition and control the areas to be ideal used. Then based on the dynamic fuzzy neural network's system identification is one of current research area hot spots. At present the image recognition and the nonlinear dynamic system identification is always the key and difficult point for people to study, but the fuzzy neural network served as the identification to provide for the identification domain newly,and the general effective identification tools.This article based on the dynamic fuzzy RBF network and the GD-FNN network, recognized these two aspects to the pattern recognition and the misalignment dynamic system to launch the research, its primary coverage was as follows:1,Improved RBF network dynamic fuzzy image identification methods: use the image on the selection identification input signal the feature vector substitution primitive image number, a fuzzy neural network number theory realize network identification. In the actual network in identification of optimized the learning algorithm, using MATLAB software platform in remote sensing image a simulation experiment was carried out to verify the fuzzy RBF networks in the excellent properties of image recognition. 2,Dynamic fuzzy neural network with extractive function which rapid fuzzy rules and the strong nonlinear dynamic recognition ability, attention by researchers. This article made the improvement to the GD-FNN network architecture to form the return generalized fuzzy neural network (GDRFNN), and has carried on the optimization to the learning algorithm. Separately using GDFNN network and GDRFNN network for the same nonlinear dynamic system network made the identification simulation. which has confirmed GDRFNN in the misalignment identification domain superiority.Due to the dynamic fuzzy neural network fused the fuzzy system and the neural network respective merit. when carries on the system identification, which can carry on the accurate pattern recognition and the nonlinear system identification by its the two's unique function, The its has broad prospects for development in the reality utilization .