透過您的圖書館登入
IP:18.221.222.47
  • 學位論文

管流類推法吸引力參數之時空分類研究-以中山高速公路北區路段為例

The Temporal-Spatial Classification of Relative Attractiveness Factor in Fluid Analogy Method – A Case Study of Sun Yat-Sen Freeway

指導教授 : 廖祐君

摘要


本研究討論,依時性管流類推法 (Time-Dependent Fluid Analogy Method, TDFAM) 於高速公路廊道所計算出的各時階出口匝道之Ae值 (Factor of Relative Attractiveness) 特性,企圖將一天中個數繁多且大小有所不同但卻相為類似的Ae值透過其於時-空分佈上的特性簡化合併,使合併後之各時-空平均吸引力參數除了在推估起迄旅次外,亦能當作廊道系統的規劃與控制方面之參考。 本研究利用高公局100年1~4月中山高速公路北區路段之每5分鐘匝道流量資料,在進行資料缺值插補後,代入依時性管流類推法計算出Ae參數值,並利用群落分析法 (Cluster Analysis) 將時空中相為類似的Ae值進行分析合併,把分類後之各時空Ae值以單一平均Ae值代表,並以此平均Ae值進行原始每5分鐘之各起迄站之起迄旅次推估,以平均絕對值誤差率與均方根誤差兩項標準檢測合併後之各時空 Ae 平均值的起迄旅次推估能力,檢測分群概念的適切性。 本研究結果顯示,在尖離峰時段下依各匝道流量特性之合併及以匝道平均一天流量特性分類為前提下將各類空間進行時間段分群之兩種概念分類方式,模式的起迄旅次量推估能力平均皆能維持在誤差20%以內,代表此兩種分群合併方式所產生的各時空Ae平均值能大致符合未合併前之時空Ae值特性,也能表現出廊道系統內各出口匝道吸引力值在尖離峰時段的變化情形。而分類後的Ae時空平均值可作為未來模式於資料時區單位維度選擇上之參考、廊道系統內之短程起迄旅次(隔站即下的流量)的即時預測及各時階各匝道吸引力值的參考指標等應用方面。

並列摘要


This study mainly discussed the characteristics of Relative Attractiveness (Ae) of each off-ramp at various time period calculated by Time-Dependent Fluid Analogy Method (TDFAM). The study attempted to simplify and integrate varied Ae values for easy reference based on characteristics analysis of temporal-spatial distribution. In this way, the mean of temporal-spatial Ae could be easily used in planning and control of corridor system in addition to the estimation of origin-destination (O-D) flows. This study used the ramp flow data at North Section of Sun Yat-Sen Freeway from January to April of 2011. After reparation of the missing data, The Ae was calculated with each 5 minutes for every off-ramp by TDFAM. Moreover, the cluster analysis was applied to group Ae values with similar characteristics under the temporal-spatial characteristics. After classification, each Ae group is represented by a mean of temporal-spatial Ae which used to estimate the O-D flow in every 5 minutes. The O-D flow estimation ability of each temporal-spatial Ae is evaluated by Mean Absolute Percentage Error Rate (MAPE) and Root-Mean-Square Error (RMSE). The study results indicated that the error of O-D flow estimation ability could be maintained within 20% in average. It also indicated the temporal-spatial Ae could represent the characteristics of the temporal-spatial Ae before integration, it can also show the time-dependent characteristics of vehicle flow changing with time in the corridors a day. The temporal-spatial Ae after classification could be taken as the reference for the dimension of time period selection in TDFAM, and the reference for the calculation of real-time O-D flow in the future and the index of the real-time Ae value for each off-ramp in each time period.

參考文獻


11. 吳健生、廖梓淋、林鈺翔,民國100年,「兩步驟類神經網路車輛偵測器遺漏資料之填補及其應用」,運輸計劃季刊,第四十卷,第一期,頁1-30
12. 汪進財、邱孟佑,民國99年,「以車流狀態為基礎之高速公路旅行時間預測模式」,運輸學刊,第二十二卷,第三期,頁261-284。
19. 張堂賢、莊國欽,民國93年,「非線性系統於高快速道路整合匝道儀控之研究」,運輸計劃季刊,第三十三卷,第四期,頁673-698。
21. 張堂賢、黃宏仁,民國97年,「車輛偵測器資料漏失之在線插補技術研究」,運輸學刊,第二十卷,第四期,頁377-404。
25. 黃宏仁,民國97年,車輛偵測器數據補償與正規化研究,國立台灣大學土木所博士論文。

延伸閱讀