基于多源数据的城市道路交通事件检测  

Urban road traffic incident detection based on multiple data

作　　者：徐涛[1,2] 徐爱功[1] 胡超魁[1] 张明月[1] 

机构地区：[1]辽宁工程技术大学测绘与地理科学学院,辽宁阜新123000 [2]同济大学智能交通运输系统(ITS)研究中心,上海200092

出　　处：《测绘科学》2011年第6期91-94,共4页Science of Surveying and Mapping

基　　金：国家高技术研究发展计划(863计划)(2008AA11Z205)

摘　　要：针对交通事件自动检测多以高速公路、城市快速路为对象以及使用数据源单一的现状,本文提出一种基于多源数据融合的城市道路交通事件检测方法。在对信号控制下交通事件引起的交通流变化进行分析的基础上,利用杭州市城区浮动车、SCATS、Citilog、OD系统提供的实时交通数据,基于CUSUM算法构建差分流量和速度交通事件检测模型。该模型可以有效抑制交通信号对于交通流的周期性影响,协同视频、行程时间构成两方式三指标的事件检测体系。实验表明,模型在高峰时段和平峰时段均能快速准确检测交通事件。According to the fact that researches of automatic incident detection mainly focus on freeway and expressway, and data resource is usually single, this paper brought forward an approach of urban road AID （Automatic Incident Detection） based on muhiple data fusion technology. This paper analyzed the change of traffic flow caused by incident under traffic signal control, after preprocessing the data from floating car, SCATS, and Citilog systems in Hangzhou City, and the paper then built differential flow rate and average velocity AID model based on the CUSUM algorithm which can suppress the periodic influence on traffic flow. The system of AID consists of two modes including video judgment and traffic parameters analysis and three indexes which comprise differential flow rate, average velocity and travel time. The experimental results showed that during peak hours and non-peak hours, the model could detect traffic incidents rapidly and accurately.

关 键 词：智能交通系统 交通事件检测 数据融合 城市道路 差分流量 

分 类 号：U491[交通运输工程—交通运输规划与管理]