The Functional Wavelet Neural Network Based on Genetic Algorithm
Ye Zhou · Jisuanji fangzhen · 2010
There are some disadvantages in traditional wavelet neural networks,such as falling into local minimum point easily.To avoid that disadvan-tage,networks are optimized by genetic algorithm (GA).A type of Genetic Algorithm based on multiple varibles coded with real and binary number is proposed in this paper.In this algorithm,real and binary numbers are presented simultaneously in a chromosome,which integrate the advantages of both.And then this genetic algorithm is put in the optimization of the functional wavelet neural network.So the more strengthened wavelet network is obtained.Simulation results show that WNN with hybrid genetic algorithm has a comparatively simple structure,and that it can both meet the precision request and enhance the generalization ability.The availability and superiority are testified through the result of the simulational experiment.