Abstract: Based on the weaving area of urban continuous tunnel, the lane change point selection behavior of weaving vehicles on the mainline was analyzed. Based on the vehicle trajectory data collected in the field, the random forest model was used to analyze the characteristics of the lane change point selection behavior of the mainline vehicle and the influencing factors of the behavior decision. Then, the support vector machine model and the random forest model were used to model and compare the lane change point selection behavior. The research results show that the main factors affecting the selection of lane change points for mainline weaving vehicles are the target vehicle state and the rear vehicle state of the lane, and the accuracy of the lane change decision model established after feature screening is higher. The support vector machine model can better describe the lane change behavior, and the prediction accuracy of the model is not less than 85%. The research results help to describe the lane change point selection behavior more accurately in the simulation model and provide support for the optimization of the weaving area of urban continuous tunnels and the formulation of assisted driving strategies.

Key words: traffic and transportation engineering; urban continuous tunnel interweaving zone; lane change point selection behavior; behavior characteristic analysis; decision model

摘要： 基于城市连续隧道交织区分析主线交织车辆换道点选择行为，以实地采集的车行轨迹数据为基础，利用随机森林模型分析主线车辆换道点选择行为特征及影响行为决策的影响因素，而后分别采用支持向量机模型与随机森林模型进行换道点选择行为决策建模和对比分析。研究结果表明：影响主线交织车辆换道点选择的主要因素为目标车辆状态以及本车道的后车状态，且经特征筛选之后建立换道决策模型的精度更高，其中支持向量机模型可以较好的刻画换道行为，模型的预测精度不低于85%。研究成果有助于在仿真模型中更准确描述换道点选择行为，同时为城市连续隧道交织区的优化和辅助驾驶策略制定提供支撑。

关键词: 交通运输工程；城市连续隧道交织区；换道点选择行为；行为特征分析；决策模型