於資料漏失下之旅行時間預估

Abstract

智慧型運輸系統在世界各先進國家發展下，對於改善交通雍塞、節省能源以及提升運輸安全已有顯著之成效。而先進交通管理系統(Advanced Transportation Management System, ATMS)為ITS發展之核心，主要功能與服務為即時偵測交通狀況與交通控制等。而隨著科技發展進步，先進旅行者資訊系統(Advanced Traveler Information Systems, ATIS)亦為近年來ITS發展之重點，主要功能為提供用路人有效且正確即時交通資訊。若能將ATIS與ATMS兩系統做資訊整合，擬定正確交通控制策略外，能提供即時且正確交通資訊至使用者。ATMS系統係主要利用偵測、通訊、控制等技術，將偵測所得的交通狀況，經由通訊網路傳輸到交控中心，中心再結合其他系統所獲得資訊擬定交通控制策略。若偵測器因外在或內在因素導致導致提供無效或偏差之資料，將會對ATMS擬定交通控制策略與ATIS系統提供使用者之交通資訊造成偏差。 因此本研究針對資料漏失，研擬兩種資料漏失插補模組，供ATMS系統與ATIS系統能提供有效且正確交通資訊給使用者。第一種為定時插補資料模組內建立方曲線內插法、線性插補法、鄰近值插補法。因車輛偵測器故障或探偵車密度不足無法收集到連續資料，透過定時插補資料模組，可有效且完善的維護資料庫。第二種為即時插補資料模組內建平均值法、傅立葉法、α−β−γ濾波器結合平均值、α−β−γ濾波器結合傅立葉法。當ATIS系統進行即時旅行時間預估，但無即時資料，透過即時插補資料模組，可避免旅行時間預估偏差。 經由本研究實驗設計，定時插補資料模組中，線性插補法為插補績效最佳方法。即時插補資料模組中，四種插補方法績效無顯著性差異，因此可依不同交通管理者需求而採用。經由定時與即時插補過後之資料供旅行時間預估模組演算，在旅行時間推估方面，皆獲得良好績效。

The development of Intelligence Transportation System among numerous advanced countries in the world has made a remarkable progress on improving traffic jams, energy conservation, and the promotion of transportation safety. The core of ITS is called Advanced Transportation Management System, ATMS, which is mainly used for the immediate traffic condition detection and control, etc. Along with recent Internet technological improvement, Advanced Traveler Information Systems, ATIS is also a key point in the growth of ITS. Providing the efficient and correct immediate traffic information to the users is the major function of ATIS. In this research provide two module to avoid missing data, the first is on-time interpolate data which including cubic interpolation, linear interpolation, neighbor interpolation. The second module is real-time interpolate data which including mean interpolation, Furier interpolation, mean interpolation combine with α−β−γ filter, Furier interpolation combine withα−β−γ filter. Acroding the experiment design, the linear is the best results in on-time interpolation and there were no difference between the four methods of interpolating in real-time interpolation. It depend the authority demend to choose interpolation method.