【作者】 张晨琛；

【导师】 贾利民； 王艳辉；

【作者基本信息】 北京交通大学 ， 交通安全工程， 2014， 博士

【摘要】 摘要：近年来随着我国逐渐加大基础设施建设的投资,高速公路的建设也达到了空前的规模,高速公路路网结构进一步形成。高速公路在给人们带来现代交通高效、快捷的全新感受的同时,也以高事故率、惊人的事故伤亡困扰着公路的管理者和使用者。然而,从目前现有的研究成果来看,尚缺乏一套较为完整的、面向全局的路网风险评估方法为道路的管理者提供决策支持。在路网风险影响因素分析层面,现有的路网风险评估方法仅针对单个影响因素或者几个影响因素进行探讨,尚不能从多尺度、多层次来表征路网的交通安全状况；在路网结构特征层面,尚没有一种方法能够在综合考虑路网的结构风险和运营风险的基础上对路网风险进行综合评估。本论文针对以上这些问题,结合国家“十一五”科技支撑计划项目课题——“交通安全信息集成、分析及平台构建技术开发与示范应用”,对高速公路网风险评估方法进行了深入研究。本文首先结合国内外研究成果和现场调研的实践经验,从“人—车—路—环—管”五个方面阐述了高速公路交通安全影响因素,建立了直接、间接和综合选取三类安全影响要素的选取依据,研究了安全影响要素的形式化表达方法,为下一步利用安全影响要素描述高速公路路网属性奠定了基础。以高速公路网的实际功能特性和路网拓扑结构性质为背景,对高速公路路网、基础路网、运营路网、网络构件进行了完整的形式化定义,并提出了网络构件路网的静动态属性的定义和计算模型,这些定义和模型可以全面综合反映路网及其构件的拓扑属性、物理属性及交通流影响等静态和动态性质。基于风险理论和高速公路网自身结构特点,给出了路网结构风险、运营风险及路网风险等概念的形式化定义,构建了高速公路网风险分析的理论和方法框架,明确了高速公路网安全性的内涵及其与路网风险的关系。为了对路网结构风险、运营风险及路网风险进行有效度量和定量分析评估,分别定义了路网结构风险测度集、运营风险测度集,并提出了这些测度的计算模型；其中结构风险测度集主要从路网的拓扑结构特性出发,包括基础路网非均匀性测度、连通性测度和抗毁性测度三类测度；运营风险测度集从路网交通流特性参数出发,包括运营路网非均匀性测度、连通性测度和抗毁性测度三类测度。基于对路网结构风险测度和路网运营风险测度的分析,定性描述了各类风险测度之间的相互关系,构建了路网风险综合评估的层次化关联结构；采用ANP网络分析方法,得到了路网风险测度指标的Shapley值,并根据最大Shapley熵原理计算得到测度指标集的gλ模糊测度,最后运用多重Choquet积分计算得到了路网风险评估值；为了明确路网风险和路网安全性之间的对应关系,借助待评估路网近两年的事故统计数据以及K-means聚类方法,对路网风险评估值进行了聚类分析,划分了路网安全性和风险等级。将灰色动态建模和小波神经网络模型相结合,提出了基于灰色—小波神经网络的多因素组合预测模型。该方法同时考虑了多个测度指标对路网安全性预测的影响,深化了对系统演化规律的认识,提高了预测精度,为多因素预测提供了一种新的实用方法。最后,结合课题情况,以北京市高速公路网为背景,全面运用本文提出的理论和方法,结果表明,运用本文提出的方法对路网的分析和预测结果与实际路网运营状态较为一致,预测精度较高,表明本文提出的路网风险测度理论、风险评估和预测模型能够较为真实的再现路网的实际运营状况,可以对路网风险和全局安全性进行有效的分析、评估和预测。更多还原

