需求驱动和网络环境下高速铁路列车时刻表的优化

Passenger travel demands of high-speed rails are characterized by the uneven spatial and temporal distribution, the deterministic and stochastic union, and the aggregative and heterogeneous coexistence. The train occupancy on a rail section or station is associated with a spatial-temporal exclusiveness and chronological priority characteristic. The matching and restricting relation between passenger demands and train services is captured by multiple factors, multiple purposes, uncertainty, and nonlinearity. The existed literature is mainly focused on train timetabling problems under a demand-invariant or single-corridor case. This project is to be developed in response to the fact that the high-speed rail network has being constructed and put into practice, the transport capacity has been substantially improved and the internet book has been generally implemented in China. The research is also to be supported by the immense passenger information in Big Data Time, and oriented by the optimal synchronization between passenger demands and train services. The attractive and concomitant relation between continuous demand-flow and discrete train-line are comprehensively considered by designing a reasonable time-space network based on queue delay and streamline fusion. The key problem is to design demand-driven passenger train timetables under multi-track-line interlaced and multi-level-train coexisted space-time networks. The mechanism of demand-supply match and flow-line coordination will also be revealed in this research. It is hoped to improve the underlying theory for designing train services of high-speed rail networks, and to solve the problem faced by the research and practice in Chinese high-speed railway transport organization, as well as to supply better transport services to the community.

高速铁路旅客需求具有时空分布不均衡、确定与随机叠加、聚集与异质共存的属性，列车对轨道区间或车站的占用具有时空排他和时序优先的特征；旅客需求与列车服务之间，构成了多因素、多目标、不确定、非线性的匹配与制约关系。国际学术界关于列车时刻表问题的研究，主要集中在单条线路的环境和需求固定的情形。本项目以我国高速铁路大规模建设并投入运营、旅客运输能力大幅度提高、网上购票普遍推行为背景，以“大数据”时代海量客流信息为支撑，以旅客需求与列车服务的最佳匹配为导向，通过构建基于排队延误和流线融合的时空网络，综合考虑连续性旅客流和离散性列车运行线之间的吸引及共存关系，系统地研究多线路交错与多级别列车共存的时空网络中、基于需求驱动的列车时刻表编制理论与优化方法，从本质上揭示轨道交通供需耦合及流线协同的机理，完善高速铁路旅客列车服务方案设计理论，提高我国高速铁路运输管理水平，向社会提供更加优质的运输服务。

列车时刻表设计，是一类具有广泛应用背景的调度优化问题，也是轨道运营管理领域最具挑战性的科学问题，其中加入需求因素并嵌入旅客行为，使问题变得异常复杂。针对不同的网络环境和时变需求，课题组深入分析了连续性旅客流和离散性列车运行线之间的吸引及共存关系，构建了基于流线融合的时空网络，通过数学构模、算法设计、数值计算和结果分析等环节，系统研究了需求驱动列车时刻表编制理论与优化方法，取得的主要成果有：1）构建了越行环境下列车时刻表问题的线性整数规划模型，将基于小时的客流需求加载到停站灵活的列车群上，加载过程同步绑定OD需求的起始站和终到站，使得客流需求加载约束对应的对偶变量依赖于两个相关联的车站，在列生成计算框架下，提出了替换单车站对偶变量的方法，以应对标准动态规划算法无法处置价格子问题的挑战。2）现有研究列车时刻表问题的文献，通常为每个列车在始发站预先指定狭窄的出发时间范围，实践中这样的设置非常困难和费力，提出了基于小时时间窗的柔性列车时刻表以解决面临的困局，设计了含有优先权搜索的交替方向乘子算法。3）介绍了如何将动车组调度转化为多车场公交车辆调度问题，为了消除来自相同或相近地点的车辆争抢执行同一任务引起的对称性，特别关注了车辆调度中指派和路径决策的协同问题，设计了结构更加紧凑的时空连接网络以降低数学模型的复杂度，构建了变量分离方法以利用任务绑定且车辆可辨的拉格朗日乘子，提出了基于有序指派的方法以提高增强模型的求解质量。4）以列车衔接次数和换乘等待时间为目标函数，构建了网络环境下列车时刻表协同的双目标优化模型，提出了基于序列性搜索的启发式算法。5）利用多维时空网络研究了新增列车运行线问题，提出了拉格朗日松弛求解方法；探究了基于旅客出行偏好的列车时刻表优化问题，设计了嵌入用户平衡的双层求解方法。6）研究了时变需求下铁路集装箱班列时刻表优化问题，提出求解模型的Benders分解算法。