| Abstract: | 近年來智慧型運輸系統(Intelligent Transport Systems, ITS)對於改善交通擁擠問題、節省能源及提升運輸安全已有顯著的成效。其中，先進交通管理系統(Advanced Transportation Management Systems, ATMS)一直是ITS發展的核心，主要功能與服務為即時偵測與交通控制；但隨著用路人對於運輸品質要求逐漸提升，先進旅行者資訊系統(Advanced Traveler Information Systems, ATIS)亦成為近年來ITS發展的重點項目，主要功能與服務為提供用路人有效且正確的即時交通資訊。未來若能將ATMS與ATIS兩系統進行資訊整合，除可有效執行擬定之交通控制策略外，亦能同時提供正確的即時交通資訊給用路人。其中，蒐集即時交通資料為運輸智慧化之基礎，負責資料蒐集之車輛偵測器，包含其設置條件、佈設位置、參數資訊、以及如何提供有效正確的交通資訊給用路人等，即為本研究欲探討之課題。
本研究構建之旅行時間推估模式，包含路段旅行時間部分和路口延滯部分。目前常見的旅行時間演算法主要包括車流理論、統計分析，以及人工智慧等方法，各方法均有其限制條件和不同的輸入、輸出項。本研究選取理論基礎完整、操作簡單的巨觀車流理論，進行旅行時間推估。一般路口延滯公式多以估計停等延滯為主，本研究加入臨近路段延滯(Approach Delay)觀念，進一步探討在不同交通情境下，延滯公式之適用性。藉由系統化的分析，找出適合台灣地區都市幹道應用之旅行時間推估模式。
此外，偵測器佈設位置對於模式績效的影響甚鉅，本研究透過實驗設計方法，以旅行時間推估模式為基準，分別從流量等級、路段長度分析因應不同交通資訊需求下的偵測器佈設準則，藉此找出適合本土的都市幹道路段路徑旅行時間推估模式及偵測器佈設位置，以提供用路人可靠的交通資訊。
模式數值分析結果顯示，旅行時間推估模式以Oh模式配合不同交通情境下的適用延滯公式之績效最佳，整體路段平均MAPE值約為15％左右。偵測器佈設組數以單一偵測器獲得之整體推估績效較為穩定且一致。而偵測器佈設位置方面，在高、中低流量以及長、短路段中，皆以佈設於路段中游處為佳。在部分特定的交通情境下，有其折衷設位置存在，即該區域內設置偵測器能同時因應交控需求，亦能滿足用路人資訊需求。此外，部分路段在加入交控偵測器後可提高模式之準確性以及推估績效，但是改善幅度不大，鑑於偵測器成本昂貴，在成本效益考量下，原則上一路段以佈設一組車輛偵測器為宜。最後，本研究歸納出在因應不同交通資訊需求的情況下之偵測器佈設準則，期供交通管理單位實務應用之參考。
Due to the rapid development of intelligent transport systems (ITS), there are apparent effects on improving traffic jam issues, saving energy and promoting safety of transportation in recent years. Among them, Advanced Transportation Management Systems (ATMS), has been the core of development in ITS all the time. The main functions include real time traffic flow detection and traffic control related activities. In addition, Advanced Traveler Information Systems (ATIS) also become the key item of development in ITS. If the information of ATMS and ATIS can be integrated in the future, it besides can carry out the traffic control strategies effectively, also can provide correct traffic information to road users at the same time.
A macroscopic traffic flow model is established to estimate travel time in this study. In addition, this study joins the concept of approach delay to discuss suitability of intersection delay models under various traffic situations. With systematic analysis, this study intends to develop a travel time estimation model that is suitable for urban arterials of Taiwan.
An experimental design was conducted for travel time estimation by using the control factors, such as level of flow and the length of link to analyze the criteria for vehicle detectors allocation under various needs of traffic information. For which can verify the travel time estimation models and the location of detectors that is suitable for urban arterials.
The MAPE values were calculated to evaluate the proposed travel time estimation models and the installation criteria for vehicle detectors. In the issues of travel time estimation, the results were promising in view of most travel time estimates are statistically accepted. The best models are Oh model to match up the suitable delay models under various traffic situations. The MAPE values are all around 15%. In addition, the estimate performances are shown to be comparatively steady and consistent while installing single detector. The optimal detector location was identified to be about mid-block position from upstream intersection, and under some specific traffic situations, there were trade-off locations for vehicle detectors installation. In addition, link travel time estimates obtained by using pair-wise detectors is slightly better than that of using single vehicle detector, however, in view of the high cost of vehicle detector, it is suggested to install one vehicle detector at most in a single link on urban arterials.
Finally, the installation criteria for vehicle detectors under various needs of traffic information are proposed, which can provide a guideline for reference to the traffic authorities. |
