State of Charge Estimation for Power Batteries in New Energy Vehicles Considering Temperature Fluctuations

To address the issue of inaccurate state of charge(SOC)estimation for power batteries in new energy vehicles in scenarios involving temperature variations, a modified battery SOC estimation method based on neural network and unscented Kalman filter(UKF)is proposed. Initially, a Thevenin equivalent circuit model with dynamic parameters is established, taking into account the effects of temperature fluctuations. The characteristics of changes in battery open-circuit voltage(OCV)under varying temperature conditions are examined to establish the relationship between battery SOC and OCV. Additionally, the alterations in battery capacity under temperature variations are analyzed. A neural network is employed to train the neural network temperature factor for battery capacity reduction corresponding to temperature changes. Furthermore, the dynamic model parameters are identified using the recursive least squares algorithm with a forgetting factor. Subsequently, the neural network temperature factor and the UKF are utilized for real-time adjustments to ensure precise SOC estimation under changing temperature conditions. The results demonstrate that the battery SOC estimation method considering temperature variations significantly enhances SOC estimation accuracy compared to traditional methods. In a low-temperature environment of -15 ℃, the proposed method enhances SOC estimation accuracy by 2.77%. © 2024 Xi'an Jiaotong University. All rights reserved.