Adaptive OCV Hysteresis Modeling for LiFePO4 SOC Estimation via Unscented Kalman Filter

Accurate state of charge (SOC) estimation for <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\text{LiFePO}_{4}$</tex> batteries is challenged by complex open-circuit voltage (OCV) hysteresis, particularly under dynamic conditions. This paper presents an adaptive dynamic hysteresis OCV model by focusing on the comprehensive experimental characterization and modeling of the hysteresis factor <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\delta$</tex>. The results reveal the dependence of the hysteresis factor <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\delta$</tex> on the SOC at current reversal points, enabling a more adaptive and accurate OCV representation. This refined OCV model, integrated with a secondorder equivalent circuit model and the unscented Kalman filter (UKF), achieves superior SOC estimation accuracy. Validation under the federal urban driving schedule (FUDS) confirms the model's capability to capture hysteresis dynamics. The maximum absolute SOC estimation error remains below 3% and RMSE values are consistently under <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{2. 5 \%}$</tex>, even under different initial SOC conditions. This paper provides a robust solution for improved SOC monitoring in demanding applications.