A Prognostic method for DC-DC converters under variable operating conditions

Prognosis of DC-DC power converters is necessary in embedded and safety critical applications to prevent further damages. However, most of the prognostic methods of power converters are focus on the critical components of the converters. Furthermore, the methods seldom consider the effect of changes in operating conditions (e.g. power supply and load). In order to address these problems, an innovative system-level fault characteristic parameter (FCP) represents the degradation status of the entire converter is extracted, and a prognostic method of DC-DC converters based on the degradation trend prediction of the FCP is proposed. Firstly, the effect of component-level degradation on the overall performance of the DC-DC converters is studied. Then, a performance parameter of DC-DC converters which is sensitive to the degradation of all critical components is chosen, and a least squares support vector machine (LSSVM) model is used to convert the performance parameter to the FCP under predetermined normal condition to eliminate the influence of changes in operating conditions. Finally, the trend prediction of the FCP is performed based on Gaussian process regression (GPR) to realize the prognosis of DC-DC converters. A Boost converter is taken as an illustrative example. Results show the feasibility and effectiveness of the proposed method.