Optimal multi-class rescheduling of railway traffic
Optimal multi-class rescheduling of railway traffic
复制标题
DOI：
10.1016/j.jrtpm.2011.06.001
复制
发表时间：
2011-11
期刊：
影响因子：
--
通讯作者：
F. Corman;A. D’Ariano;I. Hansen;D. Pacciarelli
中科院分区：
文献类型：
--
作者：
F. Corman;A. D’Ariano;I. Hansen;D. Pacciarelli
铁路调度员实时重新安排列车，以限制干扰的传播，并通过最小化与离线时间表的偏差来调节各自调度区域的交通。然而，在一个地区作出的决定可能会影响其他地区列车时刻表的质量，甚至可行性。区域控制中心协调调度员在多个区域的工作，以在全球范围内调节交通，避免全球不可行的情况。从调度员问题出发，研究了多个调度员之间的协调问题，目的是使调度员的行为趋向全局最优解。在我们的模型中，协调员可以在每个调度区域的边界处施加约束。然后，每个调度员必须通过产生符合协调员施加的边界约束的本地可行的解决方案来调度其区域内的列车。因此，协调器所面临的问题是一个双层规划问题，其中由协调器控制的变量是边界约束。我们证明了协调器问题可以用一个分支定界过程求解到最优。协调算法已在荷兰的一个大型真实的铁路网络上进行了测试，具有忙碌的交通条件。我们的实验结果表明，一个经过验证的最佳解决方案是经常发现的各种网络划分的计算时间内兼容的实时操作。
Railway dispatchers reschedule trains in real-time in order to limit the propagation of disturbances and to regulate traffic in their respective dispatching areas by minimizing the deviation from the off-line timetable. However, the decisions taken in one area may influence the quality and even the feasibility of train schedules in the other areas. Regional control centers coordinate the dispatchers’ work for multiple areas in order to regulate traffic at the global level and to avoid situations of global infeasibility. Differently from the dispatcher problem, the coordination activity of regional control centers is still underinvestigated, even if this activity is a key factor for effective traffic management.This paper studies the problem of coordinating several dispatchers with the objective of driving their behavior towards globally optimal solutions. With our model, a coordinator may impose constraints at the border of each dispatching area. Each dispatcher must then schedule trains in its area by producing a locally feasible solution compliant with the border constraints imposed by the coordinator. The problem faced by the coordinator is therefore a bilevel programming problem in which the variables controlled by the coordinator are the border constraints. We demonstrate that the coordinator problem can be solved to optimality with a branch and bound procedure. The coordination algorithm has been tested on a large real railway network in the Netherlands with busy traffic conditions. Our experimental results show that a proven optimal solution is frequently found for various network divisions within computation times compatible with real-time operations.