State of charge estimation for lithium-ion batteries via a diffusion-enhanced equivalent circuit model and an improved extended Kalman filter

State of Charge (SOC) estimation for lithium-ion batteries plays a crucial role in predicting range and managing energy in electric vehicles. However, challenges remain under complex conditions such as wide temperature variations, dynamic driving cycles, and large current fluctuations. To achieve high-accuracy and robust SOC estimation, we propose an enhanced equivalent circuit model (ECM) that integrates both solid-phase and liquid-phase diffusion electrochemical mechanisms. A systematic comparison evaluates the impact of different OCV acquisition methods on estimation performance. The methods include Low-Current testing, Hybrid Pulse Power Characterization testing (HPPC), and the Galvanostatic Intermittent Titration Technique. To address significant noise in real-world conditions, a sliding mode observer-enhanced extended Kalman filter (SMOEKF) is developed to improve system robustness. After identifying model parameters using a genetic algorithm, the voltage prediction errors of different models under HPPC conditions were compared, along with the SOC estimation accuracy under UDDS and US06 driving cycles. The results indicate that the OCV curve extracted from the rest phases of HPPC testing yields the most accurate estimation performance. The proposed enhanced ECM significantly outperforms traditional ECMs in both voltage prediction and SOC estimation. Moreover, the SMOEKF algorithm exhibits superior accuracy and robustness compared to the EKF and the adaptive EKF under both noise-free and noise-contaminated conditions. A minimum root-mean-square error of 0.17 % is achieved under the UDDS cycle and 0.07 % under the US06 cycle in the noise-free condition.