【作者】 司徒炳强；

【导师】 靳文舟；

【作者基本信息】 华南理工大学 ， 交通信息工程及控制， 2011， 博士

【摘要】 “公交优先”已成为当前城市交通发展的主题。在大力发展城市公交过程中,我们也清晰地看到公交系统的吸引力不够,公交分担率依然较低,距离我们的期望值还有较大差距。这是由于公交运营系统的可靠性一直没有得到根本提高和有效保障导致的。公交运营系统中,运行时间不稳定,编制的行车时刻表与实际运营效果相差较大,给乘客、公交企业和管理部门都带来巨大的成本损失。为了履行“公交优先”政策,通过提高计划的可靠性来应对现实中很多不确定因素的影响,使得公交运营能够在计划的预期和有效控制范围内进行,以此来提高公交分担率,就必须在公交网络运营优化环节中,深入研究城市公交网络行车时刻表编制的理论方法及可靠性问题。本论文主要从影响公交网络行车时刻表可靠性的时空二维因素的角度出发,利用优化理论、可靠性理论、复杂网络理论和时间窗等理论方法,结合公交网络协同发车的行车时刻表自身特点,研究了城市公交网络行车时刻表编制的理论方法及可靠性的关键问题。具体来讲,本论文主要工作和研究成果有如下几个方面:(1)从多个角度和利用多种方法对公交车运行时间的特性进行了详尽的剖析,包括:公交车运行时段,运行车辆,最大上/下车乘客数,运行时间比例,运行总体数值及其分布。部分重要的结论是:总停站时间一般占总运行时间的比例RDT为20%-40%,各线路到达站点的总运行时间均具有马太效应的现象,不同班次随着站点序号递增,总运行时间的差距越来越大,并且总运行时间都服从正态分布N(μ,σ2)等等。(2)研究了公交车运行时间可靠性模型及可靠性控制方法。提出了一种新的公交线路运行时间可靠性模型和公交网络运行时间可靠性模型;提出了带有Hub点的公交车运行时间可靠性的弹性冗余设计控制方法,据此提出了“大间隔所有站点法”和“大间隔Hub点法”这两种方法对应的公交线路和网络运行时间可靠性模型;对所提的模型和控制方法进行了实证检验:分析马太效应检验、非参数检验和可靠度对比检验这三方面的检验结果。(3)利用复杂网络理论分析公交网络复杂特性。从城市公交网络运营角度出发,提出了两种新的公交网络描述方法以及网络Hub点的确定方法;通过实证分析,证明广州市公交网络属于无标度网络,并列举出部分Hub点;提出了线路关联度概念和模型;然后还提出基于时刻表的公交网络Hub点演化模型构造算法,最后提出改进的有向沙堆模型来研究网络相继故障对公交网络运营可靠性的影响,发现不管线路是否进行协同发车,网络Hub点一旦受到蓄意攻击,整个公交网络的运营可靠性都会受到巨大影响。(4)提出了两套区域运营思路下的公交网络协同发车行车时刻表编制的理论模型:①考虑到站时间差的合作与竞争条件下公交网络行车时刻表优化模型(等间隔发车);②考虑乘客等车时间(换乘等车时间和非换乘等车时间)的多线组合公共交通网络发车行车时刻表模型(非等间隔发车)。所提模型都能给出Hub点时刻表,其中第二套模型方法还引入了多模式和多线路双重组合的区域运营思路。实例分析表明:两套模型都能够保障乘客需求和公交企业供给之间的平衡,以及最大限度减少双方的成本。(5)最后提出了公交网络行车时刻表可靠性双层规划模型。首先分析时间窗与Hub点时刻表的内在联系,接着提出了带时间窗和Hub点时刻表的公交网络行车时刻表可靠性的双层规划模型,下层模型为追求公交网络乘客总等车时间最小的行车时刻表模型,上层模型为追求可靠度最大的运行时间可靠性模型,提出嵌入式粒子群算法求解双层规划模型。最后,将成果应用到广州大学城多条公交线路组成的实例网络中,结果表明:非等间隔发车的行车时刻表在客流变化明显时段具有更好效果,所提模型能够给定可靠性较高的各线路在始发站和网络Hub点的行车时刻表。更多还原

