都市幹道動態旅行時間推估與交通偵測設施佈設準則之研究

淡江大學機構典藏 > 商管學院 > 運輸管理學系暨研究所 > 學位論文 > Item 987654321/33837

jsp.display-item.identifier=請使用永久網址來引用或連結此文件: https://tkuir.lib.tku.edu.tw/dspace/handle/987654321/33837

题名:	都市幹道動態旅行時間推估與交通偵測設施佈設準則之研究
其它题名:	Dynamic estimation of travel time and evaluation of the installation criteria for traffic detector on urban arterials
作者:	李鈺雯;Lee, Yu-wen
贡献者:	淡江大學運輸管理學系碩士班胡守任;Hu, Shou-ren
关键词:	先進用路人資訊系統;旅行時間;車輛偵測器;巨觀車流理論;卡門濾波理論;類神經網路;Advanced Traveler Information Systems;Travel time;Vehicle detector;Macroscopic traffic flow theory;Kalman filtering model;Artificial neural networks
日期:	2005
上传时间:	2010-01-11 04:35:25 (UTC+8)
摘要:	近年來由於都市的高度發展，交通問題日趨嚴重，民眾對於「行」的資訊需求亦是與日俱增。以先進用路人資訊系統（Advanced Traveler Information Systems, ATIS）角度探討都市幹道旅行時間議題為目前重要的研究課題之一。為提供即時交通資訊，各交通主觀機關均積極佈設各項交通偵測器，藉由交通偵測器所回報之各項交通參、變數，進一步應用各種模式轉換為對民眾具有價值之有效資訊。由於「旅行時間」對民眾而言可說是一種最直接的交通資訊，民眾在得知相關路徑的旅行時間後，在出發前可據以評估選擇所使用之交通運具；在途中亦可做為路徑選擇的依歸，故路段旅行時間可說是相當直觀、有效的交通資訊。目前國內、外有關旅行時間主要的資料來源為交通偵測器，大部分的偵測器皆可收集流量、佔有率、速度等交通變數，但是不同種類的偵測器也會因其不同特性而有所限制，因此在使用上必須通盤考量才能有較佳之資料品質。本研究以非接觸式微波雷達偵測器SmartSensor作為資料蒐集之工具，並透過不同方法論推估路段旅行時間。目前常見的旅行時間演算法主要的理論可分為三個方向：車流理論、統計分析，以及人工智慧方法。各項方法均有其限制條件及不同的輸出、輸入項，故本研究藉由比較巨觀車流理論、卡門濾波模式，以及類神經網路模式等方法論，進一步找出較適合台灣地區使用之路段旅行時間推估模式與演算法。此外，偵測器之佈設策略對於模式績效的影響甚鋸，故本研究希冀分析最合適的偵測器佈設準則，以獲得準確的旅行時間預測值。本研究藉由實驗設計，以旅行時間推估模式為基準，分別從縱向、橫向與數量等三個角度分析適合台灣地區的偵測器佈設準則，藉此希望能找出適合本土的都市幹道路段旅行時間推估模式及偵測器佈設策略，以期能提供用路人可靠且值得信賴的交通資訊。根據模式數值分析結果顯示，旅行時間推估模式以類神經網路模式之績效最佳，其MAPE值約10%。而偵測器佈設策略方面，合適的佈設位置短路段以距路段上游約100公尺為佳；長路段則以路段中游處為佳，惟該分析結果為特定路型之結論而非通則性之答案。此外，成對偵測器佈設策略可提高模式之準確性，使得模式績效提高，但是幅度不大，是否值得在同一路段上佈設兩組偵測器，鑒於偵測器之成本高昂，在成本效益的考量下，原則上以每一路段佈設一組車輛偵測器為宜。 Due to the rapid development in urban areas, traffic problem becomes serious in recent years. It turns out that travelers need more traffic information both in qualitative and quantitative perspectives. In the area of advanced traveler information systems (ATIS), the estimation of arterial travel time is one of the crucial research topics. In order to conduct effective traffic management, the government agencies have installed vehicle detectors to monitor and collect traffic characteristics (e.g., flow, occupancy, and speed) and estimate link travel time accordingly. The present research used a microwave type vehicle detector called SmartSensor to collect traffic characteristics and use three kinds of models to estimate link travel time. The research is aiming to evaluate different link travel time estimation methodologies and propose installation criteria for traffic detector on urban arterials. Because little work has been done in the area of arterial link travel time estimation and corresponding vehicle detector installation strategies, therefore we are targeting to investigate the relationship between detector location and the ability of a system to monitor traffic characteristics. To evaluate the targeted travel time models and to evaluate the installation criteria for traffic detector, RMSE and MAPE values were calculated. In the issue of travel time estimation, the results were promising in view of most travel time estimates are statistically accepted. The best model is ANN-based models. Its MAPE is 10%. In another issue, the optimal detector location was identified to be about 100 meters from upstream intersection. The study also showed that detector data obtained on one link could only represent accurate link travel time estimate on that link, and it could not be representative to an adjacent link. Finally, link travel time estimate obtained by using pair-wise is slightly better than that of using single vehicle detector, however in view of the high cost of traffic detector costs, it is suggested to install one traffic detector at most in a single link on urban arterials.
显示于类别:	[運輸管理學系暨研究所] 學位論文

文件中的档案:

档案	大小	格式	浏览次数
	0Kb	Unknown	256	检视/开启

在機構典藏中所有的数据项都受到原著作权保护.

TAIR相关文章

数据加载中.....