A Threshold Extend Kalman Filter Algorithm for State of Charge Estimation of Lithium-Ion Batteries in Electric Vehicles

Abstract

Model-based algorithms are widely employed in state of charge estimation. However, due to the complex operation environment and rich noise in the measured battery voltage and current, it is difficult to estimate terminal voltage accurately based on the battery model, resulting in great estimation deviation of state of charge (SOC). To address the issue, a threshold extended Kalman filter (T-EKF) algorithm based on the influence mechanism analysis of SOC estimation is proposed. In the proposed algorithm, the threshold for SOC, which is set by the residual non-Gauss voltage noise between the measured terminal voltage and the estimated terminal voltage, is employed to improve the robustness of EKF. Then, an experimental platform is built and two different EV operation conditions are selected to verify the algorithm on a Li-ion battery. Combining the comparison among extend Kalman filter (EKF), adaptive extend Kalman filter (AEKF), and the proposed T-EKF, some important results are obtained for the proposed algorithm. 1) The maximum SOC error of the proposed T-EKF are about 1.3% and 2.9% under urban bus operation condition (UBOC) and urban dynamometer driving schedule (UDDS), respectively; 2) the calculation cost of T-EKF is higher than EKF, but is the same as the AEKF; 3) the capacity to correct the initial SOC is maintained in the T-EKF and the oscillation on the estimated SOC curve can be effectively weakened.