An improved particle swarm optimization-adaptive square root Unscented Kalman filter algorithm for accurate state of charge estimation of lithium-ion batteries

With the rapid growth of electric vehicles and large-scale energy storage systems, accurate and stable estimation of the state of charge (SOC) of lithium-ion batteries has become a core challenge for battery management systems (BMS). This study proposes an Improved Particle Swarm Optimization-Adaptive Square Root Unscented Kalman Filter (IPSO-ASRUKF) algorithm for accurate and stable state-of-charge estimation. The algorithm integrates IPSO with an adaptive SRUKF strategy to enhance global convergence, noise-covariance adaptability, and overall estimation robustness. Its novelty lies in a collaborative “offline optimization–online adaptation” framework: IPSO employs logarithmic–opposition–elite initialization and nonlinear dynamic learning factors to improve convergence efficiency, while the adaptive SRUKF adjusts process and measurement noise covariance using singular value decomposition and a sliding-window mechanism. Experiments under three working conditions, based on a representative data selected from ten repeated tests, demonstrate RMSE values of 0.22%, 1.03%, and 1.61% for the respective conditions, significantly outperforming the conventional SRUKF, ASRUKF, and PSO-ASRUKF algorithms. These results confirm that the proposed strategy provides high accuracy, robustness, and real-time adaptability, offering reliable technical support for SOC estimation under complex operating conditions and contributing to the safe and efficient operation of electric vehicles and energy storage systems.