A Comparative Study of Parameter Identification Techniques for State-of-Charge Estimation Using Extended Kalman Filter

Abstract

The State of Charge (SOC) estimation of lithium-ion batteries (LiBs) is a crucial function for developing an accurate battery management system for electric vehicles (EVs). The Extended Kalman Filter (EKF) combined with a second-order Thevenin model is commonly used in this field due to their advantage in building a balanced solution in terms of accuracy and computational complexity. An accurate equivalent circuit model requires precise identification of the battery model's parameters. This paper compares three techniques, including curve fitting, particle swarm optimization, and genetic algorithms to study the impact of the battery model accuracy on SOC estimation using EKF. The results show that particle swarm optimization outperforms the other techniques, achieving a root mean square error (RMSE) of 0.3017% and mean absolute error (MAE) of 0.1898%.