A fault prediction method for closed-loop SEPIC converters under variable operating conditions

Abstract

Due to the critical applications of single-ended primary inductance converter (SEPIC) in various fields, it is necessary for the SEPIC converters to predict their performance trends. Closed-loop control instead of open-loop control have been adopted in SEPIC converters, and the fault prognostic methods used for open-loop converters sometimes are no longer suitable for closed-loop converters. Besides, the system-level fault feature parameters (FFP) is often influenced not only by the fault modes, but also by the variation of working conditions. To address these problems, an innovative system-level FFP represents the degradation status of the entire converter under variable operating conditions is extracted, and a prognostic method based on the degradation trend of the FFP is proposed. Firstly, the deterioration laws of some system-level parameters with the change of critical components under different operating conditions are studied. Then, a system-level performance parameter of closed-loop SEPIC converters which is sensitive to the degradation of all critical components and has regular trend is chosen, and the degradation performance parameter under rated operating condition is obtained as FFP by multivariate least-squares regression. Finally, the trend prediction of the FFP is performed based on Extreme learning machine (ELM) to realize the prognosis of closed-loop SEPIC converters. The experimental results show the feasibility and effectiveness of the proposed method.