State-of-Charge Estimation of Li-ion Battery Packs in Electric Vehicles: An Inverse Design

The battery management system of electric vehicles requires an accurate and reliable predictor of the state of charge (SoC) of the battery. The true SoC of a lithium battery is influenced by many factors, such as battery temperature, the number of charge and discharge cycles, and battery aging. Therefore, it is difficult to achieve ideal results by detecting external battery characteristics such as battery voltage, current, temperature, and internal resistance to predict SoC. To address this challenge, this article uses the non-battery external variable of speed as the input to the prediction model. After integrating the speed over time, linear regression algorithm is used to predict SoC. Furthermore, considering the effect of battery aging on the accuracy of SoC prediction, this article uses sliding time windows to update the training set, thereby achieving accurate SoC prediction. The experimental results show that the digital twin model achieves a prediction accuracy of over 95%in practical applications. In addition, as speed characterizes the external characteristics of electric vehicles, predicting SoC through speed establishes the connection between electric vehicles and electric vehicle batteries.