【作者】 王山川；

【导师】 关宏志；

【作者基本信息】 北京工业大学 ， 道路与铁道工程， 2006， 硕士

【摘要】 传统四阶段法将交通需求预测分成四个阶段进行,针对各个阶段开发的模型,特别是方式划分和交通分配阶段,一般都包含反映交通服务水平的变量。由于计算分阶段进行,导致了交通服务的变量在各个阶段并不一致。针对此问题,国外一些学者进行了联合模型的研究,但研究仅限于两种交通方式(主要是轨道交通和小汽车方式)。在我国,特别是城市交通系统,存在多种交通方式,如轨道交通、常规公交、出租车、小汽车、自行车等,因此,有必要研究多方式的联合模型。本论文以对联合模型为研究核心,其他内容围绕此展开。论文首先对方式划分模型进行了研究,利用在北京调查的非集计数据,建立了RP(Revealed Preference)模型、SP(Stated Preference)模型和融合数据模型。模型结果表明:融合模型不仅可以克服RP数据的属性范围受限和相关的问题,而且可以钝化SP数据的对政策变量的过分敏感性。在以往对交通领域中平衡问题研究的基础上,提出了多种方式运行下网络平衡的概念,即在多种交通方式运行的网络上,给定OD交通量情况下,出行者无法通过改变出行方式或出行路径,获得更小的广义费用,此种状态达到多方式运行下交通网络平衡。并根据各种交通方式运行时与小汽车流的相互影响关系,提出了交通方式分类方法,以此分类法为基础,建立了联合多方式划分和交通分配模型,进行了等价性和唯一性证明,给出了模型算法。算法中,提出使用黄金分割法修正小汽车流对第二类交通方式流的影响。最后给出了联合模型的一个算例,进一步证明了模型和算法的有效性。更多还原

【Abstract】 Traditional four-stage method divides traffic requirement forecasting into four stages. At present each model that every stage developed, generally include variables of the level of service, especially in the stages of model split and traffic assignment. Because going on stage by stage, variables are inconsistent in different stages. Some scholars have studied the combined model to solve this problem, but these studies dealt with two traffic modes (generally abut modes of subway and car). There are many traffic modes in our country, especially urban traffic system, such as subway, regular public transit, taxi, car, bicycle and so on. Therefore, study on multi-mode combined model is necessary.This paper takes the multi-mode combined model as the core, and the other contents process around this core. This article studies the model split firstly, then takes Beijing as the example, establishes the RP data models, the SP data model and the fusion data model, and it carries on the analysis of model sensitivity. The results indicate: because of the attributes correlation and the limitation of attribute levels, the RP data model the parameter of expense estimated value mark and actually not symbol; The SP data model precision is high, but some t Test values of the parameters are not very remarkable; The t Test values of the parameters in fusion data model parameter are remarkable, the model elasticity is small, the sensitivity is deactivated.Based on the former studies of the network balance, this paper proposes the concept of the multi-mode network balance: many traffic modes run on a traffic network under the designated OD matrix, travelers are unable to obtain smaller generalized cost by changing traffic modes or trip routes, this state reaches the multi-mode network balance. Then this paper formulates a classification method of traffic modes, which accords to relation that different traffic modes affect mutually while moving. Based on that classification of various traffic modes, this test proposes a combined modal split and trip distribution model of multinomial modes. And the corresponding identification and algorithm are provided. In the algorithm, this paper proposes using更多还原