【Abstract】 Bus priority has become the subject of urban traffic development. When urban public transit is developed vigorously, we clearly see that the attractiveness of public transit system is not enough, and the bus share rate remains low, where there is a large gap from the expected value. It is because that the reliability of public transit system has not been fundamentally improved and effective protected. In the bus operating system, the instability of running time and the large gap between timetable and actual preparation make the passengers, bus companies and management with a huge loss. In order to fulfill bus priority policy, it improves the reliability of plan to deal with the uncertain fators and makes the bus running under the plan and the control, to impove the public transit share rate, so we must study the theories, methods and issues of creating bus timetable and its reliability.This dissertation, from the point of spatiotemporal affect factors to reliability bus timetable, uses the optimization theory, reliability theory, complex network theory and time windows theory to study the the theories, methods and issues of creating bus network timetable and its reliability, considering with the characteristics of collaboration dispatching timetable. The main contents of this dissertation are summarized as follows:(1) It analyses the characteristics of bus running time from several angles and methods, which include running periods, vehicles, maximum of on/off passengers, proportion of running time, the overall running time value and its distribution. Some important conclusions are as follows: firstly, the ratio of dwell time to total running time (RDT) is from 20% to 40%. Secondly, the total running time, from originating station to a given bus station, display the Matthew’s phenomenon, which means that the gap between total time with every route is growing when station number is increasing. Thirdly, the total running time with every route and station are normal distribution N (μ,σ2), and so on.(2) Reliability model and control method to bus running time are studied. We propose the new reliability models of bus running time under one bus route or a bus newwork. Then, we propose a flexible redundancy control method with Hub points for the running time reliability. According to the control method, the reliability models of bus running time under route or network to“large interval with all stations”and“large interval with Hub point”are proposed. For the models and control method, we have some simulating test, which include Mattehew phenomenon test, non-parametric test and comparative reliability test.(3) We study the complexity of public transit network by use of complex network theory. From the point of operating of public transit network, we propose two new methods for describing the public transit network and how to determining the Hub points. Then, the empirical analysis proves that Guangzhou public transit network is scale-free network, and some Hub points are listed. We also propose the concept and model of route correlation, then also propose the construction algorithm of Hub point evolution model based on timetable, and finally propose an improved sandpile model to study the influence of operating reliability by cascading failure in public transit network. The results show that whether it is collaboration dispatching or not, once the Hub point is deliberate attacted, the operating relibility of whole public transit network will be greatly affected.(4) We propose two sets of bus timetbable models under regional operation public transit network. The first one is a timetable optimization model for transit network in cooperation and competition, which is considered the time disparity of arriving a same station and gives the same interval for dispatching. The second one is a timetable model for collaboration dispatching with multi routes, which is considered the waiting time (transfer waitting time and non-transfer waitting time) for the passengers and gives the different interval for dispatching. Both of the models can be gave the timetable in Hub point. The second model includes the regional operational idea with combination of multi modes and multi routes. The rusults of example show that these two models can balance the passenger demand and enterprise supply, and minimize the costs of both them.(5) In the last section, we propose a bus timetable reliability bi-level programming model. Firstly, we analyse the inner contact between time window and Hub point timetable. Secondly, we propose the bus timetable reliability bi-level programming model including the time window and Hub point timetable. The lower model is the timetable model for minimum total waitting time of passengers, and the upper model is the reliability model of running time for maximum the reliability. An embedded Particle Swarm Optimization is established to solve the bi-level programming model. Finally, the proposed method and model are applied in the public transit network in Guangzhou University Town. The results show that the timetable with non-equal interval dispatching has better results when the passengers flow change obviously, and the bi-level programming model can provide the timetable with high reliability in origin station and Hub point station for every routes.更多还原