【作者】 陈志军；

【导师】 吴超仲； 朱敦尧；

【作者基本信息】 武汉理工大学 ， 交通运输工程， 2012， 硕士

【摘要】 随着我国社会的快速发展,随之而来的道路交通安全问题也日益凸显。道路交通事故逐渐成为影响民众安全感的重要因素之一。根据我国近几年的道路交通事故统计分析,引起交通事故的主要原因中车辆驾驶员的危险驾驶行为一直占主导地位。大型车辆肇事事故是造成群死群伤事故的主要原因,其社会影响非常恶劣。根据我国道路交通事故统计年报分析,截止2009年底全国客车的保有量仅占总数的25.94%,而肇事比例却占总数的41.47%。通过分析大型车辆肇事的特大交通事故可以发现,超速、违规驾驶等是事故主要原因。为减少大型车辆肇事事故发生率,虽然国内外很多研究机构研究开发了危险驾驶行为监测系统,但现有的危险驾驶行为监测系统具有的监测功能比较单一,不能适应变化的环境,不能同时监测多种危险驾驶行为,需要研究开发具有多源信息共享、一定自主性并可以监测多种危险驾驶行为的大型车辆危险驾驶行为监测一体化系统。具体研究内容如下：介绍国内外现有车辆危险驾驶行为监测系统,分析现有危险驾驶行为监测系统存在的弊端,进而分析了研究开发大型车辆危险驾驶行为监测一体化系统的需求。引入Agent理论和多Agent信息融合理论,并介绍了Q学习算法和该算法在agent模型中的应用。在Agent理论基础上建立大型车辆危险驾驶行为监测一体化系统的4个监测agent和一个融合agent模型,并将Q学习算法融入监测agent中。在多Agent信息融合理论的基础上建立大型车辆危险驾驶行为监测一体化系统的多agent信息融合模型。并在融合算法中引入可信度矩阵和D-S证据理论方法。根据大型车辆危险驾驶行为监测一体化系统的多agent信息融合模型设计大型车辆危险驾驶行为监测一体化系统的硬件系统和软件系统。通过实车实验数据检测大型车辆危险驾驶行为监测一体化系统的可靠性和准确性。更多还原

【Abstract】 With the rapid development of our society, the problem of road traffic safety is also becoming loom large. Road traffic accidents is becoming one of the important factors that affect people’s sense of security gradually.Recent years, according to statistical analysis of road traffic accidents in our country, the driver’s dangerous driving behavior has been dominant in which cause the road accidents. Large vehicles causing trouble is the main reason which cause group of injury and death,and is bring very bad social influence. According to our country statistics annals analysis of road traffic accidents,the bus has accounted for only25.94%of the total but its accident proportion is41.47%of the total,by the end of2009. Through the analysis of some large vehicle traffic accidents which caused by large vehicles, we can find that speeding、illegal for driving and following too closely are the main reasons of the accident. To reduce the large vehicle’s accident rate,at home and broad,some research institutions developed some dangerous driving behavior monitoring system,but the existing dangerous driving behavior monitoring system has the single monitoring function which can’t adapt to the changing environment and can’t monitor a variety of dangerous driving behaviors at the same time. So we need to study a dangerous driving behavior monitoring integration system of large vehicles which can sharing many sources of information, and has certain autonomy, and also can monitor a variety of dangerous driving behaviors.There are some specific content as follows:This paper introduces the existing dangerous driving behavior monitoring system of vehicles and analyzes the deficiencies of the existing dangerous driving behavior monitoring system,and then analyzes the demand of the dangerous driving behavior monitoring integration system of large vehicles.This paper Introduces Agent theory and Multi-Agent information fusion theory,and introduces the Q learning algorithm and its application in the agent model.Based on the Agent theory,this paper establishes four monitoring agent and a fusion agent model of the dangerous driving behavior monitoring integration system of large vehicles,and integrates Q learning algorithm into monitoring agent.Based on the Multi-Agent information fusion theory,this paper establishes Multi-Agent information fusion model of the dangerous driving behavior monitoring integration system of large vehicles,and introduces the credibility matrix and the D-S evidence theory method in the fusion algorithm.According to Multi-Agent information fusion model of the dangerous driving behavior monitoring integration system of large vehicles,this paper designs hardware and software systems of the dangerous driving behavior monitoring integration system of large vehicles.Through the car experiment data,this paper detects the reliability and accuracy of the dangerous driving behavior monitoring integration system of large vehicles.更多还原