State Space Models Based Prognostic Methods for Remaining Useful Life Prediction of Rechargeable Batteries

Abstract

Rechargeable batteries are widely used in many electronic products and systems to provide power sources. Because of the influence of charge/discharge cycling and some significant battery degradation factors, such as discharge rate, temperature, depth of charge, etc., on battery health condition, battery degrades over time. In this chapter, several state space models based prognostic methods are proposed to predict battery remaining useful life. Firstly, a particle filtering based state space model for battery remaining useful life prediction at a constant discharge rate is introduced. Secondly, to improve particle filtering and its application to battery prognostics, spherical cubature Kalman filtering is introduced to provide an importance function for the use of particle filtering at a constant discharge rate. Thirdly, to extend battery prognostics at a constant discharge rate to battery prognostics at different discharge rates, a more general battery degradation model is presented. Based on the developed model, a battery prognostic method at different discharge rates is designed. Some discussions are made at last.