【作者】 李兆翠；

【导师】 刘培玉；

【作者基本信息】 山东师范大学 ， 计算机软件与理论， 2008， 硕士

【摘要】 我国是道路交通事故总数,事故死亡人数最多的国家,连续数年一直居世界第一位。交通事故已严重影响国民经济的发展,社会的稳定以及和谐社会的建设。为妥善处理交通事故,公平公正的认定事故责任人,减少交通事故,保障国家财产和人民生命财产安全,国家每年都要花费大量人力、财力用于道路交通事故的处理。交通事故的发生既有其特定的、偶然的原因,又受其他各种因素的支配。如果能够发现道路交通事故数据中各种因素之间的关联,特别是驾驶员、车辆、道路、天气、时间等可能引起道路交通事故发生的原因因素与事故本身属性中的事故类型、事故主要原因和现场等可能导致道路交通事故的结果因素之间的规律,那么交通管理部门就可以根据驾驶员、车辆、道路、天气、时间等因素来判断导致道路交通事故发生的可能性。从而辅助交通管理部门更好的工作,这对减少交通事故的发生有着十分重要的意义。本文对道路交通事故特征、导致交通事故的人、车、路等因素进行关联规则数据挖掘,并用虚拟现实技术模拟交通事故的发生过程,实现对交通事故数据的分析,从而辅助交通部门公平公正的认定事故责任。主要的研究工作如下:1.对现有的道路交通事故分析处理技术进行了探讨先介绍了常见的交通事故致因研究方法,然后介绍了数据挖掘理论、关联规则、Apriori算法以及多维多层的关联规则。最后介绍了交通事故现场模拟再现关键技术,OBB包围盒碰撞检测算法及三维仿真技术。2.给出了多维Apriori改进算法对多维Apriori算法进行改进,在进行关联规则分析前,预先指定该维的哪一个层次参与关联分析,其它层次不参与关联分析,将问题简化为单纯的多维数据模型的挖掘,从而提高挖掘的速度与效率。3.建立了基于数据挖掘的道路交通事故模型针对道路交通事故这一实际问题,采用了星型数据模型,利用多维Apriori改进算法对数据进行分析。算法采用循环迭代的策略,对数据进行操作,最后得到符合约束阈值条件的规则。利用关联规则的基本原理,对以往的道路交通事故数据进行关联分析,得出满足最小支持度和最小置信度的关联规则,查找出事故本身因素与驾驶员、车辆、道路、天气和时间因素的联系。4.设计与实现了道路交通事故分析处理系统以数据挖掘技术和虚拟现实技术为基础,设计并实现了道路交通事故分析处理系统。该系统包含三个模块,分别为事故基本信息模块、事故模拟再现模块以及事故辅助决策分析处理模块。事故基本信息模块可对事故基本信息进行存储,修改和维护。事故模拟再现模块对事故案例进行模拟再现,展现交通事故的发生过程,辅助交通管理部门进行决策。事故辅助决策分析模块利用多维Apriori改进算法,针对济南市市区的道路交通事故数据进行测试,分析交通事故发生的主要原因。并对事故信息进行分析总结,认定事故责任。对交通安全给出建议,为决策者提出具有实际意义的治理方案及预防措施提供了辅助参考。本系统以数据挖掘的相关理论为基础,利用有效的挖掘方法,从大量的道路交通事故数据中挖掘出有价值的信息。将数据挖掘应用到道路交通事故数据分析中,找出环境与事故原因、事故形态之间的联系,分析事故形成的原因。同时根据事故现场信息,采用虚拟现实技术模拟交通事故发生的全过程,辅助交通管理部门客观公正地处理交通事故,认定事故责任,提高交通事故处理的工作质量。更多还原

【Abstract】 With the most total number of road traffic accidents and death toll, China has been ranking first for several years in the world. Traffic accidents have seriously affected the national economic development and social stability as well as the building of harmonious society. To properly handle the traffic, find responsible person for the accident fairly and impartially, reduce traffic accidents, protect state property and people’s lives, the state has spent a lot of manpower and financial resources to deal with road traffic accidents every year.The occurrence of traffic accidents has its specific, fortuitous reasons, and subject to the influence of various other factors. If we can find the relations between various factors in the road traffic accident data, especially, the rule between traffic accidents reason such as drivers, vehicles, roads, weather, time, which may cause traffic accidents and the result factors such as the accident types, the main reason, scene which may cause the traffic accident result. Then the traffic management departments will judge the possibility which the road traffic accident occurs according to the drivers, vehicles, roads, weather and time factors. It can guide the traffic management department to work better, and there is great significance in reducing the occurrence of accidents.In this article it makes association rules data mining among the road traffic accident characteristics, drivers, vehicles and road factors, simulates traffic accident using virtual reality, and analyses the traffic data. Thereby it helps the traffic management departments to identify accident responsibility fairly and impartially. The major research works are as follows:1. Introducing the current techniques of road traffic accident analysis and transactionFirstly, it elaborates the common research methods of traffic accidents. Secondly, it introduces the data mining basic theory, association rules, Apriori algorithm and multi-dimensional association rules. Thirdly, it introduces the simulation and reconstruction techniques of road traffic accident, OBB bounding box collision detection algorithm and three-dimensional simulation technology.2. Proposing the multi-dimensional Apriori improved algorithmBefore analyzing the association rules, it pre-specifies which level of the dimension participates the correlation analysis, the other levels doesn’t participate the correlation analysis. Thus the problem is reduced to a simple model of multidimensional data mining, thereby enhancing Mining speed and efficiency.3. Establishing the road traffic accidents model based on data miningIn view of the road traffic accidents problem, it uses the star data model, using the multi-dimensional Apriori improved algorithm to analyse data. The algorithm uses the circulation iteration strategy to operate data, finds the rules in the restraint threshold conditions. Using the association rule basic principle, this model does association analysis to the former road traffic accident data, obtains association rules satisfied the minimum support and the minimum confidence, finds the relations among the accident factors and drivers, vehicles, roads, weather and the time factor.4. Designing and realizing the road traffic accident analysis and transaction systemTaking the data mining and virtual reality technologies as foundations, it designs and realizes the road traffic accident analysis and transaction system. There are three modules in this system, such as the basic information module, the accidents simulation and reconstruction module and the accidents assisting decision analysis and procession module. The basic information module may store, modify and maintain the basic information of accident. The accidents simulation and reconstruction module reconstructs the traffic accident, assists the traffic management departments to make decision. The accidents assisting decision analysis and procession module uses the multi-dimensional Apriori improved algorithm to test the Jinan district road traffic accidents data, analyses the main reason caused accidents. It analyses and summarizes the accidents information, recognizes the accident responsibility. Making suggestions to the traffic safety, for policy makers it provides some practical significance plan and preventive measures as supplementary reference.Based on the data mining related theory, this system uses the effective method to mine the valuable information from the massive road traffic accidents data. Appling the data mining theory in the road traffic accidents data analysis, it finds the relations between the road condition factors and the accidents, analyzing the reasons caused accidents. In the same time, it uses the virtual reality techniques to simulate and reconstruct the accident according to the accident scene information. This system can helps the traffic management departments to handle accidents and identify accident responsibility fairly and impartially, improving the traffic accident procession quality.更多还原