Comparative Study on State-of-Charge Estimation Techniques of Lithium-Ion Batteries for Hybrid Power Systems

In hybrid power systems, batteries undergo a charge-depleting (CD) mode caused by a continuous decrease in the state of charge (SOC) and a charge-sustaining (CS) mode that maintains the SOC within a specific range. These operating modes pose challenges for the accurate estimation of the SOC. This study aims to compare various equivalent circuit models (ECMs) and state estimation techniques based on the battery operation data of plug-in hybrid electric vehicles and explore an efficient accurate method considering CD-CS combined scenarios. To achieve the aforementioned objectives, the Thevenin and enhanced self-correcting (ESC) models are employed for ECMs and the extended Kalman filter (EKF) and adaptive extended Kalman filter (AEKF) are employed for state estimation techniques. The performances of the SOC estimation techniques are compared under various operating conditions. The results demonstrate that the ESC and AEKF exhibit the best estimation performance, thereby reducing uncertainty arising from the initial covariance settings. Furthermore, more improved performance is observed when an ECM is fitted based on CD-CS composite mode data compared to CD mode data onl