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 (2): 65-72.

• 智能交通系统与信息技术 • 上一篇    下一篇

基于手机信令数据的常规公交站间OD识别

于泳波*1, 2，侯佳1, 2   

  1. 1. 南京市城市与交通规划设计研究院股份有限公司，南京 210018； 2. 江苏省交通大数据与仿真平台技术工程研究中心，南京 210018
  • 收稿日期:2021-01-31 修回日期:2021-02-15 出版日期:2021-04-25 发布日期:2021-04-25
  • 作者简介:于泳波(1992- )，男，江苏南通人，工程师。

Bus Trip OD Identification Based on Mobile Phone Data

YU Yong-bo*1, 2 , HOU Jia1, 2   

  1. 1. Nanjing Institute of City & Transport Planning Co. Ltd, Nanjing 210018, China; 2. Jiangsu Transportation Big Data and Simulation Platform Technology Engineering Research Center, Nanjing 210018, China
  • Received:2021-01-31 Revised:2021-02-15 Online:2021-04-25 Published:2021-04-25

PDF

摘要：

为识别手机用户乘坐常规公交的OD，结合公交车辆GPS轨迹，在考虑常规公交换乘行为的基础上，建立基于手机用户与车辆轨迹相似度的常规公交出行识别模型，以及站间OD概率模型。通过地铁出行识别，融合手机信令数据与IC卡数据，提取包含百万样本的公交与非公交出行数据，以此作为验证集。进一步分析各参数取值、出行距离、公交线路重复系数等因素对公交出行与站间OD识别的效果。结果表明：在验证集中，常规公交出行方式识别精确率可达0.807，召回率0.912，换乘识别精确率可达0.660，召回率0.756，公交线路识别准确率可达75.5%，站间OD 识别准确率可达71.9%，参数取值的不同对识别效果影响较大。此外，出行距离越长、公交线路重复系数越低，公交线路与站间OD识别准确率越高，出行距离6 km以上、平均公交路段重复系数4 以下的识别效果最佳。

关键词: 城市交通, 公交OD, 轨迹相似度, 公交出行识别, 手机信令数据

Abstract:

This paper develops the bus trip identification model and the OD probability model based on the connection between the traveler's mobile phone data and the bus trajectory data. The proposed models help to identify the Origin to Destination (OD) of mobile phone users' bus trips considering the bus Global Positioning System (GPS) trajectory and the traveler's transfer behaviors in the identification of subway trips, the mobile phone data and Integrated Circuit (IC) card data are integrated to extract the bus and non- bus trip data with millions of samples as the verification set. The impact from major parameters on the identification accuracy are also analyzed such as trip distance, the overlap of bus routes, etc. The results show that: in the verification set, the accuracy rate of bus trip recognition is 0.807, recall rate is 0.912, transfer recognition accuracy rate is 0.660, recall rate is 0.756; the bus line recognition accuracy rate reach 75.5%, and the inter station OD recognition accuracy rate is 71.9%; different parameter values have significant impact on the recognition effect. In addition, the longer the travel distance and the lower the bus line overlap coefficient, the higher the bus lines and OD recognition accuracy would be. The recognition effect appears to be the best when the trip distance is more than 6 km and the average bus section overlap coefficient is less than 4.

Key words: urban traffic, bus OD, trace similarity, bus trip identification, mobile phone data

中图分类号: