捷運鐵路會因設備故障等一些偶發事件，造成某列車之主要延滯，接著影響後續列車的運行，形成延滯擴散現象，對旅客與鐵路營運人員帶來極大的不便，因此事件發生後之行車調度極為重要。行車調度包括改變列車運行時刻、車輛指派、司機員指派、及軌道使用。實務上，調度人員通常是在即時狀態下執行調度作業，而其依據則為行車作業手冊內所規範的相關規則。為了評估實務策略，並發展有效的調度規則，本研究構建捷運系統運轉模擬模式，來模擬各種運轉整理策略的績效。
捷運系統運轉模擬模式之主要輸入資料為計畫時刻表、車輛使用計畫、系統運行參數、以及各時段旅客起迄需求量資料。模式中包含列車流模擬、旅客流模擬、與行車調度模擬，三者間之關係必須界定清楚，並在模式中適當加以連接。若以傳統程序導向方式來構建模擬模式，程式相當複雜，而且不易擴充。本研究以物件導向技術來構建模式，以統一塑模語言(UML)作為物件分析(OOA)與設計(OOD)工具。首先，使用各種圖形，例如類別圖、活動圖、循序圖等，來描述系統內車輛、時刻表、車站、旅客等元素之間的相互關係。其次，於設計階段充分運用封裝、繼承、與多型等觀念，以利實作階段能有效重複使用程式碼，並提昇系統開發效率。最後，以物件程式語言(OOP)C++完成模式實作。
本研究探討之運轉整理策略包括：月台趕點、站間趕點、抓前車、抓後車、刪除列車、與雙人乘務等策略。系統績效指標包括旅客觀點之平均旅行時間，以及行車調度人員觀點之列車平均延滯時間。本研究並於開發的模式中，加入最佳化模組，可以對模式關鍵影響因子進行最佳化搜尋。
個案研究首先針對台北捷運淡水-新店線之單一來回運行路線進行模擬分析，以直接搜尋法 – Hooke-Jeeves演算法，搜尋延滯事件發生後執行間距控制之最佳抓車策略，並測試上午尖峰於不同時間、地點、發生不同長度的主要延滯事件，評估實施各種運轉整理策略的運行績效。其後，針對淡水、新店、中和三條路線構成之「Y字型路網」進行營運班表績效分析。
個案研究之模擬實驗結果發現：(1)延滯事件發生後，使用運轉整理策略，確實可以改善系統績效，使用單一策略以站間趕點效果較佳。(2)主延滯時間5分鐘，單純實施月台趕點與站間趕點策略，即可追回旅客因延滯所增加的旅行時間。(3)延滯事件發生後，抓前車搭配趕點策略，較抓後車效果為佳。(4)當主要延滯事件發生於行車方向後端車站，且延滯時間長度達10分鐘以上，以實施雙人乘務策略為最佳方案。(5)使用刪除列車策略，列車總延滯時間雖可縮減，但旅客平均總旅行時間則大幅增加。本研究之結論是，使用混合策略較單一策略效果為佳，從旅客績效指標觀點，抓前車是有效的運轉整理策略，由營運者觀點，則刪除列車是有效策略。

Under an abnormal condition, dispatching or rescheduling operations are centrally important for a rail transit to maintain its service reliability, because a primary delay may eventually cause severe consequent delay. Dispatching operations generally include the changes of train movements, vehicle assignments, crew assignments, and track utilization. In practice, dispatchers make dispatching plans in real time situation, in accordance with the guideline in the Rule Book. In order to evaluate practical strategies and develop effective rules, a simulation model was developed to evaluate the performance of various rescheduling strategies.
System operation plans, passenger demand, track layout, and technology parameters are the input data of the simulation model. Simulations of train flow, passenger flow, and traffic control operations are the three main elements of the model. In order to eliminate the disadvantages of model built with traditional procedural programming techniques, we propose an object-oriented approach and construct a simulation model of metro operation. The unified modeling language (UML) is used as a tool for object-oriented analysis and design. Firstly, class diagram, activity diagram, and sequential diagrams are used to depict the relationship among the elements of a system, and some of them are listed as follows: network, vehicles, time schedules, stations, and passengers. Secondly, in the stage of system design, this research uses the concepts of encapsulation, inheritance, and polymorphism in OOP (object-oriented programming) to add the reusability of software codes and improve the efficiency of the developed system. Thirdly, the system is implemented with OOP by C++ language.
The rescheduling strategies discussed in this study include speed or running time control, dwell time control, holding the preceding trains and/or following trains, cancellation of delayed trains, and double ending. The performance indices, from both passengers’ and operators’ viewpoints, are used in the study. They are: passenger travel time and the duration of train delay. The optimalization module is brought into the developed model to carry out optimal searching for key factors that mainly influence the MRT system.
