【作者】 高健；

【导师】 赵鹏；

【作者基本信息】 北京交通大学 ， 交通运输规划与管理， 2014， 博士

【摘要】 随着我国国民经济持续稳定的发展,人均收入显著提高,城市化和机动化的速度不断加快,城市的小汽车保有量不断增加,虽然城市道路也在不断的修建和拓宽,但是交通供给增加的速度远远不及交通需求,城市交通拥堵问题日益严重。此外,交通拥堵还伴随产生了交通能耗上升、交通污染严重、交通事故频发等一系列问题。面对人口、能源、环境、安全矛盾日益突出的压力,优先发展公共交通可以大大缓解我国城市交通需求增长带来的问题,有效改善城市交通拥挤状况,促使城市发展模式向节地、节能、节材转变。本论文选取公交线网、发车频率和时刻表作为优化对象,对城市公交线网进行时空协调优化,使多级公交线网布局合理,发车频率和时刻表满足要求,在时间和空间两个方面与居民的出行需求进行耦合,达到居民出行顺畅、换乘便捷、交通供给和需求平衡的目的。主要研究内容如下：(1)建立基于多智能体的城市大规模网络下居民公交出行行为仿真模型。本论文将基于活动的模型和多智能体技术相融合,以MATSim为仿真平台,建立了大规模交通网络下的居民公交出行行为模型,主要包括仿真环境要素的搭建、多智能体属性和行为规则的定义以及多智能体学习和自适应机制的建模;并以保定市为例,构建了基于多智能体的微观仿真场景,将仿真结果与实际调查数据进行比较,结果表明模型精度满足应用要求。其中,研究重点在于仿真场景的构建以及输入数据的准备。(2)以居民公交出行距离为依据,建立多级公交线网等级配置模型。本论文针对城市公交线网分层规划理论缺乏的情况,提出基于居民公交出行距离的公交线网等级配置方法。选取周转量为量化指标,建立公交线网供给周转量模型和居民出行需求周转量模型,从供需平衡的角度研究公交线网等级配置；并以河北省保定市为例,对方法和模型进行应用,验证公交线网等级配置方法的实用性。公交线网等级配置的研究可为微观的公交线网时空协调优化指明方向。(3)基于居民出行行为,建立城市公交线网时空协调优化模型,并设计启发式算法求解模型。第1阶段是进行公交线网的空间优化,以直达客流量最大为目标,生成各公交线路候选方案,随机组合候选线路,生成多个公交线网布局候选方案,接着利用MNL模型和BP神经网络研究公交空间优化对居民出行行为的影响,为下一阶段的公交时间优化奠定基础；第2阶段是在公交空间优化的基础上进行公交时间优化,利用双层规划模型对居民出行行为和公交时间优化之间的动态关系进行建模,其中居民出行行为通过多智能体技术进行仿真,最终得到各公交线网候选方案的最优时刻表；第3阶段是确定最终的公交空间和时间优化方案,在各最优时刻表方案下,计算对应公交线网候选方案的直达客流量并进行比较,选择直达客流量最大的公交线网方案以及与之对应的时刻表作为公交时空优化的最终方案。最后,利用时空协调优化理论和方法对保定市的公交系统进行了系统优化,证明了模型和算法的有效性。(4)基于微观仿真的公交时空评价方法研究和评价系统开发。以乘客为核心,针对微观仿真结果的特点,建立公交站点、公交线路和公交线网三层评价指标,并研究如何利用AHP、K-均值聚类分析和灰色关联度法进行综合评价；在理论研究的基础上,利用GIS技术开发公交评价系统,提高评价工作的有序性和高效性。评价系统在保定市的公交现状评价以及公交优化效果评价中得到了应用。更多还原

【Abstract】 With the steady development of our national economy, the individual income goes up significantly, the speed of urbanism and mechanization get higher and the car ownership increase gradually. Even though the urban road is continuously constructed and broadened, the increasing speed of traffic supply is far lower than demand, thus the traffic congestion grows seriously day by day. In addition, the traffic congestion will cause the energy consumption, traffic pollution, traffic accident and so on. Being faced with the pressure from population, energy, environment and safety, prior development transit system can remit the growing of travel demand, improve the traffic condition and convert the development modal of city to land-saving, energy-saving and material-saving.In order to address the issues in public transportation, this paper conducted a space-time coordinating optimization of urban transit network in terms of public transit network, frequency and timetable, and tried to make the layout of multilevel transit network more rational, and frequency and timetable more satisfying. The research coupled residents’ trip demands based on time and space, which made the trip convenient and traffic supply and demand balanced.(1) A model of residents’trip in large-scale transit network with the combination of multi-agent and activity-based model was established, which included the construction of simulation environment factors, attributes of multi-agent, definition of behavior rule, and modeling of multi-agent learning and self-adaptive mechanism. Herein, city of Baoding, which is located on Hebei Province, China, was used in the study. The prediction restuls from multi-agent microscopic simulation are compared with the observed data and the comparison results indicate that the prediction accuracy is applicable. The study focused on the construction of simulation scene and method of preparing the input data.(2) Public transit network hierarchy optimization model based on the residential transit trip distance. To the problem of being lack of transit network hierarchy theory, a research on public transit network hierarchy optimization based on the residential transit trip distance was conducted. From the view of balance between supply and demand, the supply turnover model and demand turnover model were both developed. The method and models were applied to transit network hierarchy optimization of Baoding. (3) A space-time coordinating optimization model of public transit was developed based on the residents’travel behavior and characteristic, meanwhile, the corresponding heuristic algorithm was proposed to solve the model. The space optimization was conducted in the Stage1. The set of candidate lines was generated with the objective of maximum direct flow, and then the set of candidate network was randomly combined. The MNL model and BP neural network were both used to study the influence of transit network layout on residents’travel behavior and the outcome would be employed into the next optimization stage. The time optimization was conducted based on the space optimization in the Stage2. The bi-level model was employed to describe the dynamic relationship between residents’travel behavior and time optimization, in which, the residents’travel behavior was simulated based on the multi-agent technology. The optimized schedule of each candidate network was put out in this stage. The final space-time schedule of public transit was determined in Stage3. With the optimal schedule of each candidate network, the direct flow of each candidate network was calculated and the candidate network with the highest direct flow was chosen as the optimal network, thus the optimal network and schedule were both found. Finally, the optimization model and algorithm were both applied to the Baoding’s transit system.(4) Evaluation method of public transit was proposed and the evaluation system of public transit was developed based on the micro-simulation. With the core of passengers and based on the features of micro-simulation results, the evaluation index system was established from three levels:station-level, line-level and network level. In addition, the AHP, K-mean Cluster Analysis and Gray Correlation Method were used for comprehensive evaluation. Based on the evaluation methods, the evaluation system was developed with the GIS technology and thus the evaluation work would be much more ordered and efficient. The evaluation system was applied to evaluating the current situation and optimized results of Baoding’s public transit system.更多还原