交通运输系统工程与信息

交通运输系统工程与信息 ›› 2021, Vol. 21 ›› Issue (1): 62-68.

• 智能交通系统与信息技术 • 上一篇    下一篇

轨迹数据驱动的单点信号交叉口运行效率评价方法

蔡晓禹*1a，1b，吕亮1a，1b，卢凯明1a，1b，唐小勇2，高志刚2   

  1. 1. 重庆交通大学，a. 交通运输学院，b. 山地城市交通系统与安全重庆市重点实验室，重庆 400074； 2. 重庆市城市交通大数据工程技术研究中心，重庆 400020
  • 收稿日期:2020-10-24 修回日期:2020-11-20 出版日期:2021-02-25 发布日期:2021-02-25
  • 作者简介:蔡晓禹(1979- )，男，四川达州人，教授，博士。
  • 基金资助:

    国家自然基金青年科学基金/National Natural Science Foundation of China(61703064)；重庆市技术创新与应用示范专项社会民生类重点研发项目/ The Social and People's Livelihood Key Research and Development Project of Chongqing Technology Innovation and Application Demonstration Projects(cstc2018jscx-mszdX0085)。

Single-point Signalized Intersection Operation Efficiency Evaluation Method Based on Trajectory Data

CAI Xiao-yu*1a，1b, LV Liang1a，1b, LU Kai-ming1a，1b, TANG Xiao-yong2, GAO Zhi-gang2   

  1. 1a. School of Transportation, 1b. Chongqing Key Laboratory of Traffic System & Safety in Mountainous Cities, Chongqing Jiaotong University, Chongqing 400074, China; 2. Chongqing Urban Transportation Big Data Engineering Technology Research Center, Chongqing 400020, China
  • Received:2020-10-24 Revised:2020-11-20 Online:2021-02-25 Published:2021-02-25

PDF

摘要：

为解决信号交叉口运行效率评价方法偏理论化、实用性不强等问题，以出租车、公交车、驾图 (车联网)多源GPS轨迹数据为基础，充分利用车辆减速、停车、加速等连续速度变化特征及位置信息，提出交叉口个体车辆排队长度、通行时间、停车次数等交通参数提取技术。基于此，构建以信号交叉口运行指数为一级指标，车辆平均通行时间，第95%分位排队长度，两次停车率为第二级指标的评价指标体系，基于高斯混合聚类模型对综合运行指数进行分级量化，最终确定五级服务水平评价标准。实例表明：采用本文方法能够比较准确地反映不同时段交叉口各转向、进口道及交叉口整体运行水平，最大排队长度与实际相对误差约15%左右，优于仿真结果；除样本不足情形以外，服务水平评价结果比仿真更贴近实际运行情况，验证了本文方法的技术可行性及评价结果的客观真实性。

关键词: 城市交通, 轨迹数据, 交叉口通行时间, 排队长度, 两次停车率, 运行效率评价

Abstract:

The existing method for signalized intersection operation efficiency evaluation is mostly theoretical and not practical. This paper proposes a method to evaluate vehicles' performances at signalized intersections including queue length, travel time, number of stops, etc. The study collected multi-source GPS trajectory data of taxi, bus and driving map (car network). and obtained continuous speed change characteristics and position information of vehicle deceleration, stop and acceleration. The performance evaluation takes the operation status of the signalized intersection, as the first-level indicator and the average vehicle travel time, the 95 percentile queue length, and the frequency of cycle failure as the second- level indicators. Based on the Gaussian hybrid clustering model, the study graded and quantified the comprehensive performance measures and determined a five- level of service evaluation criteria. The example analysis shows that the method is able to accurately reflect the performances of the turning movements, approaches, and overall operation level of the intersection during different time periods. The relative error of the maximum queue length calculated from the proposed method and the queue length in the field is about 15%, which is smaller than what obtained from the simulation. Except for some cases of insufficient samples, the level of service evaluation results is closer to the actual operation than the simulation. The example analysis verifies the technical feasibility of the proposed method and the accuracy authenticity of the evaluation results.

Key words: urban traffic, trajectory data, intersection travel time, queue length, rate of cycle failure, operation efficiency evaluation

中图分类号: