Autonomous End-to-End SOH Prediction Services for Battery Systems via Temporal-Contrastive Representation Learning

The article introduces TC-SOH, an autonomous, end-to-end service for state of health (SOH) prediction in lithium-ion battery systems, utilizing a temporal-contrastive representation learning approach. It features a modular architecture that eliminates manual feature engineering, enhancing transparency through various diagnostic techniques. TC-SOH demonstrates significant performance improvements over existing models, achieving a 1.91 times reduction in mean absolute percentage error (MAPE) and a 2.13 times reduction in root mean square error (RMSE) across four public datasets, making it a valuable tool for practitioners in battery management systems.