SOC estimation of Li-ion batteries using OCV-SOC curve estimation and adaptive EKF

Abstract

This thesis proposed a State of Charge (SOC) estimation method of lithium-ion batteries to increase a accuracy of SOC estimates. The proposed method consists of an open circuit voltage (OCV)-SOC curve estimation method and an adaptive extended Kalman filter method that use an equivalent circuit model (ECM). Using this method, model’s terminal voltage error that was caused by inaccurate OCV-SOC curve was decreased and affection of mismatched ECM parameters under large current changes was filtered. Therefore, the maximum absolute error of SOC estimates was reduced to < 1.4% and root-mean-square error decreased under 0.406%. The algorithm was developed in MATLAB and was verified with experimental measured dataset of dynamic current profile test. By using proposed method, battery-powered applications such as electric vehicle and drone can obtain highly accurate SOC values even if the load current changes dynamically.