Enhanced moving-step unscented transformed-dual extended Kalman filter for accurate SOC estimation of lithium-ion batteries considering temperature uncertainties

Accurate estimation of the state of charge (SOC) of power lithium-ion batteries is required by battery management systems (BMSs) in electric vehicles. Variations in ambient temperature have a significant impact on battery performance while in operation. This work proposes a moving-step unscented transformed-dual extended Kalman filter (MUT-DEKF) based on an adaptive forgetting factor recursive least-squares method and a second-order Thevenin equivalent circuit model (SOT-ECM) with fast and slow characterization modes to accurately estimate the SOC using lithium cobalt oxide and lithium nickel manganese cobalt oxide battery data at different temperatures and working conditions. Experimental validation of the SOT-ECM conducted under the Beijing bus dynamic stress test working condition shows that the model tracks the actual voltage of the battery with an absolute maximum error of 0.1508 V, 0.0711 V, and 0.0464 V at 0 degrees C, 25 degrees C, and 50 degrees C, respectively, demonstrating the characteristic nonlinear behavior of battery systems. Additionally, the SOC estimation results reveal that the MUT-DEKF exhibits strong noise feedback, weight correction, and robustness. The proposed MUT-DEKF accurately estimates the SOC by utilizing two different battery data collected at different temperatures and operating conditions. The proposed method demonstrates low mean absolute error and root mean square error values of 0.1604% and 0.1922% respectively, indicating high accuracy, faster estimation, and robustness. This makes it an efficient approach for SOC estimation in BMS operations.