【Abstract】 ABSTRACT:As China enlarges investment in infrastructure construction gradually,expressway construction has reached an unprecedented scale and expressway network structure has formed further in recent years.When expressway bring people brand-new feeling about efficiency and speediness of modern traffic,they also bother road managers and users by high accident rates and amazing accident casualties.However,according to existing research achievements,the lack of a complete evaluation method about risk of expressway networks,which faces the whole situation,provides decision support for road managers.In the aspect of analyses about factors affecting risk of expressway networks, existing evaluation methods related to risk of expressway networks only direct at discussions about a single affecting factor or several affecting factors and fail to express traffic safety situations of expressway networks from several dimensions and different layers.In the aspect of features about expressway networks structure,there have been no method that can give comprehensive evaluation on risk of expressway networks based on consideration of structural risk and operational risk of expressway networks.On the basis of these problems,this thesis combines with’Traffic Safety Information Integration,Analysis as well as Development and Demonstrative Application of Platform Construction Technology’,a task of’the11th five-year’ science and technology support program,to study method of risk evaluation of expressway networks.First of all,this thesis combines with domestic and foreign research achievements and practical experience in on-site surveys,expounds factors oriented traffic safety of expressway in five aspects including’ people-cars-roads-environment-management’ and establishes three kinds of bases about selection of features of factors oriented traffic safety of expressway,i.e.,direct,indirect and comprehensive selections.In addition,it studies formalization expressing method about features of factors oriented traffic safety of expressway.All of these lay a solid foundation for us to use features of factors oriented traffic safety of expressway to construct properties of expressway networks.Under the background of practical function and topology features of expressway networks, complete formalized definition of the expressway network, physical network, opertional network are given. And the definition and calculation model of components and network’s static and dynamic features are proposed, which could comprehensively reflect the network and the components’ static topology attribute,physical property and dynamic property of traffic flow.Based on risk theory and structural features of expressway networks, complete formalized definition of structural risk of expressway networks,opertional risk of expressway networks and risk of expressway networks are given, analytical framework of risk of expressway networks is constructed,and concept connotation of traffic safety of expressway networks and its relationship with risk of expressway networks is put forward.In order to measure risk of expressway networks, this thesis defines measure set about structural risk of expressway networks and measure set about operational risk of expressway networks, respectively.In detail,measure set about structural risk mainly start with topological structure of expressway networks,including three kinds of measure,i.e., heterogeneity measure,connectivity measure and survivability measure of physical networks.Measure set about operational risk begin with traffic flow characteristic parameters of expressway networks,including three kinds of measure,i.e.,heterogeneity measure,connectivity measure and survivability measure of operational networks.Based on analysis about structural risk measure and operational risk measure of expressway networks,it gives qualitative description about interaction and relationship among risk measure and constructs a hierachical structure about risk evaluation on expressway networks.Furthermore, it obtains Shapley value of risk measure indexes of expressway networks by using ANP network analysis method.According to maximum Shapley entropy principle,it achieves gλ fuzzy measure of measure index sets by calculation. At last, it gets values of risk evaluation on expressway networks by applying multiple Choquet integral computation.To express corresponding relation between risk and safety of expressway networks clearly,it implements clustering analysis for values of risk evaluation on expressway networks and divides safety and risk grades with the help of statistical data about accidents in expressway networks in recent two years and K-means clustering method.Additionally,it combines grey dynamic modeling with wavelet neural network model, proposes a multi-factor combination prediction model based on grey-wavelet neural network.The method considers impacts of several measure indexes on prediction about safety of expressway networks simultaneously,deepens cognition of system evolution laws, improves prediction accuracy and provides a new and practical method for multi-factor prediction.Finally,it combines with situations of the task, the theory and method proposed are applied in Beijing Expressway. The results show that the analysis of network and prediction outcomes basically accord with actual operation states of the expressway network and prediction accuracy is high,which indicates that risk measure theory,isk evaluation and prediction model of the expressway networks can reflect actual operation situations of the expressway networks actually, and can analyze, assess and predict the risk and holistic safety of expressway networks effectively.更多还原