Comparative evaluation of advanced SoC estimation and active cell equalization techniques in electric vehicles

Abstract

Advanced State of Charge (SoC) estimation and active cell equalization techniques are essential for improving the accuracy, safety, lifespan, and energy efficiency of lithium-ion batteries in Electric Vehicles (EVs). However, inaccurate SoC estimation, cell imbalance, battery degradation, reduced service life, safety concerns, and inefficient energy utilization remains significant challenges for Battery Management System (BMS). This paper presents a comparative evaluation of advanced SoC estimation and active cell balancing techniques using the Extended Unscented Kalman Filter (EUKF), Multi-Adaptive Square Root Extended Kalman Filter (MA-SREKF), Model Predictive Control (MPC), and Robust Model Predictive Control (RMPC). Under nonlinear operating conditions, the proposed MA-SREKF achieves superior SoC estimation accuracy, reducing the maximum estimation error to ± 1.5% compared with EUKF. For active cell balancing, RMPC outperforms MPC by reducing SoC error fluctuations from ± 8 to ± 3%, resulting in faster convergence and improved balancing accuracy. The comparative results demonstrate that the MA-SREKF and RMPC provide enhanced robustness, reliability, and computational efficiency, making them well suited for real-time BMS applications in EVs.