Battery Degradation Modelling and Prediction with Combination of Machine Learning and Semi-empirical Methods

Battery energy storage systems (BESS) are being widely deployed as part of the energy transition. Accurate battery degradation modelling and prediction play an important role in BESS investment and revenue, planning and sizing, operational monitoring, and warranty check-ups. Complex operational behaviors and system variability make the battery degradation modelling and prediction more challenging. In this paper, we propose a novel methodology of combining a machine learning model with semi-empirical equations to have a robust state of health (SOH) estimation. Together with cycle decomposition analysis, this model can be deployed for real-time degradation analysis of BESS.