Fault Diagnosis of Rectifier Circuit Based on Improved PSO Algorithm

Aiming at the problem of multiple fault types and low diagnostic accuracy of power electronic circuits, a method of optimizing probabilistic neural network based on wavelet packet decomposition and particle swarm optimization was proposed. Firstly, the simulation model of three-phase bridge rectifier circuit was established, and the fault voltage signal was decomposed and reconstructed by three-layer wavelet packet decomposition technique, and the characteristic values were extracted and normalized. Then PSO was used to optimize the probabilistic neural network to find the appropriate smoothing factor, and the data was trained and diagnosed. Finally, compared with the unoptimized probabilistic neural network, the simulation results show that the method is superior to the unoptimized probabilistic neural network on both training effect and diagnosis accuracy.