移動閉塞區間制下混合派翠網路應用於捷運列車運轉整理策略之模擬研究

A Study on Simulation of Rapid Transit Dispatching and Operation Strategies with Moving Block System by Using Hybrid Petri Nets

林琮欽(Tsung-Chin Lin)

指導教授 ： 陶治中

淡江大學/管理學院/運輸管理學系碩士班/碩士(2009年)

https://doi.org/10.6846/TKU.2009.00711

摘要

臺北捷運木柵延伸線將於98年6月底正式通車，成為國內軌道運輸系統使用移動閉塞行車制之先例。在過去制度中，列車一般由軌道之閉塞區間進行定位，無法精確得知列車所處之位置，造成軌道空間利用上之浪費。本研究旨在建立一套能反應移動閉塞行車制度之模擬模式，供營運者在不需以實體列車測試下進行策略分析，進而作為營運上之參考。

本研究以移動閉塞行車制度為前提，建構包含旅客、列車、營運者之模擬模式，以充分反應各層面之互動。過去捷運列車調度之模擬模式大致可分為不連續事件模擬以及連續性時間模擬。不連續事件模擬有著快速求解、處理便利之優勢，然其無法描述系統細部模組互動。本研究考量到移動閉塞行車制下車與車互相影響之特性，決定以連續性時間模擬作為模式構建之依據。並基於模糊理論能反應調度措施對列車影響之優勢，選擇以模糊跟車方法取代傳統行車動力學。

鑒於派翠網路能傳達多種資訊之優勢。本研究將模糊跟車規則之推論過程以模糊派翠網路進行呈現，再以時間派翠網路架構車站內旅客之上下車行為，並將兩者混合以描述列車運行與停站時旅客行為之切換。同時營運者可分別透過改變模糊派翠網路之模糊指標以及時間派翠網路之時間區間影響列車與旅客之行為。

本研究以臺北捷運木柵線與內湖線(包含木柵線區段)作為模擬對象，探討在引進移動閉塞行車制後可能之運轉策略、最短發車班距、延誤發生時採行之調度策以及月台旅客過多造成車站壅塞時之反應策略之績效。

研究結果發現：
1.捷運公司實測之最短班距72秒，經模擬驗證後認為應可再下修至68秒。透過本研究提出之營運模式，能在有條件之狀況下進一步縮短至61秒。
2.延誤發生後之班距回復時間受發車班距長短、時刻表調整門檻、時刻表調整程度、車站駐車時間等多方因素影響。
3.提出旅客適應性之調度策略，能在不縮短發車班距之情形改善旅客候車時間。

關鍵字

移動自動閉塞 ； 派翠網路 ； 模擬模式

並列摘要

Taipei Nei-Hu Rapid Transit Line applying the first Moving Automatic-Blocking System (MAS) in Taiwan has been officially in operation since 4th July 2009. Comparing with MAS, current train control systems using Fixed Automatic-Blocking System (FAS) result in problems of insufficient railway capacity. This study aimed at developing a simulation model to conduct several scenario surveys on operational strategies for Taipei Nei-Hu Rapid Transit Line.

Based on Petri Nets approach, a simulation model considering interactions among passengers, trains and the rapid transit operator was developed. Referring to literature review, there are three types of simulation models of train dispatching for urban rapid transit systems – discrete event, continuous time and the hybrid simulation. Discrete event simulation models have the advantages of quick solution and convenience of handling; however, it could not describe the detailed interaction of models. Considering continuous train positioning by using MAS, continuous time simulation model was chosen. Besides, considering the influence of dispatching measures on train, fuzzy theory was used to depict train moving behavior instead of using traditional train movement dynamics.

In addition, Petri Nets can transmit different kinds of information, the inference process of fuzzy train moving rule by Fuzzy Petri Nets (FPN) was used. Time Petri Nets (TPN) was used to describe passengers’ behavior of getting on and off within the stations. These both approaches were used to describe passengers’ change in behavior as trains move and stop at stations. Besides, operators could influence trains and passengers’ behavior by changing the fuzzy index of FPN and the time interval of TPN.

Taipei Mucha Line with Neihu Line was taken as the empirical study to investigate possible operation strategies such as rescheduling, the shortest headway, performance of dispatching when delays occur and responsive strategies toward stations with crowded passengers.

The results of this study revealed:
1.The shortest headway in Taipei Nei-Hu Rapid Transit Line field trial was 72 seconds, however, with numerous simulation tests it could be reduced to 68 seconds.
2.The restoring time when delays occur was influenced by factors such as the duration of headway, threshold in adjusting timetable, adjustment extent of timetable and duration that trains stop at stations.
3.A dispatching strategy of passengers’ adaption which could reduce the time of waiting without reducing train headway was suggested in this study.

並列關鍵字

Moving Auto-blocked System(MAS) ； Petri nets ； Simulation Model

參考文獻

9. 劉嵩瀚(2006)，「不同行車制度下台鐵區段容量分析模擬模式之研究」，淡江大學運輸科學研究所碩士論文，民國95年。

5. 李治綱、陳朝輝、簡聰裕(2002)，「捷運鐵路列車延滯事件發生後行車調度策略之模擬分析」，運輸計劃季刊，第31卷，第2期，頁299-321，民國91年。

12. 謝興盛(2003)，「捷運列車延誤時班距調整模式之模擬分析－以台北捷運中、高運量系統為例」，國立成功大學交通管理科學研究所博士論文，民國92年。

6. 張恩輔(2002)，「捷運系統運轉整理之模擬分析」，國立成功大學交通管理科學研究所碩士論文，民國91年。

13. Cai, G., Zhao, S., Jia, L., Ye, Y., Li, X. (2008), “Approach of Train Group Operation Conflict Prevention Reasoning Based on Fuzzy Time Petri Nets,” Networked Computing and Advanced information Management, 2008, Vol. 2, pp. 272-277