Research on fault prediction method of power electronic circuits based on particle swarm optimization non-homogenous grey model

Aiming at the low accuracy of current methods for power electronic circuit fault prediction, this paper propose a fault prediction technique of power electronic circuits. The approach is based on characteristic parameter data and particle swarm optimization non-homogenous grey model (PSO-NGM). Taking the Buck-Boost converter circuit as an example, the fault prediction of power electronic circuits is achieved. Firstly, it is an optimal choice to select the output voltage as monitoring signal, and then extracte the average voltage and ripple voltage of history and the current time as characteristic parameters. Additionally, PSO-NGM algorithm is used to predict the future characteristic parameters. Finally, fault prediction of power electronic circuits is achieved by comparing the predictive characteristic parameters with the threshold value. The experimental results indicate that the PSO-NGM algorithm is especially accurate in predicting the average voltage and ripple voltage with a error less than 0.3% and 1.3% respectively, which is ideal for fault prediction. The new method can trace the characteristic parameters´ trend and can be effectively applied in fault prediction of power electronic circuits.