文章编号： 1672-5328（2015）06-0073-07

王雪松，李佳，钟楚君
(同济大学交通运输工程学院，上海201804)

摘要： 信号控制交叉口由闯红灯引起的交通事故具有较高的伤亡率。为深入了解闯红灯行为的影响 因素，利用人工记录与视频观测结合的方法对上海市4 个信号控制交叉口进行调查。基于流行病学 影响因素研究方法选择未闯红灯行为作为对照组与闯红灯行为对比，以排除无关因素。对调查数据 进行行为编码，考虑到各个交叉口的异质性，基于随机效应逻辑斯特模型在交叉口层面分析驾驶环 境、驾驶人特点、车辆特征对闯红灯行为的影响。结果表明，本地驾驶人、男性驾驶人、客车更容 易闯红灯，流量大会增加闯红灯的可能性；此外，是否为本地车辆、车内是否有乘客对闯红灯行为 也有显著影响。

关键词： 信号控制交叉口；闯红灯；随机效应模型；对照组；行为编码

中图分类号： U491

文献标识码：A

Drivers' Red Light Running Behavior Analysis

Wang Xuesong, Li Jia, Zhong Chujun
(School of Transportation Engineering, Tongji University, Shanghai 201804, China)

Abstract： Injury and fatal crash rates are generally higher at signalized intersections due to the red light running problems. To investigate red light running behavior, a research was conducted by collecting data manually and automatically with video cameras at four signalized intersections in Shanghai. Based on the investigation methods commonly used in epidemiology studies, this research compared driving behavior between vehicles running red light and vehicles not running red light to eliminate other influential factors' effect. The different driving behaviors were coded for analysis. Considering the differences in each intersection's design and traffic flow, the research analyzes the impact of driving environment, characteristics of the drivers and vehicles on red light running behavior based on Random Effect Logistic Regression Model. The results show that local, male drivers and passenger vehicle are more likely to run red light and the probability of red light running increases when traffic volume is higher. In addition, whether the vehicle is local and whether there are passengers in the vehicle also have significant impact on red light running behavior.

Keywords： signalized intersection; red light running; random effects model; control group; behavior code