中断条件下高铁列车运行调整优化模型与算法研究  

Optimization model and algorithm for high-speed railway train operation adjustment under disruptions conditions

作　　者：赵文强 周磊山[1] 白紫熙 韩昌 ZHAO Wenqiang;ZHOU Leishan;BAI Zixi;HAN Chang(School of Traffic and Transportation,Beijing Jiaotong University,Beijing 100044,China;School of Logistics,Beijing Wuzi University,Beijing 101149,China)

机构地区：[1]北京交通大学交通运输学院,北京100044 [2]北京物资学院物流学院,北京101149

出　　处：《铁道科学与工程学报》2025年第3期979-990,共12页Journal of Railway Science and Engineering

基　　金：北京市自然科学基金-丰台轨道交通前沿研究联合基金资助项目(L231026);北京市教育委员会科研计划项目(KM202410037001)。

摘　　要：高速铁路运营过程中,各种突发事件的发生可能导致列车通行中断,为解决该问题,提出一种结合日常运输组织工作中常用调度手段的一体化调整优化方法,包括取消列车、短编组列车重联、启用热备车等调度手段,在考虑列车运行图基本约束的基础上,增加考虑车底运用约束与旅客需求约束,构建列车运行图与动车组运用的一体化调整优化模型,以实现在运用调度手段优化列车运行图的同时,兼顾对动车组交路计划的考虑。在此基础上,设计了深度学习驱动的智能遗传优化算法,改进交叉变异的选择模式,经过深度学习得出在不同输入条件下的最佳的交叉变异选择方案,从而提高了遗传算法的质量和效率。为了验证模型和算法的有效性和实用性,以京沪高铁全线的23个车站为基础设计了一组实验,以120min的中断时长为例,算法求解耗时1971s,求得加权后总目标函数为12151079。通过求解结果分析,与传统的遗传算法相比,该算法的求解结果在旅客总晚点时长方面降低了11.7%,耗时降低了10.8%。随后通过设置不同的中断区间,对不同中断条件下的列车运行调整进行分析,并设置不同的调度手段组合,对比分析不同调度手段组合在目标函数优化效果和求解时间上的差异,为铁路运营部门提供不同的决策方案。以上研究结果表明,深度学习驱动的智能遗传优化算法可以快速且有效地优化列车通行中断问题,为调度人员提供优质可行的解决方案,保证铁路资源的有效利用。To address the issue of train service disruptions caused by unexpected incidents during high-speed railway(HSR)operations,an integrated adjustment optimization method combining common scheduling strategies was proposed.These strategies include train cancellations,coupling of short trains,and the deployment of standby train-sets.In addition to considering the basic constraints of the train timetable,the method incorporated constraints related to train-set utilization and passenger demand.This could lead to the construction of an integrated adjustment optimization model for both the train timetable and train-set circulation,aiming to optimize the train timetable while also accounting for train-set scheduling.On this basis,a genetic algorithm enhanced by deep learning was designed,improving the selection mechanism for crossover and mutation.Deep learning was used to derive the optimal crossover and mutation selection strategy under different input conditions,thereby enhancing the quality and efficiency of the genetic algorithm.An experiment involving all 23 stations on the Beijing-Shanghai HSR was conducted to validate the effectiveness and practicality of the proposed model and algorithm.Taking the 120 minutes’disruption duration as an example,the algorithm took 1971 seconds to solve,and the weighted objective function was calculated to be 12151079.Compared to the traditional genetic algorithm,the proposed algorithm reduced the total passenger delay time by 11.7% and the computational time by 10.8%.This article set different disruption duration and analyzed the adjustment of train operation under different disruption conditions subsequently,and set different combinations of scheduling methods to compare and analyze the differences of these methods in the optimization effect on the objective function and solving time,providing different optional solutions for railway operation departments.The results suggest that the genetic algorithm enhanced by deep learning can efficiently and effectively optimize train service

关 键 词：高速铁路 列车运行调整 列车运行图 深度学习 改进的遗传算法 

分 类 号：U292.4[交通运输工程—交通运输规划与管理]