【作者】 杜海涛；

【导师】 蒋阳升；

【作者基本信息】 西南交通大学 ， 交通运输规划与管理， 2013， 硕士

【摘要】 城市道路立交桥作为解决城市交通节点问题的重要措施近年来在各个城市,特别是大型城市得到了广泛的应用。确定立交桥的形式是城市道路立交桥设计与建设前期最重要的环节。立交桥的选型由于受到多种不同因素的影响,从根本上上来说,是一个多指标的决策问题。现行的城市道路立交桥选型方法多是交通工程设计人员根据经验的判断确定若干可行的立交方案,然后通过立交桥的通行能力计算及方案的比选来确定最终的桥型。而实际上,由于传统的立交桥桥型分类方法大多过于粗略,备选方案的可靠性很大程度上依赖于设计人员的个人经验及自身水平而缺乏必要的理论依据,设计人员对于影响城市道路立交桥选型的因素往往考虑不充分等原因,城市道路立交桥在投入使用之后的使用效果并不理想。针对以上问题,本文系统的研究了城市道路立交桥分类体系、探讨了选型影响因素,构建了一套城市道路立交桥选型方法理论。首先,本文在分析传统的城市道路立交桥分类体系的基础上,根据立交桥左转匝道的出口方式、入口方式、是否存在小环道、是否存在匝道的交织情况等四个方面对城市道路立交桥的类型进行了重新界定,确立了立交桥的分类体系。根据确定的立交桥类型,采用二进制编码方式对不同的立交桥形式进行了编码。其次,本文详细分析了影响城市道路立交桥选型的各种主要因素及其影响机理,对可以量化的的因素,如通行能力和路网的影响进行定量化的描述,建立数学表达式；对于不能量化的因素,找到影响因素与立交桥桥型之间的变化规律。在此基础之上,基于定量化的因素建立了立交桥选型的双层规划模型。考虑到路网节点对路网设计双层规划模型的重要影响,在模型中加入了路网节点转向延误。其中,上层模型以带转向延误的路网阻抗最小为目标,通过分散搜索和改进的自适应遗传算法的组合算法来求解；下层模型为带转向延误的随机用户均衡模型的配流问题,采用连续平均法进行求解。通过求解双层规划模型,选取其中前若干最优种子作为立交桥选型的备选方案。对于确定的备选方案集,利用不可定量化的立交桥影响因素作为决策指标,采用多指标决策理论方法,来确定最终的桥型。最后通过Visual C++软件对模型进行仿真设计求解。更多还原

【Abstract】 The Urban interchange has been widely used as an important measure to solve the problem of urban intersection in many cities, especially in big cities. The type selection of interchange is the most important link in both pre-construction and design of interchange, and its basis is to establish a system which is used for its classification and coding. Fundamentally speaking, the type selection of interchange is a multiple attribute decision making problem as it is affected by many different factors. Existing selection methods of urban interchanges are that transportation engineers select a series of possible interchange proposals based on their work experience, and then calculate the traffic capacity of possible prosposals’and compare them to determine the last proposal. But in fact, the effect of urban interchanges is not very good because the common classification methods of interchange are too rough, and the selection options are largely dependent on engineers’personal experience and their ability which are lack of necessary theoretical basis, and the factors which influencing urban interchange type’s selection are usually not considered fully. To solve the above problem, this paper systematically studies the classification system of urban interchange and the factors which influencing urban interchange type’s selection, and then establishes a set of urban interchange selecting method theory.Firstly, on the basis of analyzing traditional urban interchange classification system, types of urban interchange is redefined and the interchange classification system is established according to exits and entrances of the left turning ramp as well as whether existing small ring roads and weaving section of ramps. After confirming the types of interchange, the author adopts binary coding method to encode for different overpass types. Different interchange forms and coding are illustrated and we find the adopted coded way is easy and simple.Secondly, based on the analysis of the factors, influencing urban interchange type’s selection, and their impact mechanism, a quantitative description for the quantifiable factors, such as capacity, the effects of network is given. Change Rules between other unquantifiable factors and urban interchange are found.On this basis, a Bi-level Programming model is built which consider of quantifiable factors. Taking into the important impact of the node of road network on Bi-level Programming model, the node steering delay of road network is brought into the model. The upper level model is objective to the minimum of impedance which includes steering delay and can be solved by the combing Algorithms of Scatter Search Algorithms and Improved Adaptive Genetic Algorithm. The lower level model is a traffic distribution problem under Stochastic User Equilibrium Model which includes steering delay and can be solved by the Method of Successive Averages. The selection options of urban interchange are obtained by solving the Bi-level Programming Model. And then, a Multiple Attribute Decision Making theory method is used by’taking the influence factors which cannot be quantified as decision-making index. At last, the models will be solved by Visual C++software.更多还原