Abstract: The traditional state of charge (SOC) prediction of power batteries is carried out under ideal experimental conditions, which may not be able to adapt to the real complex and changeable driving conditions. This work relies on the national big data alliance for new energy vehicles, using a data-driven approach, using real driving data to predict the current SOC of power batteries. The power battery type used in this paper is lithium iron phosphate battery, the alarm group and the prediction group are designed, and the logistic regression algorithm is used to extract the battery characteristic information of the low residual power SOC, and the model is trained, and the trained model is used to predict. The results show that the proposed model has high accuracy for low SOC diagnosis.

Key words: New energy vehicles; Power batteries; Low state of charge; Logistic regression algorithm; Risk prediction

摘要： 传统的关于动力电池荷电状态（SOC）预测是在理想实验条件下进行的，这未必能适应真实复杂多变的行驶工况。本工作依托新能源汽车国家大数据联盟，利用数据驱动的方法，使用真实的行车数据对动力电池当前荷电状态进行预测。文章采用的动力电池类型为磷酸铁锂电池，设计了报警组与预测组，采用逻辑回归算法对低剩余电量荷电状态的电池特征信息进行提取，并训练模型，使用训练好的模型进行预测。结果显示，文章提出的模型对低荷电状态诊断具有较高的准确率。

关键词: 新能源汽车；动力电池；低荷电状态；逻辑回归算法；风险预测