Survival Analysis with Machine Learning for Predicting Li-ion Battery Remaining Useful Life

TL;DR

This paper addresses Li-ion battery RUL prediction by reframing it as a survival analysis problem and introducing a hybrid framework that converts voltage time-series into time-to-failure data using path signatures. It trains five survival models (Cox, CoxTime, CoxPH, DeepHit, MTLR) to estimate failure-free probabilities over time, evaluated on Toyota and NASA datasets with metrics including T-AUC, C-Index, and IBS. The results show that Cox-based models generally deliver the most reliable discrimination and calibration, while MTLR offers a strong balance between ranking and probability estimation, demonstrating the framework's practical value for predictive maintenance in EVs and industrial settings. Overall, the integration of transient voltage features through path signatures with diverse survival models advances robust, uncertainty-aware RUL predictions for complex battery degradation dynamics.

Abstract

Battery degradation significantly impacts the reliability and efficiency of energy storage systems, particularly in electric vehicles and industrial applications. Predicting the remaining useful life (RUL) of lithium-ion batteries is crucial for optimizing maintenance schedules, reducing costs, and improving safety. Traditional RUL prediction methods often struggle with nonlinear degradation patterns and uncertainty quantification. To address these challenges, we propose a hybrid survival analysis framework integrating survival data reconstruction, survival model learning, and survival probability estimation. Our approach transforms battery voltage time series into time-to-failure data using path signatures. The multiple Cox-based survival models and machine-learning-based methods, such as DeepHit and MTLR, are learned to predict battery failure-free probabilities over time. Experiments conducted on the Toyota battery and NASA battery datasets demonstrate the effectiveness of our approach, achieving high time-dependent AUC and concordance index (C-Index) while maintaining a low integrated Brier score.

Figures (10)

• Figure 1: Toyota battery voltage change during charging (top) and discharging (bottom)
• Figure 2: Time-to-failure status distribution across Toyota cell groups.
• Figure 3: NASA battery voltage change during discharging
• Figure 4: Time-to-failure status distribution across NASA battery groups.
• Figure 5: Comparison of model performance on NASA data