路徑旅行時間為駕駛者路徑選擇之重要資訊,國內業務主管單位受預算經費之限制,替代道路上無法全面佈設偵測器;目前國內外應用偵測器資料對旅行時間推估,多以高速公路為研究對象,主要因為高速公路的車流行為較為單純;在號誌幹道系統上,車流另外受號誌之干擾而停等,產生延滯的情況與高速公路環境顯著不同;因此過去研究以偵測器資料為主之高速公路理論模式,並無法適用於號誌化之幹道系統,需要大幅的修正。 本研究有鑑於此,在偵測器佈設不足之前題下,以倒推起迄車流量之概念,架構路徑上各路段之相互關係,並利用遞迴演算的卡門濾波(Kalman Filter)模式作為基礎,以即時偵測器蒐集之資料為檢核點之限制,嘗試建立出在資訊不完全(Incomplete)條件下之路徑旅行時間預測模式,並以省道台1線及台15線為範例路網進行分析,同時討論偵測器密度、飽和度、事件與資料融合等不同情境下動態路徑旅行時間的預測績效。 Dynamic travel time forecast on an alternate path is recognized as valuable information for drivers to switch routes. However, insufficient installment of detectors especially on arterials is usually suggested by local government due to the limited budget. Therefore, model that can solve problems of dynamic travel time forecasting on alternate paths with only incomplete detector data is asked with great attention. This paper suggests dynamic origin-destination matrices estimation approach to construct the link flow dynamic relationships, and applies adaptive Kalman filter process to adjust the error. Finally, a case study of alternate path travel time information is provided with deliberating experimental design to illustrate the performance of suggested approach under different levels of the layout density of vehicle detector installation, saturation flow and with/without the incident and data fusion.