Smart Prognostics for Li-Ion Battery Reuse: Machine Learning-based Remaining Useful Life Estimation in Second-Life Applications

This study presents the development of machine learning-based surrogate models for estimating the remaining useful life (RUL) of repurposed Li-ion batteries in second-life applications. After their initial deployment in demanding environments such as electric vehicles, these batteries are often retired once their capacity drops below a defined threshold (typically 80% of the nominal value). However, they still retain substantial usable capacity, making them viable for less intensive roles such as stationary applications. A key challenge in this context is the accurate prediction of battery RUL to ensure safe, efficient, and reliable operation over time. In this work, surrogate models are trained on publicly available battery degradation data to learn complex aging patterns and estimate the number of charge/discharge cycles remaining before the battery reaches its end of life. The results demonstrate that the proposed modeling approach achieves high prediction accuracy, thereby supporting improved control and lifecycle management of second-life battery systems.