Fault prediction method based on SVR of improved PSO

Fault prediction raises more and more concern because it can predict the fault to refrain from large calamity. As time pass by, system performance is frequently changed in engineering practice. Therefore, it is necessary to build a supporting mechanism to conduct dynamic fusion and set up a forecasting model to track and forecast system performance. This paper puts forward a new fault prediction method, named support vector regression (SVR) with improved particle swarm optimization (IPSO) algorithm. In solving time series and nonlinear regression problems, the support vector regression model has been applied proverbially. To gain the best global optimizer in SVR, it will employ the IPSO optimize the parameters. IPSO-SVR algorithm´s simulation results preview that it not only has a better prediction ability than traditional SVR, but also keeps a rapid convergence compared to the standard PSO.