演化式幹道號誌連鎖策略之研究

Evolutionary Signal Coordination Algorithm for Arterial Traffic Control

楊祖維(Tsu-Wei Yang)

指導教授 ： 張堂賢

國立臺灣大學/工學院/土木工程學研究所/碩士(2008年)

https://doi.org/10.6342/NTU.2008.02559

摘要

幹道系統為構成都市交通路網之重要部分，故交通工程師於解決都市交通路網之問題，常藉由經驗法則將網路系統劃分為幹道系統或獨立路口，再進行各子系統最佳化號誌時制運算，最後進行整體路口號誌時制之整合。然目前國內外號誌時制軟體與許多動態號誌控制系統，大多將所輸入之幹道系統設定為相同群組，並調整群組內各路口週期為相同或整數倍，卻忽略幹道系統內各路口交通型態之差異，對於時制設計過程中最重要、最根本的群組劃分問題，鮮少深入探討。
本研究重新考慮群組劃分之因素，除了簡化具有相異週期之相鄰路口所面臨到的時差遞移問題，並能減少幹道系統過多的延滯。由於群組劃分後各子系統路口時差以「絕對時差」進行設計，將無法直接進行相異週期之群組間路口時差最佳化求解，本研究遂以「相對時差」進行各路口間時差設定，並推導一「續進路口數期望值」模式，以改善傳統綠燈帶寬法無法依據各路口續進方向之交通需求量，有效分配各路口雙向續進能力，且無法由綠燈帶寬得知系統內車輛真實的續進能力之問題。
本研究設計一套可動態輸入交通參數以產製號誌最佳化時制之系統，為了避免演算過程落入區域最佳解之情形，且符合動態號誌控制系統產製時制計畫之效率，本研究以基因演算法為模式求解運算核心，依據各子系統之特性設計合適的基因演算架構，以延滯最小為目標運算各路口最佳週期、時比，並利用本研究推導之續進路口數期望值為基準進行群組劃分與最佳化時差設計，最後即時運算產製幹道系統最佳化時制計畫。
最後本系統與實務上常使用之最佳化號誌時制軟體Synchro進行比較分析，分析結果顯示本系統所產製之時制計畫，各路口平均延滯值與Synchro軟體各有優劣，但在續進路口數期望值指標上，本系統皆比Synchro軟體所產製之結果較佳，且本系統可藉由動態交通參數輸入即時產製最佳化時制計畫，以改善Synchro軟體僅能離線運算之缺失。

關鍵字

號誌控制 ； 群組劃分 ； 相對時差 ； 續進路口數期望值 ； 基因演算法

並列摘要

The arterial system is the important part of the traffic network in the urban, so the traffic engineers often divide the network system into the isolated intersection or arterial system with the experience rule. Then, make optimal design at all intersection iming plan in the subsystem. Finally, combine and adjustment the timing plan of all intersections in the system. However, many of the optimization timing design software or dynamic timing control research are neglect the different of traffic type of every intersection in the system. They are not deep research the most important system partition problem in the timing design process.
If the intersections of every subsystem are designed the offset by absolute offset after system partition, this method is unable to design the both intersections offset among difference groups. This research design each intersection offset by relative offset, and derive a “the expected value of progressive intersection number model”. This research is for the sake of improving the traditional bandwidth method which is unable to accord with progression demand of every intersection to assign the progression ability effectively, and know the vehicles’ real progression ability in the arterial system.
This research designs a programming system that can input traffic parameter and produce the optimal timing plan dynamically, in order to accord with the efficiency of the dynamic timing control system produces the optimal timing plan, and avoid the calculation result is local extreme value. This paper adopts genetic algorithms as the operation core in this programming system. The suitable genetic algorithms structure is made according to every characteristic of subsystem. Base on minimum delay produce each intersection optimal cycle time and split time. According to the progress indicator, to divide system and design the optimal offset. Finally, dynamic produce the optimal timing plan of the arterial system.

並列關鍵字

timing control ； system partition ； relative offset ； the expected value of progressive intersection number ； genetic algorithms

參考文獻

14. 朱小玲（2005），「交控任務邏輯及績效評估之研究」，台灣大學，碩士論文。

23. 馮正民、邱裕鈞（2004），「研究分析方法」，建都文化事業股份有限公司。

1. Allsop, R.E. （1971）,”A diagram showing practicable settings for a fixed-time traffic signal and their effect on delay”, Transportation Research, Vol.5, pp.59-65

3. Chang, T.H. and Sun, G.Y.,（2004）, “Modeling and Optimization of an Oversaturated Signalized Network”, Transportation Research Part B 38，pp.687-707.

4. de Oliveira, D. E., Bazzan, A. L. C., & Lesser, V. (2005), “ Using cooperative mediation to coordinate traffic lights: A case study”, AAMAS 2005, pp.463-470