Remaining useful life prediction of lithium-ion batteries by considering trend filtering segmentation under fuzzy information granulation

A remaining useful life (RUL) interval prediction method for lithium-ion batteries (LiBs) based on fuzzy information granulation is proposed in this paper to meet the different requirements under different operating conditions. The segmentation strategy of considering the time series trend is developed for fuzzy granulation to overcome its shortcomings that cannot distinguish the time series containing different degradation trends. In order to predict the RUL interval of LiBs, the health indicator (HI) with high indirect correlation with capacity is extracted by analyzing the charge and discharge characteristics of LiBs, and the extracted HI is fuzzy granulated into two subsequences of upper and lower bounds after applying the proposed trend segmentation strategy. On this basis, the two subsequences are noise-reduced by the variational mode decomposition (VMD), and then modeled and predicted by using a gated recurrent unit (GRU). According to the two prediction sequences above and below, the prediction results can be constructed to realize the RUL interval prediction of LiBs. Comparison experiments based on public battery datasets show the superiority of the proposed prediction method for LiBs.