交通車流具有動態變化特性，且交通資訊為用路人行為決策之重要參考依據，可提供駕駛者選擇適當之路徑與出發時間，以避免交通擁擠，並以最短之旅行時間到達目的地，隨著先進旅行者資訊系統之出現與進步，短路段與運輸走廊之旅行時間亦越形重要，而利用即時之交通資料預測未來旅行時間，則是未來先進旅行者資訊系統不可或缺之交通資訊。 本研究分為兩階段進行，第一階段利用模擬的方式，產生交通資料，作為資料之產生器，並且作為研究最後之旅行時間之驗證部分，第二階段則是採用三層、完全連結及前向式的網路架構，配合倒傳遞演算法來建立不同交通車流型態下之旅行時間預測模式，期望能透過偵測器所偵測之交通資料，提供精準之旅行時間預測，以作為用路人路徑選擇或是出發時間決策判斷之依據。 經由反覆的校估與測試，由研究結果得知，本研究所構建旅行時間預測模式，預測效果良好，於高速公路旅行時間預測方面，可提供交通相關單位預測旅行時間參考之雛形。 Traffic flow are provided with dynamic variation characteristic, and traffic information is an important consultation when drivers have to make a decision. Traffic information will allow drivers to select appropriate routes and departure time to avoid congestion and arrive the destination using the shortest time. With the advent of Advanced Traveler Information System, the prediction of short-term link and corridor travel time has become increasingly important. Therefore, it is necessary to forecast future travel time effectively for Advanced Traveler Information System and users. There are two phases in this research. First, we use simulation to produce traffic information in first phase. Second, we use artificial neural network with three layers, fully connected and feed-forward, and backpropagation algorithm to build forecasting highway travel time models. Using simulation to produce the relative data of traffic character detected by vehicle detector sequentially for the base of artificial neural network training and testing. After repeatedly correcting and testing, the effect of forecasting model constructed in the research is very well .As to the result, this research can be provided to forecast travel time in real-time highway travel time estimation.
