A transferable long-term lithium-ion battery aging trajectory prediction model considering internal resistance and capacity regeneration phenomenon

Accurately predicting the remaining useful life (RUL) of lithium -ion batteries (LiBs) is crucial for improving battery management system design and ensuring device safety. However, achieving accurate long-term predictions of aging trajectories is challenging due to error accumulation in multi -step ahead forecasts. This study shows that considering future internal resistance (R), which is related to the aging process, and the capacity regeneration phenomenon (CRP) that occurs during aging can help reduce error accumulation. Specifically, we propose a hybrid method that incorporates future R and CRP to predict the aging trajectories and RULs of LiBs. Experiment results demonstrate: (1) for the same charging/discharging policies and battery types, the proposed method can accurately predict the aging trajectory and RUL using only the first 20 cycles' data (approximately 5% of the complete data); (2) for different charging/discharging policies and battery types, with transfer learning, the proposed method can predict the aging trajectory and RUL using the first 40 cycles' data. These results demonstrate that the proposed model is both accurate in long-term prediction and robust for estimating the aging trajectory and RUL across various datasets.