Prognostics for lithium-ion battery operating under different depth of discharge using hybrid method

Lithium-ion battery is a widely used power source in electronic system and its degradation has a very serious impact on the system performance and reliability. In this paper, a novel hybrid prognostics method is developed for the battery operating under different depth of discharge (DOD). Unlike other prognostics schemes, in the proposed design, unscented Kalman filter (UKF) and a fitting method are utilized to generate a prognostics result and a series of raw error data, the error data is adopted by complete ensemble empirical mode decomposition (CEEMD) method and relevance vector machine (RVM) to correct the UKF-based prognostics result. Note that the previously reported prognostics schemes seldom consider the effect of DOD on battery degradation nor the prognostics error. Comparative experiments based on proposed scheme, UKF and RVM are conducted to verify the effectiveness and performance of the proposed hybrid method.