Fuzzy neural networks, a method of fuzzy optimization design based on neural networks, is one of active branches in the intelligent control theory. Under the background of multimedia data communication which is widely applied in long-distance digital measure and control systems, the thesis discusses the application of compensative fuzzy neural network(CFNN) and fuzzy hopfield neural network(FHNN) in the telecommunication, and proves that is a practical and effective method. The combination system of the compensative fuzzy theory and neural networks introduces fuzzy nerve cells, which make networks can either adjusting input &output fuzzy subjection functions appropriately or optimizing the fuzzy inference dynamically by the help of compensative logic arithmetic. This system can be applied in the fusion of multi-sensor, image data compaction, and networks congestion control, which greatly enhanced the robust, stability and working speed of the networks system. In addition, the FuzzyHN-based algorithm, designed by pole configuration, can significantly improve the filter, which proves that it is a practical and effective method in image restoration.
|