【作者】 姚宝珍；

【导师】 杨成永；

【作者基本信息】 北京交通大学 ， 道路与铁道工程， 2011， 博士

【摘要】 随着城市化进程的推进,城市公交系统不断向立体、多元化方向发展。城市公交枢纽作为各种交通方式衔接的物理载体,其布局和运营效率是否合理直接关系到整个线网是否通畅。因此,本文从枢纽的衔接角度出发,对枢纽的空间布局及运营调度优化进行了较为深入的探讨。本文的主要内容如下：(1)城市公交枢纽的空间布局优化研究。合理的空间布局是发挥枢纽功能和作用的基础,以最低的空间消耗提高整个网络的运输效率。由于公交枢纽空间布局优化问题是非线性优化模型,是一个NP-hard问题,如果备选枢纽数量较大将很难求解出合理的布局方案。为了提高模型的求解精度和减少计算时间,根据枢纽布局的主要影响因素,建立了基于魅力度的备选枢纽筛选模型以缩小枢纽优化的解空间。然后,在备选枢纽筛选的基础上,分别提出单枢纽布局优化模型和多枢纽布局优化模型。由于多枢纽布局优化模型是一个多目标优化模型,本论文开发了基于排序法的多目标遗传算法对其进行求解。最后通过大连市主城区的数据对模型和算法进行了检验。结果显示,基于魅力度的备选枢纽模型得到的备选枢纽非常符合大连市主城区的实际情况。同时,多枢纽优化结果表明,对于公交换乘乘客来说,枢纽间的组合效用最大,可以提供更顺畅的的公交服务。(2)城市公交枢纽时刻表优化研究。公交枢纽静态调度是公交运营者主要的日常工作。时刻表优化是静态调度中最重要的组成部分,其直接或间接决定了车辆排班、司机调度等工作。因此,本文针对枢纽内的公交线路的运营特点,以枢纽内对等待时间影响最大的线路(例如,轨道线路或大间隔的公交线路等)作为基准线路,使其他线路与基准线路之间实现最大同步换乘。考虑到公交车辆(列车)运行的随机性,特别是常规公交车辆,本文引入一个松弛时间来完善枢纽内公交线路的衔接程度。然后,从优化单枢纽时刻表入手,拓展到多枢纽的时刻表优化研究,并开发基于启发式算法———SCE-UA的求解算法对枢纽内公交线路的时刻表进行了优化。最后,基于仿真分析对模型和算法进行了检验,结果显示,多枢纽联合优化的效果优于每个枢纽的单独优化,而且,增加松弛时间的优化模型的鲁棒性要更好。(3)城市公交枢纽的动态调度决策研究。公交车辆的运营环境非常复杂,经常会受到很多随机因素的干扰。通过采取有效实时调度措施,减弱由于干扰带来的影响,从而恢复枢纽内公交线路的正常运营。本文针对动态调度对车辆运行信息的实时性要求高的特点,提出公交车辆运行时间预测模型,并通过衰减因子来提高模型的预测精度。同时考虑到公交车辆运行状况是动态的,准确预测最佳的松驰时间可以减少乘客的等待时间,因此,本文提出松驰时间预测模型。然后基于车辆运行实时信息基础上,分析基于枢纽的动态调度的特点,提出以最小化乘客等待时间为目标的动态调度模型。考虑到动态调度模型是一个复杂问题,采用遗传算法对模型进行了求解。最后通过实例对模型和算法进行了验证。本文在枢纽布局优化、时刻表优化以及动态调度等方面有一定的创新,对城市交通规划的研究和实践具有一定的借鉴意义。更多还原

【Abstract】 With the development of urbanization process, the urban transit system has been developing rapidly. Public transportation hubs are the physical carriers of urban transit system. Whether the layout and operational efficiency of the hubs are reasonable directly influences the connection of the whole network. Thus, according to the connection of hubs, this paper focuses on location optimization and operation scheduling of public transportation hubs. The main contents of this paper are as follows: (1) The location optimization of public transportation hubs. Proper location of hubs is the basis to achieve its function. The transport efficiency of the whole network should be improved at the lowest consumption. As the optimization process of hubs location is related to many factors, it is difficult to get a reasonable solution if there are a large number of candidate hubs. To improve the accuracy of the solution and reduce the computational time, a candidate hub location model based on fascination is presented. Then, on the basis of candidate hubs model, an optimization model of a single hub location and of multi-hub location are also proposed. As the multi-hub location optimization model is a multi-subjective optimization model, a multi-subjective genetic algorithm based on a ranking method is adopted to solve the multi-hub location optimization model. The model and the algorithm are examined by the data of Dalian city and the results indicate that the candidate hubs gained by the candidate hubs model are in accordance with the actual situation of Dalian city. And the proposed multi-hub location model is more effective than the single hub location model for proving more transit service.(2) The timetable optimization of public transportation hubs. Public transportation hubs operation is the main work of transit operator. Timetable optimization is the most important part of the operation, which directly or indirectly determines the vehicle scheduling, driver arrangement, etc. Therefore, according to the operation features of transit routes in hubs, this paper considers the route which plays the strongest effect on the waiting time in hubs as the basic line to achieve the maximum synchronization of other routes with the basic line. In order to reduce the randomness effect of arrival time of transit vehicles, some slack time is proposed. Thus, the timetable model of the multi-hub is also proposed based on the timetable model of the single hub. SCE-UA, an evolution algorithm, is used to solve this model. Finally, the model and the algorithm are examined by simulation analysis. The results demonstrate that the effect of multi-hub timealbe optimization is better than that of each hub timetable optimization and the robustness of the optimization model with the slack time is better than that of the optimization mode without the slack time.(3) The research on dynamic control strategies of public transportation hubs. The running environment of transit vehicle is very complicated and often affected by many stochastic factors. The influence can be weakened by taking effective measures to restore the normal operation of transit routes in hubs. Due to the real-time feature of dynamic dispatching, the prediction accuracy of transit vehicle running time prediction model can be improved with a forgetting factor. Considering that the condition of transit vehicles is dynamic and the accurate prediction of optimal slack time can ensure a good convergence of transfer routes and reduce the waiting time of passengers, the slack time prediction model is also presented. And on the basis of real-time vehicles operational information, the features of dynamic control strategies based on the hubs are analyzed, and a dynamic dispatching model aiming at minimizing the waiting time of passengers is proposed. Since the dynamic dispatching model is a complicated problem, a genetic algorithm is also used to solve the model. Then the model and the algorithm are examined with data of Dalian city.Finally, a summary is given and some contributions are discussed.更多还原