Comparative study of Unscented and Extended Kalman Filtering methods for State-of-Charge Estimation of Lithium-Ion Battery in EVs

Battery Management Systems (BMS) are one of the most important systems that manage a large number of battery cells while ensuring safe and reliable operation. Prediction of state-of-charge of the battery should be much accurate to ensure vehicle run safely and reliably. Complex chemical reactions inside the cell determine the nonlinear relationship between cells' Open Circuit Voltage(OCV) and State-Of-Charge(SoC). SoC also gets affected by temperature, charging-discharging hence it is difficult to predict. Therefore, the paper establishes a 3-RC precise model of Lithium-ion battery and proposes the Unscented Kalman Filtering(UKF) method for SoC estimation. Comparison of estimated and actual SoC is done using Simulink® as a simulation platform. The Root Mean Square Error(RMSE) and Mean Absolute Error(MAE) of UKF and the extended Kalman filter at different ambient temperatures are compared. It is found that UKF shows superior performance than EKF in all aspects.