| 摘要: | 高速列車的管理工作包含例行維修(如車輛迴轉以清理)與災害發生時之應變措 施等，由於高速列車中上述管理工作非常重要，台灣高鐵公司雖然已有相關列車班表模型，但是此模型尚未完整，因此，基於相關的文獻回顧，其他學者並沒有針對環狀的列車系統，例如列車的迴轉運作等，這些運作在台灣高鐵是非常重要的。本研究根據台灣高鐵的相關需求與運作方式提出一個最佳化排班模式，結合整數與動態規劃並使用CPLEX 軟體求其最佳解，當鐵道發生災害時，高鐵單位會對鐵路系統重新排程，所以本研究的模式亦支援重新排程的最佳化。本研究應用敏感度分析針對不同的參數觀察目標值的影響，參考台灣高鐵的列車資料，使用兩線軌道、128 服務、29 輛列車與8 個車站，經由實驗結果表示，本研究所求得的最佳解與高鐵公司的時刻表一致(程式執行時間為0.10 秒)，因此，此模式可以應用模擬並分析相關的列車系統，提供災害發生時即時的決策支援。 Managing circulation of trains, including regular inspection, car cleaning times and turning back operations, has become important due to the scarcity of railway company resources. The Taiwan High-Speed Railway (THSR) already has cyclic patterns of daily train circulation, but these patterns have not been modeled yet. Moreover, based on a review of the literature, researchers in the railway field have never considered train circulation, especially in HSR systems, even though it is important. This research proposes a scheduling optimization model that has the capability to accommodate not only basic requirements such as railway topology, traffic rules, and user requirements, but also train circulation as well. Mixed integer and dynamic programming have been chosen to solve the model under CPLEX. In addition, railway systems are often characterized by high traffic density and heterogeneous traffic that is sensitive to disturbances; thus, rescheduling activity for updating an existing schedule in response to disruptions is needed. This research has applied sensitivity analysis in order to identify how disturbances propagate in the original timetable and which actions to decide in order to mitigate the impact instead of cancelling many trains. Assumptions as well as input and output values are configured by using real data from THSR,which used two lines, 128 services, 29 trains, and eight stations. The model has obtained a timetable result as good as the real timetable in a short computation time (that is, 0.10 second). Sensitivity analysis results could determine critical infrastructure and parameters that are sensitive to disturbances. Therefore, it could be a good simulation analysis for predicting the effect of disruptions on the timetable without doing real experiments such as trains being disordered and overtaken. |
