【作者】 余勇；

【导师】 陈德旺；

【作者基本信息】 北京交通大学 ， 交通信息工程及控制， 2008， 硕士

【摘要】 随着城市快速交通迅猛发展,城市快速路路网结构日益复杂,交通事件也日益增多。由于快速路通行量大又有较高的稳定行驶车速,比普通的城市道路相对封闭,比城市间的高速公路流量大,一旦发生交通事件,如果不能得到及时的处理,就非常容易造成交通拥堵,因此准确、快速的判断交通事件的发生对整个快路网的畅通运行有着十分重要的现实意义。交通事件自动检测是高速公路/快速路智能交通管理系统的重要组成部分,提高事件自动检测算法性能可以显著改善事件管理系统的运行效果。本论文所依托的是北京市自然科学基金资助项目《北京快速路交通流模型与交通诱导关键技术研究》(8052016)的一个研究内容,同时也是北京四通智能交通系统集成有限公司项目《北京市快速路智能交通管理平台》的一个子模块。文中阐述了国内外交通事件自动检测方法的研究现状,目前的事件自动检测算法各有侧重点,没有一种的性能完全优于其他算法。本文在分析了交通事件对交通流产生的影响的基础上,针对现有算法各种的优缺点,综合考虑发生交通事件时交通流数据在两维空间上、时间上的异常变化,提出了一种易于实际应用的三维集成事件自动检测算法。该算法设计思想是当交通事件发生时在横向维上考虑相邻车道间流量和速度的变化,在纵向维上考虑上下游断面占有率的变化,在时间维上考虑占有率在占有率-流量模型上的等级变化,然后集成结果报警。本文还详细介绍了其他子模块集成判断模块和参数更新模块的设计思想和工作步骤。为验证算法,本文采用北京快速路实测交通流数据和事件信息对本文算法进行有效性分析和实用性分析,结果表明本文算法基本满足实际应用需要。最后在明确系统目标的基础上,利用Visual C++和oracle数据库实现了实际应用的算法软件。本文中涉及到的研究思路、研究方法是针对有固定检测器路段交通事件自动检测方法的一个探索,可以为交通事件自动检测算法提供一定的参考。更多还原

【Abstract】 With the rapid development of urban rapid transport, urban road network structure becomes more complex and traffic incidents also become much more. Because there is a lot of high-speed traffic on urban freeway, which is more closed than ordinary urban road and more crowded than inter-city freeway, urban freeway will congest heavily if the incidents can’t be cleaned in time. So an accurate, rapid detection of the happening of traffic incidents has a very important practical significance to the unobstructed network. Traffic automatic incident detection (AID) is an important part of freeway emergency management system, and high-performance automatic incident detection algorithm can significantly improve the system’s effect. The thesis stems from a research content of Beijing Natural Science Foundation-funded projects named as "Research on Traffic Flow Model and Key Technologies of Traffic Guidance for Beijing Urban Freeways" (8052016) and also is a sub-project of "intelligent traffic management system in Beijing" which is a project from Beijing STONE Intelligent Transportation System Integration CO., LTD.After an extensive literature review of existing AID techniques, the paper introduces the key points of existing algorithms and finds there is not an absolute perfect algorithm. On the basis of the analysis of traffic incidents’ impact and according to the temporal and the two-dimensional spatial abnormal variability of traffic flow as a result of traffic incidents, the paper proposes an easily applicative AID algorithm based on 3D integration. The design idea is that when a traffic incident happens, the algorithm integrates the change of volume and speed of adjacent lanes in the horizontal dimension, the change of occupancy of upstream and downstream section in the vertical dimension and the change of occupancy levels of the flow-occupancy model in the temporal dimension to alarm incidents. The paper also describes design ideas and working procedures of integrated judgment module and parameters update module in the details.Then in order to verify the algorithm, the paper analyses the validity and practicability of the algorithm with the detector data from Beijing freeway, and the result shows the integrated algorithm can meet basic needs of practical application. And on the basis of the clear system objective, the paper uses the Visual C++ and Oracle technology to program an applicative software. The research ideas and the algorithm proposed in this paper is a method exploration of traffic automatic incident detection algorithm of fixed detectors and may provide certain theoretical reference.更多还原