Enhancing Lithium-Ion Battery State-of-Charge Estimation Across Battery Types via Unsupervised Domain Adaptation

Abstract

Accurate estimation of the state-of-charge (SOC) in lithium-ion batteries (LIBs) is paramount for the safe operation of battery management systems. Despite the effectiveness of existing SOC estimation methods, their generalization across different battery chemistries and operating conditions remains challenging. Current data-driven approaches necessitate extensive data collection for each battery chemistry and operating condition, leading to a costly and time-consuming process. Hence, there is a critical need to enhance the generalization and adaptability of SOC estimators. In this paper, we propose a novel SOC estimation method based on Regression-based Unsupervised Domain Adaptation. We evaluate the performance of this method in cross-battery and cross-temperature SOC estimation scenarios. Additionally, we conduct a comparative analysis with a widely-used classification-based unsupervised domain adaptation approach. Our findings demonstrate the superiority of the regression-based unsupervised domain adaptation method in achieving accurate SOC estimation for batteries.