透過您的圖書館登入
IP:3.146.176.254
  • 學位論文

利用動態理論預測高速公路之車輛平均速度

Mean Velocity Forecast for Highway System Using Kinematic Theory

指導教授 : 連豊力

摘要


智慧型運輸系統中,先進交通管理及資訊系統之動態交通控制,唯有依賴可描述系統狀態的監視技術,自動蒐集交通狀況資料,並具良好的預測能力,才能預先處理系統的改變,進一步修正控制策略,以達到運輸系統能更有效率的運作。可見交通預測能力的良窳是ITS成功與否之重要關鍵,而交通資訊預測則是一個重要的課題。 由於交通資訊會因交通組成、車道分佈、道路分類、路線設計、交通管制方式、道路容量、天候、氣溫、環境地形等因素有所影響,要加以精確預測並不容易。因此,已有許多研究利用時間序列分析、迴歸分析、類神經網路等方法探討此一課題。然而,現有技術運用在交通預測皆著重在單一地點的預測,不同地點之間的交通關係較少討論,導致高速公路系統架構與車流移動行為在預測過程中被忽略。 本論文提出一個以動態理論為概念的交通預測方法。此預測方法分析了車流分佈狀況,及車流與車速之間的真實關係,利用道路上車輛在時間的變化下其在空間中移動的情形,先後預估未來車流與車速的資訊。本研究採用宏觀的交通參數,結合空間與時間的交通變化,提供一種簡易且具有交通特性的預測方法。最後,以模擬方式呈現此預測模式的結果。

並列摘要


Traffic forecast is one of fundamental abilities in intelligent transportation system (ITS) and advance traffic management system (ATMS). Accurate traffic information prediction is also crucial to modify traffic management strategy before traffic state change. Existing traffic forecast techniques use traffic information concerned only at single location without neighborhoods. Highway framework and traffic stream behavior are usually ignored in traffic forecast. This thesis proposes the concept of kinematic theory for a traffic forecast model. The model can be used to analyze traffic stream distribution and realistic relation between flow and velocity, and then observe vehicle motion in space by time variation to predict flow and velocity in sequence. This study adopts macroscopic traffic parameters and combines traffic variation in time and space to provide a simple forecast method with traffic property. Furthermore, some simulations demonstrate the predicted result for the forecast model.

參考文獻


[22: ITS Taiwan]
[1: Jou et al. 2003]
[2: Voort et al. 1996]
M. Van Der Voort, M. Dougherty, and S. Watson, “Combining Kohonen Maps with ARIMA Time Series Models to Forecast Traffic Flow,” Transportation Research Part C: Emerging Technologies, Vol. 4, No. 5, pp. 307-318, 1996
[3: Yang 2005]

延伸閱讀