摘要 2011年，我国发布了第一个国家级的道路交通安全专项规划——《道路交通安全“十二五”规划》，提出了十二五期间的道路交通安全目标。道路交通系统是一个复杂的动态系统，道路交通安全建设需要全社会的共同努力。科学合理的道路交通安全目标能够起到良好的导向作用，提高政府和公众的道路交通安全意识，凝聚各方力量共同参与，是改善道路交通安全的行之有效的重要措施。道路交通安全目标的决策需要实现科学化、规范化、民主化。相较于国外发达国家，我国的道路交通安全目标决策工作起步较晚，实践经验较少，相关研究成果也比较少。因此，研究适合于我国国情的道路交通安全目标决策的理论、方法和工具已经成为我国道路交通安全管理的客观需要。 本文以道路交通安全目标为研究对象，从我国道路交通安全管理工作的实际情况出发，运用道路安全工程、计量经济学、计算机科学等学科中的有关理论与方法，研究服务于县级以上行政区域的道路交通安全目标决策支持系统。 论文首先对道路交通安全目标决策及其支持系统的概念进行了界定，然后分析了我国道路交通安全目标决策的流程、目标设立所需考虑的主要相关因素。在此基础上，从用户和功能两个方面深入分析了系统的需求，由此设计了系统的结构及工作流程。 在进行了系统的总体设计之后，论文对系统涉及的三项道路交通安全目标决策问题进行了研究，包括道路交通安全影响因素分析、道路交通安全趋势预测和道路交通安全目标的分解。 在道路交通安全影响因素分析中，论文运用变异系数、相关系数和面板模型从国家和省级区域两个层面分析了经济、人口、车辆、道路、交通量与道路交通安全之间的年际关系、总体关系、省际关系和短期关系。经济因素包括国民生产总值、人均国民生产总值、机动化水平和城镇化率；人口因素包括总人口、驾驶人数、65岁以上人口数、农村人口数、城镇人口数；车辆因素包括机动车保有量、民用汽车拥有量、营运汽车拥有量、摩托车数量；道路因素包括公路里程、等级公路里程、城市道路长度、城市道路面积；交通量包括公路客、货周转量。论文还根据我国道路交通安全的宏观政策和交通事故肇事数据从正反两方面探讨了管理因素对道路安全的影响。 在道路交通安全趋势预测中，针对道路交通安全目标决策的需求，研究了事故起数、受伤人数、死亡人数、万车死亡率和10万人口死亡率的预测模型。构建的ARIMA预测模型依据事故本身的规律建模，适用于数据的获取局限性较大的情况；构建的多元变结构协整模型以多项宏观影响因素为输入变量，拟合了包含结构突变的长期协整关系，适用于考虑目标衔接的预测；构建的ECM模型综合考虑了国民生产总值与道路安全的长期和短期关系，具有较高的预测精度。基于以上3种模型构建的组合模型在大多数情况下能进一步提高预测精度。 在道路交通安全目标分解中，分省际和省内两种情况进行了研究。对于省际目标分解，提出了“确定基数、分配消减量、计算分解目标”的三步法。重点研究了第二步中的目标分解原则、目标分解的指标体系以及基于信息熵的多目标线性加权分配模型。通过目标分解实例对该方法进行了分析与评估。在省际目标分解方法的基础上，研究了省内安全目标分解的指标体系和指标权重计算方法。 最后，总结了本文的主要研究成果，简要探讨了进一步的工作方向。 关键词：道路安全，安全目标，决策技术，预测技术，目标分解 |
In 2011, Chinese first national road traffic safety plan was developed, which is “the 12th Five-year Plan of Road Traffic Safety”. In the plan, a national road safety target was set. Road traffic system is a complicated dynamic system and requires the efforts of the whole society to improve its safety. A scientific and feasible road safety target is an effective measure which is able to lead the participants to improve road safety and raise the awareness of governments and the public. The decision for road safety targets needs to be scientific, standardized and democratic. China started late on decision work for road safety targets and has less experience and relative research results compared with the developed countries. Therefore it is the objective need of our road traffic safety management to explore and study the theories, methods and tools of road traffic safety management based on the situation of our country. The study object of this paper is road traffic safety targets. According to the practices of the actual conditions of Chinese road safety management, this paper studies the decision support system for road traffic safety targets at and above the county level using theories and methods of road safety engineering, econometrics and computer science. First, the concepts of road traffic safety targets decision and its support system are defined. The procedure of Chinese decision work for road safety targets and the main factors considered in target setting are analyzed. Based on the analysis, the user requirements and the functional requirements are analyzed. And software structure and work flow of the system are designed. After the overall design of the system, three main problems about setting road safety targets are studies, including influencing factor analysis, trend prediction and target decomposition of road traffic safety. In influencing factor analysis, the relationships of annual change, overall change, inter-provincial change and short-term change between road safety and several factors such as economy, population, vehicle, road and traffic volume are studied using variable coefficient, correlation coefficient and panel data model. The study is conducted at a national level and provincial region level as well. Economic factors include gross domestic product, per capita gross domestic product, motorization level, urbanization rate. Population factors include total population, driver numbers, population over 65 years old, rural population, urban population. Vehicle factors include motor vehicle ownership, civilian car ownership, operation vehicle ownership and number of motorcycles. Road factors include highway mileage, grade highway mileage, urban road length and urban road area. Traffic volume factors include highway passenger turnover and highway freight turnover. The influence of management factors on road safety is discussed based on macro policy and traffic accident data. . In road safety trend prediction, several prediction models are developed aiming to solve the needs of road safety target decision. The models are used to predict accident number, death number, injury number, mortality rate per 10000 vehicles and mortality rate per 100000 population. The ARIMA prediction model is built on the basis of traffic accident own laws and can be used in low data accessibility. The multivariate structural change co-integration prediction model has several macro influencing input variables and fits the long term co-integration relationship based on structural change, which has relatively high prediction accuracy. The combination model based on the three above models gives a higher prediction precision in the majority of cases. In target decomposition, the inter-provincial and intra-provincial target decomposition are studied respectively. The three-steps method which are “baseline data determination, reduction number distribution, decomposed target calculation” and the principles for target decomposition are proposed. In the second step of the method, the index system and the multi-objective linear weighting allocation model based on information entropy are proposed. The method is analyzed and evaluated by the case study of national road safety target decomposition. Based on inter-provincial target decomposition, the index system and weight calculation method for intra-provincial target decomposition are studied. Finally, an application example of the system is given and the main research findings are summarized and the following work is discussed briefly. Key words: road safety, safety target, decision technology, perdition technology, target decomposition |
博士论文