【作者】 吕能超；

【导师】 刘正林； 严新平；

【作者基本信息】 武汉理工大学 ， 载运工具运用工程， 2010， 博士

【摘要】 大型活动进场和散场期间,短时间内急剧产生的出行需求和汇聚的交通流给场馆周边的交通基础设施带来了巨大压力,会产生交通拥堵和停车场不足等问题。这种临时突变的交通矛盾不可能完全依靠新增道路和停车场等设施来解决,实施合理的交通组织和管理措施能够有效地利用其他道路和停车场资源,缓解大型活动期间的交通供需矛盾。然而,如何从需求和供给入手,考虑出行者的路径选择行为,制定高效合理的交通组织和管理方案是值得研究的问题。考虑了大型活动交通组织和管理中交通规划者和出行者之间的相互影响,应用了基于双层规划理论的优化方法,对大型活动期间停车票价优化问题、停车诱导信息优化问题以及散场期间网络单行优化问题开展了系统研究。主要完成了以下的研究工作：(1)研究了双层规划模型在大型活动交通组织和管理方案制定中的建模方法以及模型的求解算法。为避免大型活动交通组织和管理过程中可能出现的诡异现象,研究了交通网络设计的双层规划优化方法在交通组织和管理方案优化建模种的应用。在优化模型中,将交通组织优化问题的规划者和出行者分别考虑成上层模型和下层模型的决策者,利用双层规划模型描述交通组织管理方案制定过程中规划者和出行者的博弈过程。针对不同优化问题,分别用准动态交通分配模型、准动态随机用户均衡模型描述需求波动情形的出行者的路径选择行为。针对双层规划非线性、难求解的特点,采用粒子群算法和量子进化算法求解双层规划优化问题。(2)基于双层规划模型框架,建立了大型活动停车票价优化模型,实现了停车票价对不同类型停车场的需求调控。分析了大型活动小汽车出行方式的停车-换乘/步行多方式出行的特性,对小汽车出行方式的出行网络进行了转换,将停车场后一段费用相对稳定的路段转化为停车场的吸引度,使得原出行停车场选择问题可以用出行分布/交通分配模型描述。在此基础之上,采用基于Logit效用模型描述出行者选择停车场的行为,考虑了驾驶时间、停车费用、公交费用和行走费用等主要因素,并以此作为双层规划模型的下层模型。分析了大型*本论文研究得到国家973项目“城市交通系统的组织优化与控制”(2006CB705505)和973项目“城市生命体承载系统的健康识别和调控理论与方法研究”(2005CB724205)支持。活动出行总时间组成部分,以及停车票价和停车场容量约束,以此作为双层规划的上层模型。设计了基于离散粒子群的算法求解双层规划模型上层问题并采用凸组合算法求解下层问题。(3)基于双层规划模型框架,建立了大型活动停车诱导信息优化模型。在停车场票价规划的基础上,探索了大型活动不完全诱导信息发布的优化方法。在原有停车场网络基础上,将停车场转化为虚拟的路段,得到一个新的扩展网络,扩展后的普通网络能有效实施最短路径算法和交通分配算法。应用诱导信息反应转移率模型描述出行对信息的反应,以此作为出行者对停车时间的估计基础,并用准动态随机用户均衡模型描述出行路径选择行为,作为双层规划模型的下层模型。建立了停车诱导信息优化的双层规划模型,并使用量子进化算法求解优化模型。(4)基于双层规划模型框架,建立了大型活动疏散网络临时单行优化模型。针对大型活动散场时交通需求不对称的特点,建立双层规划模型优化疏散网络的单行设置。采用准动态用户最优交通分配模型描述需求波动条件下的用户路径选择行为,以此作为双层规划模型的下层模型；分析了背景交通和疏散交通的出行约束,建立了单行问题的数学规划模型,以此作为双层规划模型的上层模型。设计了粒子群算法求解此类网络优化问题。(5)设计并实现了大型活动交通组织决策支持系统原型。对大型活动交通组织和管理决策支持系统进行了需求分析和模块划分,并设计了系统所需要的数据表。基于Visual C#.NET和ArcGIS Engine开发了基于GIS的决策支持系统,实现了停车票价优化模块、停车诱导信息优化模块和疏散网络的单行优化模块。更多还原

【Abstract】 During special events there are increased traffic demand and traffic flow around the stadium road network in a short period, resulting in traffic congestion and parking shortage. The temporary traffic problem can’t be solved by constructing roads and parking facilities. It is effective to develop reasonable traffic organization and management plan to relieve the traffic supply and demand conflict by using other roads and parking resources. How to develop reasonable and efficient traffic organization and management plan according to demand and supply is worth studying. In this dissertation, interaction between traffic planners and participators is considered when we develop traffic organization and management optimization model for special events. An optimization approach based on bi-level programming system was proposed. It was applied in parking fare optimization problem, parking guidance information optimization problem and evacuation network one-way optimization problems for special events. The main study works in this dissertation are as following:(1) Bi-level programming model system framework as well as its solving algorithm are researched in special events traffic organization and management. To avoid Braess strange phenomenon in traffic network organization and management process, the bi-level programming of traffic network design optimization problem is used for traffic organization and management optimization. In the model, traffic planners and travelers of optimization problems are considered as the upper model and the lower model decision-makers separately. The game process of planners and travelers is described by bi-level programming model. Quasi-dynamic traffic assignment model and quasi-dynamic stochastic user equilibrium model are used respectively to describe traveler’s route choice behavior for different optimization problems under dynamic demand circumstance. Because of Non-linear and solving difficulty characteristics of bi-level programming, advanced Particle Swarm Optimization (PSO) and Quantum Evolutionary Algorithm (QEA) are applied for upper level problem solving to get a better solution. The research has been supported by the National Basic Research Program of China (No. 2006CB705505) and National Basic Research Program of China (No.2005CB724205) (2) Special Events parking fare optimization model is established and parking demand for different parking lots is control by parking fare. Based on special events Parking-Ride/Walk multi-trip characteristics analysis, traffic network conversion is implemented. The after-parking links are converted to attractiveness of parking lot, because the travel prices of these links are relatively stable. In this way, the original parking choice problem can be described by trip distribution/traffic assignment model. Based on network conversion, Logit-based effectiveness model is applied to describe the behavior of travel and parking choice. The effectiveness model takes into account driving time, parking fees, bus fees, walk costs and other factors. This model is the lower level model of bi-level programming model. By Analysis of special events total travel time object, parking fares and parking capacity constraints, the upper level model is established. The proposed parking fare optimization model is solved by Discrete Particle Swarm Optimization (DPSO) and lower level assignment model is solved by Convex Combination Method (CCM).(3) Special events Parking Guidance Information (PGI) optimization model is established. Based on the the parking fare plan research, we studied the incomplemention guidance information optimization method, which aims to make system travel-time minimum. By considering parking lots as dummy links, the original traffic network with parking lots can be converted into a new expanded common network. In this way, the shortest path algorithm and traffic assignment can implement effectively in the network. The reaction rate of guidance information is used to describe traveler’s response of information, which is the basis of parking time prediction. The stochastic user equilibrium model is used to describe the route choice behavior and as the lower level. Bi-level programming-based PGI optimization model was proposed and the upper level is resolved by QEA.(4) Special events temporary one-way optimization model is developed. Based on the asymmetric traffic demand characteristics of special events evacuation, the traffic network temporary one-way is configured by optimization model. The route choice behavior under dynamic demand circumstance is described using Quasi-Dynamic User Optimal traffic assignment model, which is also the lower model of bi-level programming model. Based on the object and constraints of background traffic and evacuation traffic, the network one-way Mathematic Programming is established, which is also the upper level model of the bi-level programming model. An algorithm based on DPSO and Frank-Wolf (FW) is designed to solve the proposed model.(5) The special events traffic organization and management Decision Support System is developed. Based on system needs analysis and module division, the database as well as data table needed for the system are designed. Special events traffic Organization and Management Decision Support System by Visual C#. NET and ArcGIS Engine. The parking fare optimization model, PGI optimization model and network one-way optimization model are implemented in corresponding module.更多还原