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): 75-81.

• 智能交通系统与信息技术 • 上一篇    下一篇

基于电子支付数据的公交车厢满载率实时估算方法

韦清波*1，苏跃江2, 3，高媛1，杨敬锋4，莫竣杰1   

  1. 1. 广州市公共交通数据管理中心，广州 510620；2. 华南理工大学，土木与交通学院，广州 510641； 3. 广州市交通运输研究所，广州 510627；4. 中山大学，广州 510275
  • 收稿日期:2020-09-01 修回日期:2020-10-22 出版日期:2021-02-25 发布日期:2021-02-25
  • 作者简介:韦清波(1984- )，男，广东高州人，高级工程师。
  • 基金资助:

    广州市科技计划项目/Science and Technology Program of Guangzhou(201804020012，201903010101)。

Real Time Estimation of Bus Loading Rate Based on Bus Electronic Payment Data

WEI Qing-bo*1, SU Yue-jiang2, 3, GAO Yuan1, YANG Jing-feng4, MO Jun-jie1   

  1. 1. Guangzhou Public Transport Data Management Center, Guangzhou 510620, China; 2. School of Civil Engineering and Transportation, South China University of Technology, Guangdong 510641, China; 3. Guangzhou Transport Research Institute, Guangzhou, 510627, China； 4. Sun Yat-sen University, Guangzhou 510275, China
  • Received:2020-09-01 Revised:2020-10-22 Online:2021-02-25 Published:2021-02-25

PDF

摘要：

针对公交监测和调度中要求实时掌握车厢满载率，以及“一票制”无法获取乘客下车信息等问题，构建基于数据驱动的组合模型，在乘客上车时即推断其出行OD站点，进而融合多源数据实现车厢满载率的实时估算。提出以K近邻算法为组合模型的核心，针对K近邻推断率过低等问题，研究在更大空间维度分析乘客出行规律并推断下车站点的方法，有效提升历史数据的利用率和下车站点的推断率；此外，针对偶发型乘客缺少历史规律数据的情况，充分利用站点下车客流量先验概率随机分配，实现电子支付乘客OD的全样本推断。利用跟车调查法对不同线路、不同班次的车厢拥挤度进行验证。结果表明，模型计算结果与实际结果相符，能够反映出不同线路、不同站段之间的车厢拥挤水平变化。

关键词: 城市交通, 车厢满载率, 组合模型, 实时估算, 大数据

Abstract:

Monitoring and scheduling bus operations require real- time grasping of bus loading rate. However, it is normally difficult to obtain the information of passenger drop-off stations in the one-ticket payment system. This study developed a portfolio model to estimate passenger's origin-destination (OD) station after passenger boarding the bus, and then performed the real-time estimation of vehicle loading rate using integrated multi- source data. The portfolio model is based on K-nearest-neighbor algorithm. Considering that there are still portion of passengers whose drop-off station are hard to be estimated, this study used a travel-area-estimation algorithm to obtain passengers' travel rules in a larger spatial dimension. This method effectively improved the estimating rate and utilization rate of historical data. In the occasional case that some historical data of passengers are missing, the study utilized the prior probability of dropoff flows at stations to perform the full sample estimation of passenger's OD in electronic payment system. A carfollowing survey was then used to verify the estimation result of different bus lines and shifts. It shows that the estimated bus loading rate is consistent with the actual results, which reflects the changes of the bus loading level.

Key words: urban traffic, bus loading rate, portfolio model, real-time estimation, mega data

中图分类号: