This study proposes an Extended Kalman Filter (EKF)-based estimation technique that considers battery aging characteristics to more accurately estimate the State of Charge (SOC) of lithium-ion batteries. With the increasing adoption of electric vehicl...
This study proposes an Extended Kalman Filter (EKF)-based estimation technique that considers battery aging characteristics to more accurately estimate the State of Charge (SOC) of lithium-ion batteries. With the increasing adoption of electric vehicles and the increasing research on the management, reuse, and recycling of used batteries, the importance of technologies for accurately assessing and utilizing the state of used batteries is growing. To determine the recyclability and reuse of these batteries, research is being conducted on reliable SOC estimation techniques that can accurately diagnose the current state of used batteries. However, existing SOC estimation techniques fail to adequately account for aging characteristics, such as increased internal resistance, decreased effective capacity, and voltage response changes, resulting in limitations in their applicability in reuse environments.
In this study, experiments were conducted on Samsung SDI's INR21700-40T cylindrical lithium-ion batteries. This study developed an ECM that reflects aging characteristics and validated an Extended Kalman Filter (EKF)-based SOC estimation technique based on this ECM. First, battery voltage and current data were collected through charge and discharge experiments, and based on this data, ECM parameters for each aging stage were identified. Furthermore, aging indicators such as capacity loss and internal resistance increase were quantitatively evaluated according to the KS C IEC 62660-3 standard, and changes in battery characteristics due to battery deterioration were reflected in the model.
Next, an EKF-based SOC estimation algorithm was developed based on the developed aging-incorporated ECM. The SOC estimation performance under dynamic driving conditions was verified using the Urban Dynamometer Driving Schedule (UDDS) driving cycle in a MATLAB/Simulink environment. This demonstrated that the proposed EKF-based SOC estimation technique maintained stable estimation performance even under aging battery conditions.
A comparison of EKF-based SOC estimation performance before and after aging was performed revealed that the proposed aging-incorporated EKF outperformed the conventional EKF even under conditions of battery deterioration, maintaining a stable SOC estimation error (RMSE) of approximately 1%. In particular, by updating the RC parameters and the SOC–OCV relationship according to the aging status, we were able to effectively ensure convergence and consistency of SOC estimation despite changes in voltage response over time.
Furthermore, through module-level experiments, we measured the voltage of each cell based on real-time current input and applied the EKF. This demonstrates that the proposed EKF-based estimation technique can perform stable real-time SOC estimation even in environments similar to actual vehicles.
In summary, this study experimentally demonstrates the effectiveness of an EKF-based SOC estimation technique that systematically reflects characteristic changes due to battery aging. Furthermore, it provides a technological foundation for its expansion into SOC diagnosis algorithms for reused batteries.