Application Of BP Neural Network Based On Genetic Algorithm To The Design Of Digital Filter

The technology of filter has been eighty years' history, and has developed quite perfect signal technology. At present, because of some problems of factual system, diversified optimized accounts are applied on optimized design of classical filters, and become emphasis of research and acquire some fruit. In china, the research of GA(Genetic Algorithm) optimizing neural network and applying them to design filter acquire some fruit. GA and NN have all-right research foreground in optimized design of digital filter because of some excellence of the combination of them. The combination of them can improve filter's technology and some problems of designing of classical filters.In this thesis, the advantage and disadvantage of BP neural network and GA (Genetic Algorithm) are analyzed, then novel algorithm of BP neural network based on Genetic Algorithm is proposed. In this novel algorithm, their ability of local and global search for optimum solutions are used and GA is used to optimize the learning of BP neural network's connection weights. Then , this novel algorithm is applied in the design of digital filter and validated.FIR Digital Filter and IIR Digital Filter are widely used in a variety of signal processing problems. Actually, their design is a optimum design of function's approach. In this thesis, Several LP/HP FIR and IIR filters are designed based on the proposed novel algorithm, and experimental simulations are performed to justify the effectiveness of the novel algorithm. Simulation results demonstrate that as compared with corresponding results yielded from other exiting design algorithms, the filters design by this novel algorithm has less shake in passed band and stoped bassed.