Li-Ion Battery SOC Estimation using Improved AEKF with Adaptive Window Width Adjustment

Precise estimation of the state of charge (SOC) in lithium-ion batteries plays a vital role in preventing overcharge and overdischarge, which helps maintain the safety and reliability of the battery. The existing adaptive extended Kalman filter (AEKF) algorithm updates the noise covariance matrix by calculating the voltage residuals within a fixed-width window. However, an inappropriate window width may compromise SOC estimation accuracy or even lead to the non-convergence of the algorithm. In this work, an improved extended Kalman filter (IAEKF) algorithm is proposed, which calculates the variance of voltage residuals within the current window to assess the fluctuations of the voltage residuals. Based on this assessment, the window width is fine-tuned by either increasing or decreasing its size. Finally, the noise covariance matrix is updated in realtime according to the adjusted window width. Statistical results demonstrate that, under various temperature conditions, the proposed algorithm effectively enhances SOC estimation accuracy even when the initial SOC value deviates by as much as 50 % from the actual value.