Quantifying Battery Degradation Patterns with Sparse Regression Models

Abstract

Understanding and predicting battery degradation is vital for extending the lifespan and improving the reliability of energy storage systems, especially in electric vehicles and renewable energy applications. Traditional modeling techniques often struggle with the high dimensionality and multicollinearity present in battery datasets. This paper introduces a sparse regression framework to model and quantify battery degradation patterns accurately. Sparse regression models, particularly Lasso and Elastic Net, are leveraged to handle high-dimensional features, enforce variable selection, and ensure model interpretability. The proposed approach is validated through extensive experiments using publicly available lithium-ion battery datasets. Key degradation indicators such as internal resistance, capacity fade, and state of health are predicted with high accuracy, offering critical insights into the long-term behavior of batteries. The results demonstrate that sparse models outperform conventional regression techniques in both predictive performance and computational efficiency, enabling scalable and explainable diagnostics for battery health monitoring.