A Novel Approach for Real-Time Estimation of State of Charge in Li-Ion Battery Through Hybrid Methodology

Rechargeable batteries are essential components for modern energy systems and electric vehicles (EVs). Accurate estimation of State of Charge (SOC) plays a pivotal role in the reliable operation and efficiency of battery systems. While various methods have been developed to improve SOC estimation, there remains significant potential for further enhancement. This paper presents a hybrid SOC estimation technique specifically designed for EV battery management systems (BMS). The proposed method effectively mitigates the impact of cell deterioration, achieving high-precision SOC estimation. SOC serves as a critical parameter in BMS decision-making. This study integrates the Adaptive Extended Kalman Filter (AEKF) with a Li-ion cell model and the Coulomb Counting technique. Given the computational complexity inherent to AEKF and the susceptibility of the Coulomb Counting method to noise, their combination offers a novel approach characterized by improved accuracy and reduced complexity. The method was validated through extensive simulations in MATLAB-Simulink and experimental testing using a hardware test bench. The results were compared to those of the unscented Kalman filter-based SOC estimation, adaptive integral correction-based methods, and machine learning-based methods. The proposed adaptive strategy shows a 70% reduction in complexity compared to DEKF while achieving an SOC estimation accuracy of up to 1.02%.