交叉路口交通量預測模式之研究

Abstract

隨著大眾生活水準的提高，私人運具數量不斷上升；另一方面，由於現代社會的分秒必爭，駕駛者往往希望能夠在最有效的時間裡，完成其旅次。為了協助上述的目標，其中一個重要關鍵，則是短期交叉路口交通量預測，因此，短期交通量預測能力的好壞，成了影響交通未來發展的重要因素之一。 有關交通量的預測有極大量的研究。特別是在過去十年中，有許多不同的方法論和技術針對此課題作研究。本研究之目的期能改進短期交通量模式的預測能力。目前國內針對短期交通量預測的研究，大多相信交通量變化型態乃是沒有規律之非線性系統，此乃由於國內外使用時間序列分析法作為預測交通量模式，其「時階設定」(time-intervals)大多採用某一特定時階，如：5分鐘、15分鐘、30分鐘。 但實際上，交通量的多寡應和號誌時制之內容有關，本研究將採用時間序列分析法，將時階設定為號誌週期之整數倍，以探討交通量變動之規律性。易言之，本研究將針對「時階設定」和「預測的基礎數據」兩項，構建交通量預測模式，期能找出交叉路口交通量預測之適當方法。

Because growing of living standards grows, the number of private automobile increase at the same time. Also, in recent days, “time” is money, all drivers wish to arrive the destination as fast as possible. In order to research the goal, the short-term forecasting is a key point. The forecasting ability becomes the main factor of the future for transportation. There are a lot of studies for traffic volume forecasting, especially the pass decade. There are variety methodologies and technologies aiming at this issue. In this research, it is expected to improve the traffic forecasting ability. Now, most internal research of short-term forecasting believe that the change of traffic volume is non-linear. And most research use the fixed “time step”. Such as: five minutes, ten minutes, fifteen minutes. However, the volume of traffic should have relations with the control of traffic light. In this research, I will use the time-series models, and set the time interval as the integer times of traffic light in order to understand the regular pattern of traffic volume.