Optimizing state of charge estimation using thevenin models and extended Kalman filter

Precise battery modeling plays a critical role in enhancing State of Charge (SoC) estimation and optimizing Battery Management Systems (BMS) for electric vehicles and energy storage systems. This study presents a comparative evaluation of three Thevenin equivalent circuit models, first-order (1RC), second-order (2RC), and third-order (3RC) with a focus on parameter optimization and estimation accuracy. The models are calibrated using discharge data from the Hybrid Pulse Power Characterization (HPPC) test, obtained from the CALCE dataset. Optimization techniques are applied to minimize voltage estimation errors, and model performance is assessed in terms of dynamic behavior replication, voltage accuracy, and computational demands. Results show that higher-order Thevenin models improve accuracy in capturing battery dynamics, although they introduce greater computational complexity. The 3RC model offers the best balance, achieving the lowest Mean Squared Error (MSE) and providing a more accurate representation of transient voltage responses. The Extended Kalman Filter (EKF) is integrated with each model using MATLAB/Simulink to enhance SoC estimation under real-time conditions such as dynamic load profiles. These findings highlight a trade-off between modeling accuracy and computational cost, with the 3RC model emerging as the most effective for applications requiring high-precision SoC estimation. The results support the use of optimized Thevenin models combined with EKF as a robust solution for advanced BMS design.