Study of Fault Classification Based on PSO Wavelet Neural Network

In this paper, a novel studying-training algorithm of wavelet neural-network based on particle swarm optimization (PSO) was presented. Then it was compared with the traditional gradient descent algorithm by the fault classification experiment. The simulation result get a conclusion that the wavelet neural-network trained by PSO not only reduces the iterations, but also get the better convergence precision. It is further indicated that PSO algorithm is fit to train the wavelet neural-network and the optimized network has good signal classification ability.