SOC Estimation of Power Lithium Battery Based on RGC and Multi-innovation UKF Joint Algorithm

Abstract

A second-order RC equivalent circuit model was established to accurately estimate the state of charge (SOC) of power lithium battery. The model parameters were identified online using the recursive gradient correction (RGC) algorithm, enhancing the real-time performance of parameter identification. Building on the unscented Kalman filter (UKF) algorithm, a multi-innovation unscented Kalman filter (MIUKF) algorithm was proposed by incorporating the multi-innovation identification theory. This approach overcomes the impact of ignoring historical errors in traditional Kalman filter algorithms on estimation accuracy, thereby accelerating the algorithm’s convergence to the true value and improving its accuracy and stability. The algorithm was validated under various operating conditions. The results indicate that, compared to the UKF algorithm, the MIUKF algorithm exhibits superior performance in estimation accuracy and anti-interference capability, enabling precise SOC estimation for lithium batteries in vehicles.