在台灣, 高速公路上交通車流有經常性壅塞之現象,為發展先進旅行者資訊系統,實有其必要透過預測模式提供駕駛人可靠的旅行時間預測資訊做為參考。由於我國高速公路上重車與客運車數量佔交通量比率大且變異程度高,致使欲取得較可靠的高速公路旅行時間預測資訊變得困難且複雜。在旅行時間變異程度較大的情況下下,相對突顯出即時預測資訊的重要性,透過預測模式輔以蒐集短期間之旅行時間資料,即時更新預測結果,勢必更能掌握未來路況的發展趨勢。 本研究應用灰色系統理論包括灰預測模式、灰關聯分析於高速公路旅行時間預測上,測試分析預測模式在每種不同旅行時間分佈型態的預測績效與優劣處。並針對預測模式之缺點進行分析並提出修正方法,依預測長度將其拓展為分預測、時預測與日預測形式,探討未來結合電視新聞、廣播、先進旅行者資訊系統、車上電腦、資訊可變標誌等媒介之旅行時間預報系統發展。 研究結果證實以灰色預測模式作為旅行時間預測方法,除了所需要的交通資料甚少、其建立模式過程亦較其他預測理論簡單、迅速,且績效相當良好,尤其是應用於短期預測上,未來可發展應用於即時旅行時間預測上。
In Taiwan, Traffic on freeway is very heavy and often congested. For deploying of Advance Traveler Information System, a prediction model to provide drivers with adequate predicted travel time is essential. Because of the variant traffic compositions of heavy vehicles and intercity buses, the prediction of travel time becomes more complicated and difficult to get a reliable predicted travel time. The variance of travel time is high and the predicted result of travel time should be revised in a short interval. Therefore, this study applies grey system theory including grey prediction model and grey relational analysis to test its performance on different traffic flow conditions. Additional, this study proposes some revised methods of grey travel time prediction model because there are still some shortcomings in predicting the highly fluctuated data series. According to the predicted value of freeway travel time, this study extends to develop the model of prediction of minutes, prediction of hours, prediction of days and discusses the development of a travel time prediction broadcasting system including TV News, Advance Traveler Information System, vehicle computers, CMS, and Website. This study proves that the grey theory for prediction just needs not only fewer data but shorter time to built models than the other methods. Its performance is very good to apply to the real-time prediction in the future.