高速公路路段速率推估之研究

Abstract

近年來智慧型運輸系統ITS(Intelligent Transportation System)的發展已在國內蓬勃發展，在交通部及各地方政府通力合作之下，已慢慢地由中央推廣至地方。從早期較先發展的高速公路即時速率、事件顯示系統，以及國際機場之班機到離資訊系統，發展至今高速公路部份除可提供速率資訊外，即時影像資訊民眾亦可隨時掌握。各縣市政府亦積極在推動公車動態資訊系統、交控系統之規劃與建置工作，並且慢慢地發展各地方政府之交通資訊中心，希望能提供各項交通資訊訊息給民眾，以滿足民眾之需求。 『便民即時交通資訊系統』已提供高速公路即時路況資訊，其資料來源主要為高公局偵測器資訊，當偵測器故障時則該路段之速率將無法提供，有鑑於此，本研究乃蒐集高公局偵測器資訊，並建立歷史速率資料庫，及構建各種速率推估方式，以因應偵測器故障而無法提供即時路段速率時，可從歷史資料庫或經模式推估取得預估速率，提供高速公路穩定、完整的路段速率，進而提供後續研究預估車輛旅行時間之用途。 本研究內容包括： 1.建置高速公路歷史速率資料庫，以供後續研究使用。 2.分析路段速率變化情形，作為將來監控偵測器是否異常之依據。 3.構建迴歸推估模式，找出最佳推估方程式。 4.分析比較不同推估方式之差異，建立不同狀況下何種推估方式準確度較高。

In last decade, ITS (Intelligent Transportation System) in Taiwan has obtained significant progress. Through the cooperation among MOTC and transportation agencies in local governments, ITS infrastructure constructions are specially encouraged. In the early stages, the Taiwan Area National Freeway Bureau could only provide real-time traffic speed and traffic event display on certain freeway segments, and international flight arrival and departure information. Till now, she can also provide the real-time image of traffic flow situation to the public. Besides, there are many local governments install many kinds of APTS and ATMS projects to provide more helpful traffic information to satisfy the needs of public in their modern life. “The real time traffic information system” can provide many useful freeway information, which are mainly collected by traditional vehicle detectors installed on freeways. However, if the detectors are broken down, the system is not able to provide any information. If there is no back up alternative to gather quasi- realtime information, this service will not be reliable. The thesis is trying to develop a methodology to recover the lost of traffic speed data on certain freeway sections. The thesis has collected daily speed data on certain freeway sections in three minutes interval as the historical data. Moreover, empirical analysis models are provided to recover the lost of speed data by retrieving the similar traffic situations from the historical database. The thesis has shown that it is a reasonable solution to ensure the real-time traffic information system can be function appropriately under the existing equipment constraints. It is hoped that the empirical models used in the thesis can be further applied to the study of trip travel time estimation. There are four major contributions of this thesis, 1.Building up the standard historical freeway traffic speed database in Taiwan. 2.Analyzing the variation of the speed data on certain freeway sections, which can be a good comparing base to check if any on-site vehicle detectors are malfunction. 3.Identifying the appropriately regression model to estimate the lost of freeway speed data. 4.By comparing the estimation results among different models to understand the difference of feasibility for each model under different conditions.