Wavelet Neural Network Based Transformer Fault Diagnosis Using Particle Swarm Optimization

The wavelet network (WN) which inherited learning ability of artificial neural network and localization characteristics of wavelet transform is the novel and efficient model of nonlinear signal processing developed in last years, and has good convergence and robustness. In this paper, the basic theory of wavelet networks is introduced. In order to improve efficiency and accuracy of fault diagnosis, we use the particle swarm optimization algorithm to optimize the initial connection weights and thresholds of wavelet neural network. The result shows that the wavelet neural network based on the particle swarm optimization is good in the iteration number, function approximation error and network performance, and the application in transformer fault diagnosis of the particle swarm wavelet neural network confirms the effectiveness of the method.