A Similarity-Based Approach for Diagnosing the Aging of Lithium-Ion Batteries in Second Life Combining Time Series and Machine Learning

Modelling aging in the second life of lithium-ion batteries (LiBs) is challenging due to the complexity of degradation mechanisms that lead to capacity loss and internal resistance increase, as well as uncertainty and variability in the operational and environmental conditions to which the batteries are exposed. In this work, we propose a similarity-based approach for diagnosing the aging of LiBs in their second life, which combines time series analysis and machine learning to help identify trends and patterns in the aging process. This approach overcomes the intrinsic nonlinearity nature of the LiB aging trajectory in the second life while adapting to varying operational and environmental conditions. Knees or inflection points defining the first, second, and non-usable lives of the batteries are also identified, offering insights into degradation mechanisms and thus supporting thermal management and optimal user-pattern tasks to extend the LiBs’ lifetime.