Predicting the Remaining Useful Life of Lithium Ion Batteries based on the Support Vector Regression-particle Filter

Abstract

Lithium ion batteries (LIBs) are widely used in a variety of electronic equipment, and reliable operation of LIBs is vital for maintaining equipment functionality. Therefore, the development of methods for predicting the remaining useful life (RUL) of LIBs, which are designed to provide a warning prior to battery failure, is essential for ensuring normal operation of electronic equipment. This paper presents an approach for predicting the RUL of LIBs based on the support vector regression-particle filter (SVR-PF). First, an LIB capacity fade model is established for the analysis of parameters involved in capacity fade. Next these parameters are used to establish an RUL prediction model for LIBs to predict the RUL, and to update the probability density of the model during offline analysis. SVR-PF is applied, rather than the commonly employed standard particle filter, to improve prediction precision.