A Comparison Study on the Selection of the Optimum Electrical Equivalent Circuit Model for Lithium-Ion Batteries and Parameter Accuracy Analysis by the Least Squares Method

The most commonly used methods for state of charge (SoC) estimation in lithium-ion batteries are Coulomb counting and data-driven approaches. However, both methods require high computational complexity and are associated with significant hardware costs. In many studies, equivalent circuit-based approaches that can be implemented on low-cost, resource-constrained processors have been explored as a solution to this problem. For this purpose, the Thevenin equivalent circuit model (ECM) was taken as reference and different electrical models such as 1RC, 2RC, and 3RC have been analyzed. The reliability of analyses conducted using equivalent circuits is directly related to the accuracy of the equivalent circuit parameters. Additionally, the number of parallel RC networks in the equivalent circuit positively impacts the accuracy of the analysis results. However, increasing the number of RC networks also raises the computational demand of the analysis. Therefore, significant efforts have been devoted to estimating battery parameters and identifying the most optimal circuit configuration. In this study, a nonlinear least squares method (LSM) is employed to minimize computational complexity while maintaining optimal accuracy. Furthermore, the impact of the number of parallel RC networks on accuracy is comparatively investigated. The accuracy of the estimated parameters is evaluated through US06 and UDDS driving cycle tests. The 1RC, 2RC, and 3RC models are compared using Root Mean Square Error (RMSE) and Mean Absolute Error (MAE) values. As a result of the study, parameter estimations were achieved with over 98% accuracy.