Robust joint estimation of state of charge and electrochemical parameters for Lithium-ion batteries under model uncertainty and measurement outliers

This paper presents a joint adaptive robust extended Kalman filter (JAREKF) for simultaneous estimation of the state of charge (SOC) and equivalent circuit model (ECM) parameters in lithium-ion batteries. Unlike conventional approaches that assume fixed or slowly varying ECM parameters, the proposed method embeds resistances and capacitances directly into the augmented state vector to account for parameter uncertainty. In this study, a dual-level robustness mechanism is introduced within two time scales. In the fast time scale, the voltage outliers are suppressed through innovation-based adaptive weighting. Then in the slow time scale, the updates are gated by estimation uncertainty to prevent model mismatch or aging-induced drift from degrading SOC accuracy. Experimental validation under dynamic load profiles, room-temperature operation, and artificial sensor faults demonstrates that the algorithm achieves SOC estimation errors within 1.32% using only noisy terminal voltage measurements. It also reliably tracks gradual variations in polarization resistance and capacitance in the ECM. The method is well suited for practical embedded battery management systems requiring long-term reliability, where adaptability to cell aging and robustness against sensor faults and model uncertainties are critical.