臺灣博碩士論文加值系統

高速公路已成為國人國內移動方式的首選，然而車輛日益增多所帶來的塞車問題也是必須克服的重要環節。為有改善交通亂象以及保障用路人的旅行品質，交通部台灣區國道高速公路局引進計程電子收費系統，並於國道上架設收費門架，除了節省進入收費站的等待時間，收費門架也具備收集車輛資料功能。本研究的重點在於如何有效利用車輛資料建立旅行時間預測模型，使用路人利用替代道路免於塞車，浪費不必要的等待時間。

本研究以計程電子收費系統的公開資料為研究資料，經由資料預處理從路段、時間、車輛資料建立旅行時間預測模型，比較各模型結果選出適宜台灣高速公路的預測系統。

在模型建立上使用時間序列分割日期與時段以及聚類對資料進行預處理，以平均行駛車速和車流量做為輸入變數，再以KNN、支援向量機、類神經網路三種分類器建立旅行時間預測模型。研究結果發現以聚類進行資料預處理能大幅降低預測誤差值，因為聚類能有效去除離群值在模型訓練時帶來的影響。

Highway has become people’s first choice for the country moves in Taiwan. However, an important part of the growing problem of traffic congestion caused by vehicles also must be overcome. To make traffic problem better and to protect road occupants’ traveling quality, Taiwan Area National Freeway Bureau, MOTC introduced Electronic Toll Collection not only saving time in passing through the toll but also collecting data from cars for using efficiently. This study focus on using the data from cars to establish traveling–time prediction model and make road occupants use alternative roads from traffic congestion.

Research data comes from Electronic Toll Collection’s open data. By the data preprocessing, from sections, time, and data of cars establish traveling–time prediction model and compare each one to choice the best highway prediction system in Taiwan.

To establish traveling–time prediction model, this study use time series model to split date and time period and cluster for data preprocessing then input average speed and traffic flow as variables to three classifiers like KNN, support vector machine, and neural network. The experiment result shows preprocessing data by clustering can reduce deviation value by a wide margin because of clustering may exclude outliers in training model.

第一章 緒論...............................................1
1.1 研究背景與動機.........................................1
1.2 論文架構...............................................2
第二章 文獻探討...........................................3
2.1 旅行時間預測模型相關文獻...............................3
2.2 分類器相關文獻.........................................4
2.2.1 KNN簡介.......................................................4
2.2.2 類神經網路簡介................................................. 5
2.2.3支援向量機（support vector machine, SVM）簡介.......................6
第三章 研究方法...........................................8
3.1 實驗資料和資料預處理變數...............................8
3.1.1 選取實驗資料................................................... 8
3.1.2 時間序列........................................................9
3.1.3 聚類...........................................................10
3.2 預測驗證標準..........................................11
3.3 模型建立流程..........................................11
第四章 實驗結果..........................................13
4.1 實驗環境..............................................13
4.2 小客車時間序列結果比較................................13
4.3 小客車聚類結果比較....................................24
4.4 不分車種時間序列結果比較..............................29
4.5 不分車種聚類結果比較..................................39
4.6 模型驗證...........................................................40
4.7 旅行時間預測模型之資訊管理意涵.....................................46
第五章 結論與未來研究方向................................49
參考文獻..................................................50
附錄一 小貨車時間序列處理實驗數據........................................52
附錄二 大客車時間序列處理實驗數據........................................68
附錄三 大貨車時間序列處理實驗數據........................................84
附錄四 聯結車時間序列處理實驗數據.......................................100

附錄五 各車種聚類處理實驗數據...........................................116

[1]高速公路計程收費通行量
http://stat.motc.gov.tw/mocdb/stmain.jsp?sys=100&funid=a3206
[2] 卓訓榮、周幼珍、王晉元、蔡繼光，1999，高速公路旅行預測－以k-NN法
及分群法探討，中華民國運輸協會98年學術論文研討會
[3] 李穎，2001，類神經網路應用於國道客運班車旅行時間預測模式之研究，國
立成功大學交通管理學系碩士學位論文
[4] B.L Smith, M.J Demetsky，1997，Traffic flow forecasting: comparison of modeling approaches，Journal of transportation engineering，123 (4)，261-266
[5] Brian L Smith, Billy M Williams, R Keith Oswald，2002，Comparison of parametric and nonparametric models for traffic flow forecasting，Volume 10, Issue 4, Pages 303–321
[6] Lee, Y.I., 1998, Prototype of Traffic Information Center of Seoul, 5th ITS World Congress
[7] Hellinga, B., G. Knapp, 1999, Automatic Freeway Incident Detection using Travel Time Data from AVI Equipped Vehicles, 6th World Congress on Intelligent Transport System
[8] Ding, Y., Chien, S. I., and Wei C.H., 2000, Dynamic Transit Arrival Time Prediction Using Link-Based and Stop-Based Artificial Neural Networks (Revised Version), Submit to Transportation Research
[9] Zhang, X. Y., and J.A. Rice, 2003, Short-Term Travel Time
Prediction: Transportation Research Part C, Vol.11, pp.187-210,
[10] Rice, J., and E.V. Zwet, 2004, A Simple and Effective Method for Predicting Travel Time on Freeways, IEEE Transactions on Intelligent Transportation Systems, Vol. 5, No. 3
[11] Rice, J., and E.V. Zwet, 2005, A Simple and Effective Method for Predicting Travel Time on Freeways, IEEE Transactions on Intelligent Transportation Systems, Vol. 5, No. 3
[12] Chang, G.L., Zou, Nan, Wang, Jianwei, 2006, Development and Filed
Evaluation of a Real-Time Travel Time Prediction System, Maryland Department of Transportation State Highway Administration, Draft Final Report
[13] Hsung-Jung Cho, Ming-Te Tseng, 2007, Shockwave Detection for
Electronic Vehicle Detectors, Computational Science – ICCS pp.273.4-282
[14] T.M. Cover, P.E. Hart,1967,Nearest Neighbor Pattern Classification, IEEE Transactions on Information Theory, vol. IT-IS, no. 1, pp.21-27.

[15] Warren S. McColloch and Walter Pitts, 1943, A Logical Calculus of the Ideas Immanent in Nervous Activity
[16] 楊錦欽，2002，類神經網路運用於型材矯直之研究，國立國立中興大學機械工程學系碩士學位論文
[17] Cortes, C., & Vapnik, V., 1995, Support-vector networks. Machine learning,20(3), 273-297.
[18] 林宗勳，support vector machine簡介 http://www.cmlab.csie.ntu.edu.tw/~cyy/learning/tutorials/SVM2.pdf
[19] 交通部台灣區國道高速公路局，2013，高速公路中長程旅行時間預測模式之建立與應用
[20] J.B McQueen, 1967, Some methods for classification and analysis of multivariate observation, University of California, Los Angels
[21] Tofallis, 2015, A Better Measure of Relative Prediction Accuracy for Model Selection and Model Estimation, Journal of the Operational Research Society, 66(8), 1352-1362
[22] Chen, M., and I.J. Chien, 2001, Dynamic Freeway Travel Time Prediction Using Probe Vehicle Data: Link-based vs. Path-based, Transportation Research Board 80th Annual Meeting
[23] Hall, D. L, 1992, Mathematical Techniques in Multi-sensor Data Fusion,
Boston, MA: Artech House
[24] 交通部台灣區高速公路局，交通管理控制系統
http://www.freeway.gov.tw/Publish.aspx?cnid=87&p=79