Long-term Remaining Useful Life Prediction of Lithium Battery based on Grey Theory

Abstract

With growing popularity of lithium batteries in power systems, remaining useful life prediction of batteries is gaining importance. A long-term remaining useful life prediction approach for lithium-ion batteries based on grey theory is presented in this paper. Firstly, an optimized discrete grey model is used to track the battery overall degradation trend. Then, Markov chain is implemented to compensate the error of grey model, which is mainly sourced from fluctuations of curve caused by capacity regeneration. Finally, relevance vector regression is utilized to provide the final remaining useful life value and its probability. The universality and accuracy of this approach are validated with battery data under different discharge rates and temperatures.