Modeling and Forecasting Battery Degradation using Scientific Machine Learning for Sustainability

The transition towards sustainable energy systems is dependent on the performance of lithium-ion batteries, particularly in electric vehicles (EV). As battery degradation impacts the lifespan and overall sustainability of these vehicles, predicting degradation is vital for optimizing battery usage and reducing environmental impact. This paper presents a Scientific Machine Learning (SciML) framework using Universal Differential Equations (UDEs) to model and predict the state of health (SoH) of lithium-ion batteries over time. By capturing degradation phenomena, this approach allows better battery management and prolonged battery life, as well as supporting sustainable consumption and production patterns, aligning with UN’s 12th Sustainable Development Goal. The results display the UDE’s effectiveness in forecasting battery degradation, with a mean squared error of 9.90, showing its ability to reduce environmental impact by lowering battery waste and the need for frequent replacements.