An Aging-Aware SOC Estimation Method for Lithium-Ion Batteries using XGBoost Algorithm

Abstract

An accurate state-of-charge (SOC) estimation for a lithium-ion battery is highly dependent on the knowledge of aging, which is usually costly or not available through online measurements. In this paper, novel aging-aware features which can simultaneously characterize battery aging and SOC are extracted from the discharging process. Then, the extreme gradient boosting (XGBoost) algorithm combined a stage division is applied to acquire the nonlinear relationship model between the proposed features and the battery SOC through the offline training. The proposed method does not require the initial SOC value, which implies that the SOC can be estimated by the trained model from any operating states of a battery. Moreover, a random sampling test to simulate the online real-time SOC estimation verifies that the proposed method is effective and potential to be applied in the battery management system.