Abstract: To Classify drivers' driving styles to measure the impact of different driving styles on the energy consumption of pure electric vehicles. Using principal component analysis and K-means to divide driving style types, and using support vector machine (SVM) to build models for training. Then, based on the AVL Cruise simulation model, the bus running energy consumption data is output, and the energy consumption differences of drivers with different driving styles are compared. The results show that after the model labeled prediction of the test set, the classification accuracy rate was 93%, Under the same route, the aggressive driver has more energy recovery. Compared with the normal driver, the fuel consumption per mileage of the aggressive driver and the cautious driver increased by 2.81% and 4.44% respectively. Taking advantage of the degree of influence, measures such as screening and training of drivers are taken accordingly, thereby reducing the energy consumption of pure electric bus operation.

Key words: Driving style; Pure electric bus; Driving energy consumption; Principal component analysis; SVM; AVL Cruise simulation model

摘要： 对驾驶员的驾驶风格进行分类，衡量不同驾驶风格对纯电动车能耗的影响。运用主成分 分析和 K-means 划分驾驶风格类型，利用支持向量机（SVM）建立模型进行训练；然后基于 AVL Cruise 仿真模型输出公交车运行能耗数据，比较不同驾驶风格驾驶员的能耗差异。结果 表明，模型对测试集进行标签预测后，分类准确率在 93%；同一线路下，激进驾驶员能量回 收较多，对比正常型驾驶员，激进型和谨慎型驾驶员单位里程油耗分别增加了 2.81%、4.44%。 利用影响程度，相应采取对驾驶员进行筛选培训等措施，进而降低纯电动公交车运行能耗。

关键词: 驾驶风格；纯电动公交车；行驶能耗；主成分分析；支持向量机；AVL Cruise 仿真模型