Early Prediction Method for Remaining Useful Life of Retired Batteries in Second-life Applications

Lithium-ion battery, which degrades to 80% of its rated capacity, will no longer be used in new energy vehicles. For better echelon utilization of these retired batteries, it is necessary to accurately predict their remaining useful life (RUL). Previous studies have achieved high accuracy in RUL prediction depending on a large amount of data or complex models, which put forward high requirements on data acquisition and computing resources. This paper aims to use early degradation data and simple models for RUL prediction with physical and statistical interpretation. Sliding window method is used to obtain more data information, while various matrices including multi-dimensional feature matrix, combination feature matrix, and fusion feature matrix are constructed. After the corresponding function processing and transformation processing, some of these matrices have better performance on RUL prediction than the capacity matrix. The combination matrix is the most effective in early prediction for RUL of retired batteries.