Abstract: In order to accurately obtain the micro boarding and alighting data of bus passengers, to mine the residents’ bus travel characteristics, and to optimize the bus planning and operation decision-making, the bus passenger flow information collection technology was researched, which was based on the bus GNSS positioning information and the MAC address of the Wi-Fi module of the passenger mobile terminal. Focusing on the problem of redundant interference of a large number of MAC addresses during vehicle operation, an online passenger identification algorithm was proposed. Through analyzing the interference conditions, the MAC address of bus passengers online was recognized based on five feature levels of classification calculation, namely, the number of MAC address occurrences, K-means clustering, online duration, online displacement, and the maximum acquisition time interval. Integrating vehicle positioning information, combined with the opening rate correction coefficient, the bus OD was calculated, and the accurate boarding and alighting locations of some passengers were obtained. The research results indicate that, the results of the manual car-following survey verify that the proposed system can better provide the most timely and accurate bus travel data for the bus operation managers, compared with the bus IC card survey method, image (video) recognition and other methods.

Key words: intelligent transportation; bus passenger flow statistics; online passenger identification algorithm; GNSS information; MAC address

摘要： 为准确获取公交乘客微观上下车数据，挖掘居民公交出行特征，优化公交规划和运营决策，基于公交车GNSS定位信息及乘客移动终端WiFi模块的MAC地址，研究公交客流信息的采集技术。重点针对车辆运行过程中的大量MAC地址冗余干扰问题，提出在线乘客识别算法。通过干扰情况分析，分别基于MAC地址出现次数、K均值聚类、线上时长、线上位移、最大采集时间间隔，五种特征层次分类计算，识别在线公交乘客MAC地址。整合车辆定位信息，结合开启率修正系数，计算公交OD，并获取部分乘客准确上下车地点。研究结果表明：通过人工跟车调查结果进行验证，该系统相较于公交IC卡调查法、图像（视频）识别等方法，可以更好地为公交运营管理者提供最及时、准确的公交出行数据。

关键词: 智能交通；公交客流统计；在线乘客识别算法；GNSS信息；MAC地址