Adaptive dead-zone dual sliding mode observer for reliable electrochemical model-based SOC estimation

Accurate state-of-charge (SOC) estimation is essential for the safe and efficient operation of lithium-ion batteries in electric vehicles and stationary energy storage systems. Electrochemical models enable high-fidelity SOC estimation, yet their practical deployment is challenged by nonlinear voltage dynamics, parameter drift, and the risk of divergence in joint state–parameter estimation. This paper proposes an adaptive dead-zone dual sliding-mode observer (ADZ-DSMO) based on an improved extended single-particle (IESP) model. In the proposed framework, a state SMO estimates SOC-related states, while a parameter SMO adapts dominant model parameters online. A Lyapunov-guided, online-computable adaptive dead-zone is developed to gate parameter updates, thereby suppressing error-propagation-induced divergence. Experimental validation on a Panasonic NCR18650PF cell under 1C, 3C, Urban Dynamometer Driving Schedule (UDDS), and New European Driving Cycle (NEDC) profiles shows that the proposed method achieves lower SOC error than the conventional dual sliding-mode observer (DSMO) and the IESP-based dual extended Kalman filter (DEKF). Under correct initialization, the SOC mean absolute error (MAE) remains below 0.25% under 1C, UDDS, and NEDC, and reaches 1.13% under the more demanding 3C condition. Under a 30% initial SOC bias, the proposed method maintains MAE of 0.13–0.25% under 1C, UDDS, and NEDC, and achieves 1.46% MAE under 3C, with faster convergence than the DEKF in all tested cases. Moreover, selective parameter updating reduces the average computation time per sampling step compared with that of the always-updating DSMO, indicating improved real-time feasibility. Robustness is further evaluated under temperature variation and model-generated aging conditions.