Estimation of State of Charge (SOC) for Lithium Iron Phosphate Batteries Based on SHEKF Algorithm

In response to the issue of inaccurate state of charge (SOC) estimation of batteries due to changes on environmental temperature and noise under complex operating conditions, this study takes an Extended Kalman Filter (EKF) method combined with Adaptive Forgetting Recurrent Least Squares (AFFRLS) for online SOC estimation of batteries. The algorithm enhances robustness by adaptively updating the matrix of observation noise using Sage-Husa integrating into EKF algorithm, addressing the divergence issues traditionally encountered in SOC update on the LiFePO4 battery platform. Finally, the algorithm is validated through Dynamic Stress Test (DST) conditions at various temperature points, showing SOC estimation errors consistently within 3% when temperatures are above 10°C.