Precise state-of-charge estimation for LIBs: a cutting-edge nonlinear model approach with enhanced robustness and reliability

Precise state-of-charge (SoC) prediction is essential for optimising the performance, safety, and longevity of lithium-ion batteries (LIBs) in battery management systems (BMS). However, traditional prediction tactics, including Kalman filters and sliding mode observers (SMOs), struggle with sensor noise, model uncertainties, and external disturbances, leading to inaccuracies in real-world applications. This study proposes a nonlinear battery framework integrated with a Luenberger observer enhanced by H-infinity (H∞) optimisation to boost SoC prediction accuracy and robustness. The H∞ framework effectively mitigates disturbances, while sensor fault prediction enhances reliability under varying operational conditions. The recommended tactic is computationally efficient and suitable for real-time SoC prediction. Empirical outcomes validate the superior accuracy and stability of the recommended approach, achieving prediction errors that are up to 3.8% lower than those of conventional SMOs. The findings demonstrate potential for next-generation BMS applications, particularly in electric vehicles (EVs) and energy storage systems. Future work will focus on adaptive parameter prediction techniques to boost performance under real-world battery ageing conditions.