A Comparative Study of Test Profile-Based Parameter Identification for State-of-Charge Estimation Using Extended Kalman Filter

Abstract

This work investigates and compares four test profiles, including incremental OCV, federal urban driving schedule (FUDS), US06, and beijing driving schedule test (BJDST), for parameter identification of a second-order Thevenin model for lithium-ion batteries. The parameters are extracted with the particle swarm optimization (PSO) and validated through cross-testing using an extended Kalman filter (EKF) for state of charge (SOC) estimation. The root mean square error (RMSE) and mean absolute error (MAE) metrics are utilized for performance assessment, both for terminal voltage and SOC. The results demonstrate that the BJDST-based parameters yield the lowest terminal voltage RMSE (1.44%) and MAE (0.81%), while the FUDS-based parameters achieve the lowest SOC RMSE (0.27%) and MAE (0.20%). The study highlights the importance of dynamic profiles, such as FUDS and BJDST, to identify generalizable model parameters.