Enhancement of an Electric Vehicle’s State of Charge Estimation Using an Extended Kalman Filter

The state of charge (SOC) of a vehicle battery must be estimated by its battery management system (BMS). To achieve precise SOC estimation, the paper investigates model-based methods, namely the Thevenin model and the second-order RC model, in conjunction with the extended Kalman filter (EKF) approach. The Thevenin and 2RC models are described in detail, highlighting their significance in SOC estimation. The EKF approach is introduced as a valuable tool to refine the SOC predictions obtained from both models. Integrating these models and the EKF approach into a MATLAB Simulink program enables comprehensive evaluation. By comparing the empirical SOC method with the estimations from the Thevenin and 2RC models using the EKF approach, the study finds that the maximum SOC error for the Thevenin model is 0.793%, while for the 2RC model, it reduces to 0.414%. This research underscores the efficacy of the second-order RC model along with the EKF approach in achieving accurate SOC estimation for vehicle batteries, which holds significant implications for optimizing battery performance and longevity in electric vehicles.