提供用路人有效之旅行時間，將有助於用路人在出發時間、路徑選擇上做出正確判斷。而校車為一固定路次、單一班次、無前車資訊之通勤車輛，且對於校車使用者而言，如果不能有效掌握校車行車資訊，將耗費更多的旅行時間與旅行成本才能到達目的地。本研究透過全球衛星定位技術，獲得校車即時行車資料，並利用資料庫系統構建校車行車資料庫，並以車輛行車速度作為輸入變數，進行希爾伯特-黃轉換中的經驗模態分解，再利用灰預測GM(1,1)模式針對各分量進行預測，求得校車未來之行車速度，以此估算校車到達下一停靠站旅行時間，結合歷史資料推估模式，應用歷史資料推估後續各站之旅行時間，計算整體路徑旅行時間，並建構校車位置與預估到站時間語音查詢系統。經由測試結果比較，發現本研究所構建之旅行時間預測模式，在平均絕對誤差時間與平均絕對誤差百分比部分均優於只使用歷史資料推估法，顯示本研究模式有助於提升旅行時間預測的準確度，而根據本研究之問卷結果顯示，提供語音查詢服務能夠降低乘客之安全預留時間，有助於乘客縮短旅行時間。 To provide Travel time information for commuters, it will assist the commuters decision the departure time and routing. The school bus is the commuted transportation which runs with regular route and single runs of scheduled bus. If The commuters do not know well the driving information of school, it will result more travel time and cost. The research use the Global Positioning system receive school bus travel data, and input the school bus speed data to proceed Empirical Mode Decomposition, then employ the Grey Forecasting GM(1,1) Model to prediction IMF and calculate the next station arrival time, finally combine with Historical Profile Approach Model prediction the path travel time. Compared with the experiment results, the researcher set up both the prediction models of travel time, the testing result to appear in the Mean Absolute Error Time and Mean Absolute Percentage Error, are better than Historical Profile Approach. It indicates that this research model can improve the accuracy of travel time. Also, according to the result of questionnaire survey, it shows the broadcasting enquiring system can help customers to save their travel time.
