文章编号： 1672-5328（2019）04-0105-09

周继彪1, 2，赵鹏飞3，董升1，张水潮1
(1.宁波工程学院建筑与交通工程学院，宁波浙江315211；2.同济大学交通运输工程学院，上海201804；3.北京 工业大学建筑工程学院，北京100124)

摘要： 为科学评估地铁车站内行人的拥挤状态，针对车站内部客流分布特征，采用改进的蚁群聚类 算法，提出了行人拥挤状态等级划分方法。以西安地铁自动售检票系统的历史客票数据为研究对 象，利用其17 个车站内的拥挤强度、拥挤持续时间以及拥挤影响范围等进行行人拥挤分级，分别 得到进站、出站以及进出站三种状态下的聚类分级结果。结果表明：全日地铁客流呈现出明显的M 形分布，早高峰集中于8:00—9:00，晚高峰集中于18:00—19:00；三种状态下行人密度累积频率分 布曲线符合对数函数，表明客流量随着客流密度的增加而降低。计算结果与实际客流的变化规律一 致，说明基于改进蚁群聚类算法对车站内的行人拥挤等级进行划分是可行的，能够反映地铁车站内 部行人的拥挤程度。

关键词： 交通工程；拥挤分级；蚁群聚类算法；客流特征；地铁车站

中图分类号： U491

文献标识码：A

Pedestrian Congestion Levels at Subway Stations with Ant Colony Algorithm

Zhou Jibiao1, 2, Zhao Pengfei3, Dong Sheng1, Zhang Shuichao1
(1.School of Civil and Transportation Engineering, Ningbo University of Technology, Ningbo Zhejiang 315211, China; 2.School of Transportation Engineering, Tongji University, Shanghai 201804, China; 3.College of Architecture and Civil Engineering, Beijing University of Technology, Beijing 100124, China)

Abstract： To accurately evaluate the pedestrian congestion state at subway stations, this paper develops a classification method for pedestrian congestion with an improved ant colony clustering algorithm based on the distribution characteristics of passenger flow. With the historical data from the Xi'an subway automatic ticketing system, the paper classifies the pedestrian congestion levels based on the degree of overcrowding, duration and scope of overcrowding at the 17 stations, which also yields the clustering results for passenger flows at ingress, egress, and mixed ingress and egress conditions. A clear M-shaped passenger flow during 24 hours is emerged with the morning peak concentrated between eight and nine o'clock and the evening peak concentrated between 18 and 19 hour. The cumulative frequency curve of passenger density for the three congestion levels is consistent with the logarithmic function, which shows that the passenger flow decreases with the increase of passenger density. The consistency between the calculation results and actual passenger flow data indicates that the classification of pedestrian congestion levels at subway stations with the improved ant colony clustering algorithm actually reflect the overcrowding situation in the real world.

Keywords： traffic engineering; classification of congestion level; ant colony clustering algorithm; passenger flow characteristics; subway stations