Insight into the whole from the part: Redefined state of health for lithium-ion batteries based on optimal charging fragment search

Supplemented with accurate state of health to evaluate the battery degradation level is an urgent and hot issue in the battery safety field. However, conventional state definitions and mainstream estimation methods are resistant to reliable implementation in complex real-world applications. In this work, we explore the possibility of redefining the state of health by using directly measurable charging data and accordingly validate its superior performance in identifying degradation. Over 51,600 cyclic samples from 63 cells with multiple charging protocols underpinned this investigation. Firstly, dedicated to alleviating the disparities between incomplete charging cycles in practice, this paper proposes the optimal charging fragment search method for feature screening based on the sliding window and data processing technologies. Subsequently, the distinct period feature parameters are fused to generate a redefined state of health. The validation results indicate that the redefined state of health effectively imitates the degradation trajectory of the original version, as evidenced by the correlation coefficients higher than 0.9. Thereafter, no modeling effort is involved in the proposed method, which has tremendous benefits in computational efficiency and robustness. The research findings highlight the promise of using directly measurable charging data to redefine the key state of batteries.