The case study first pinpoints on only Tamsui line to carry out simulation analysis, adopting the direct searching method – Hooke and Jeeves’ algorithm to find out the most effective strategy of conducting headway control to deal with delay. And it tests the performance of different rescheduling strategies under different time frames, locations, and delays during peak hour in the morning to evaluate the effectiveness of rescheduling operations. After that, the case study takes the Y-type MRT network, consisting of Tamshui line, Hsintien line and Chungho line as materials to evaluate the effectiveness of operating schedule.
The results of the research are as follows：(1) Rescheduling strategies definitely improve the operation when delay occurs. Running time control outperforms other strategies in adopting only a single method. (2) Given the delay is 5 minutes, implement of dwell time control and running time control will be sufficient for catching up the increased travel time caused by the delay. (3) When delay occurs, the two methods mentioned above can be enhanced better when accompanied by holding preceding trains than holding following trains. (4) Implement of double ending would be the best approach when primary delay takes place in the later part of the operation route and the time delayed is over 10 minutes. (5) Cancellation of the delayed trains can reduce the total time delayed, but the average travel time of passengers would intensively increase.
We can conclude that the holding strategy is effective only for the passenger’s indices while cancellation for the operator’s. With regard to passenger’s travel time, to use compound strategies is generally better than to use an individual strategy.

A.期刊論文
1. 李治綱、陳朝輝、簡聰裕，民國91 年6 月，「捷運鐵路列車延滯事件發生後行車調度
策略之模擬分析」，運輸計畫季刊，第三十一卷第二期，頁299-322。
2. Lee, C.K. and Chen, C.H., “Driver Scheduling and Rostering of Taiwan Railway
Administration”, Journal of the Eastern Asia Society for Transportation Studies, Vol.5,
pp292-306, October, 2003.
3. Chen, C.H., Lee, C.K. and Chang, E.F., “Simulation Analysis on the Dispatching Operation
of Rapid Rail Transit”, Journal of the Eastern Asia Society for Transportation Studies,
Vol.5, pp323-338, October, 2003.
4. 陳朝輝、李治綱，「應用物件導向技術於捷運系統運轉模擬之模式構建」，運輸學刊(已
接受)
附錄-58
B.研討會論文
1. 李治綱、陳朝輝、簡聰裕，「捷運鐵路運轉整理之模擬研究」，新世紀軌道運輸學術研
討會論文集，民國89 年10 月，本文獲得該研討會論文獎。
2. 李治綱、陳朝輝、郭彥秀，「台鐵司機員排班與輪班問題之研究」，第五屆運輸網路研
討會論文集，民國89 年10 月。
3. 李治綱、陳朝輝、簡聰裕，「都市鐵路運轉整理之研究」，中華民國運輸學會第十五屆
學術研討會論文集，民國89 年12 月。
4. Lee, C.K. and Chen, C.H., “Simulation Analysis on the Rescheduling Operation of a
Transit Line”, 9th World Conference on Transportation Research, 2001.(第九屆世界運輸
研究研討會，獲國科會機票補助至韓國漢城發表)
5. 李治綱、黃範哲、陳朝輝，「鐵路行車與車輛調度之模擬分析」，中華民國第十六屆運
輸年會及學術研討會，民國90 年11 月。
6. Lee, C.K. and Chen, C.H., “A Simulation model for Metro’s Rescheduling Operation”，台
北捷運博覽會，民國91 年4 月。
7. 李治綱、陳朝輝、張恩輔，「捷運系統運轉整理之模擬分析」，中華民國運輸學會第十
七屆學術研討會論文集，民國91 年12 月。
8. Lee, C.K. and Chen, C.H., “Simulation Analysis on the Rescheduling Operation of a
Transit Line”, World Transport Research - Selected Proceedings of the 9th World
Conference on Transport Research (July 22-27, 2001, Seoul, Korea), Pergamon, CD-Rom
edition, Nov 4, 2003.
9. 陳朝輝、李治綱，「捷運系統延滯事件發生後行車調度策略最佳化之模擬分析」，中華
民國運輸學會第21 屆學術研討會論文集，民國95 年12 月。
C.國科會專題研究報告
1. 陳朝輝（2002-2003），「捷運網路模擬與運轉整理分析之研究」，國科會專題研究報告
（編號：91-2211-E-035-030）。
2. 陳朝輝（2003-2004），「行車調度最佳化之研究」，國科會專題研究報告（編號：
NSC92-2211-E-035-027）。
3. 陳朝輝（2004-2005），「應用物件導向模式於捷運網路運轉模擬分析」，國科會專題研
究報告（編號：NSC93-2211-E-035-021）。