文章编号： 1672-5328（2025）03-0054-08

徐士伟1, 2，陈鹏1, 2，何鸿杰1, 2，宋程1, 2
(1. 广州市交通规划研究院有限公司，广东广州510030；2. 广东省可持续交通工程技术研究中心，广东广州 510030)

摘要： 微观仿真中城市轨道交通车站乘客出行行为与设施服务时间参数标定存在精度不足的问题。 针对通勤型、枢纽型及特殊型三类车站，提出了一种基于视频采集与图像识别技术的参数调查方 法，实现了车站内乘客行为数据的自动化采集与指标统计。以广州市城市轨道交通车站为例开展的 精细化参数调查表明：三类车站呈现明显差异化特征，通勤型车站在设施服务时间方面表现最优； 枢纽型车站不仅服务时间最长，且乘客携带大件行李的比例最高；特殊型车站的各项参数特征均介 于前两类车站之间。通过对图像识别方法进行效率验证和仿真案例的应用分析，验证了该参数调查 方法的有效性，为精细化仿真研究车站乘客行为提供了可靠的数据依据。

关键词： 城市轨道交通车站；乘客参数；微观仿真；特征挖掘；图像识别

中图分类号： U491.2+6

文献标识码：A

Passenger Parameters Characterization at Urban Rail Transit Stations for Microscopic Simulation

XU Shiwei1, 2, CHEN Peng1, 2, HE Hongjie1, 2, SONG Cheng1, 2
(1. Guangzhou Transport Planning Research Institute Co., Ltd., Guangzhou Guangdong 510030, China; 2. Guangdong Sustainable Transportation Engineering and Technology Research Center, Guangzhou Guangdong 510030, China)

Abstract： There is a lack of accuracy in the calibration of passenger travel behavior and facility service time parameters in microscopic simulations of urban rail transit stations. This paper proposes a parameter survey method based on video collection and image recognition technology for three types of stations: commuter, hub, and special-purpose. The approach enables automated collection and indicator analysis of passenger behavior data within stations. A detailed parameter survey involving a case study conducted at a rail transit station in Guangzhou demonstrates significant variation among the three station types: commuter stations exhibit the most efficient facility service times; hub stations record the longest service times and the highest proportion of passengers carrying large luggage; and special-purpose stations have intermediate characteristics across all parameters. Efficiency validation and simulation-based application confirm the effectiveness of the image recognition technology-based parameter survey method, providing a reliable data foundation for refined simulation study of passenger behavior at stations.

Keywords： urban rail transit stations; passenger parameters; microscopic simulation; feature extraction; image recognition