Improved square root cubature Kalman filter for state of charge estimation with state vector outliers

In this paper, a square root cubature Kalman filter based multi-innovation least square (SRCKF-MILS) algorithm is proposed for state of charge (SOC) estimation. First, a second-order resistor-capacitor (RC) equivalent circuit model is constructed to approximate the dynamic performance of the battery. Then, the multi-innovation least square and the square root cubature Kalman filter are used to interactively estimate the battery model parameters and SOC. Since the state vector contains outliers, the strong tracking filtering (STF) theory is introduced to adaptively adjust the gain matrix. The STF theory combined with SRCKF-MILS algorithm can improve the robustness of the algorithm. Simulation examples show the effectiveness of the proposed algorithms.