Electric vehicle’s battery cell modeling and state of charge sstimation

Abstract

The rapid growth in electric vehicle (EV) adoption underscores the necessity for precise battery management systems (BMS) to ensure safety, efficiency, and longevity of lithium-ion batteries, the leading technology in EV battery packs. As EV technology evolves, the need for accurate State of Charge (SOC) estimation becomes increasingly critical, influencing battery performance, lifespan, and operational safety. Despite advancements, current SOC estimation models struggle to predict SOC accurately under diverse real world conditions, often due to simplifying assumptions or controlled testing environments. Existing models fail to capture dynamic voltage behaviors influenced by charge-discharge cycles, leading to potential inaccuracies in SOC prediction under practical scenarios. This research develops a robust SOC estimation model for EV batteries, integrating advanced battery modeling and Extended Kalman Filter (EKF) methodologies. The study seeks to improve SOC estimation accuracy and reliability, addressing the complex and nonlinear behaviors of lithium-ion batteries in varying operating conditions. Three distinct battery models, (one-state, two-state, and three-state model), each with increasing complexity and fidelity in SOC prediction are employed in this research. Using real world battery performance data, these models are refined through EKF implementation, allowing real-time SOC estimation under variable conditions. The study reveals that model complexity directly correlates with SOC estimation accuracy. The three-state model, while computationally demanding, achieves the highest accuracy The two-state model strikes a balance between accuracy and resource efficiency. Conversely, the one-state model is appropriate for low-stakes applications that do not require high SOC precision. Results highlight that different applications may require varying levels of model complexity to align with their accuracy and resource demands.