Comparison between two model-based algorithms for Li-ion battery SOC estimation in electric vehicles

• Published in: Simulation modelling practice and theory Vol. 34; pp. 1 - 11
• Main Authors: Hu, Xiaosong, Sun, Fengchun, Zou, Yuan
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.05.2013
• Subjects: Algorithms; Battery modeling; Electric batteries; Electric vehicles; Errors; Extended Kalman filter; Lithium-ion batteries; Lithium-ion battery; Noise; Robustness; State estimation; Lithium-ion battery; Electric vehicles; Battery modeling; State estimation
• ISSN: 1569-190X; 1878-1462
• DOI: 10.1016/j.simpat.2013.01.001

► A nonlinear battery model is identified for Li-ion battery state estimation. ► Two SOC-estimation filters using the established model are compared. ► The novel robust filter has a smaller error bound than does the standard filter. ► The novel robust filter has a better robustness against system noise statistics. Accurate battery State of Charge (SOC) estimation is of great significance for safe and efficient energy utilization for electric vehicles. This paper presents a comparison between a novel robust extended Kalman filter (REKF) and a standard extended Kalman filter (EKF) for Li-ion battery SOC indication. The REKF-based method is formulated to explicitly compensate for the battery modeling uncertainty and linearization error often involved in EKF, as well as to provide robustness against the battery system noise to some extent. Evaluation results indicate that both filters have a good average performance, given appropriate noise covariances, owing to a small average modeling error. However, in contrast, the REKF-based SOC estimation method possesses slightly smaller root-mean-square (RMS) error. In the worst case, the robustness characteristics of the REKF result in an obviously smaller error bound (around by 1%). Additionally, the REKF-based approach shows superior robustness against the noise statistics, leading to a better tolerance to inappropriate tuning of the process and measurement noise covariances.