A strong robust state-of-charge estimation method based on the gas-liquid dynamics model

Model-based strategies for estimating the state-of-charge (SOC) of Li-ion batteries are essential in real-time applications, such as electric vehicles and large-scale energy storage. However, based on existing models, developing estimation methods with strong robustness to initial and cumulative errors, high SOC estimation accuracy, and adaptability to sparse data remains challenging. Herein, the modeling principles of the gas-liquid dynamics model are systematically clarified, and a SOC estimation method based on this model and a dual extended Kalman filter with a watchdog function is proposed. The proposed method is comprehensively compared with general extended Kalman filter and dual extended Kalman filter methods under five working conditions. The results indicate that all three methods based on the gas-liquid dynamics model have good estimation accuracy, with a maximum SOC error of 0.016 under correct initial conditions. But the proposed method has significant advantages in robustness to large initial errors, cumulative errors, and sparse data. This study provides new insights into efficient online SOC estimation.