A unified disturbance rejection design framework for robust reduced-coupling state estimation in lithium-ion batteries

Model-based state-of-charge (SOC) estimation is widely used for lithium-ion batteries (LIBs), where high estimation accuracy in conventional frameworks is achieved by designing additional estimators to compensate for uncertainties in system parameters, such as capacity, polarization voltage, and internal resistance. However, such multi-estimator framework introduce strong coupling and complex feedback paths, which may undermine the robustness and stability of the estimation in practical application. To overcome this limitation, this paper proposes a novel robust estimation framework based on decoupled state modeling, where all states are estimated using a unified estimator, thus eliminating the complexity of the traditional multi-estimator network. Within the framework, an adaptive step size regulation strategy based on the golden section search (GSS) is introduced to eliminate the need for manual step tuning and improve the accuracy of model updating. In addition, to enhance the robustness of the unified estimator, an active disturbance rejection framework (ADRF) is incorporated to achieve real time disturbance compensation. Experimental results demonstrate that the proposed method achieves high estimation accuracy under diverse operating conditions. Compared with conventional model-based approaches, it exhibits better robustness and stable estimation performance.