Prediction of Multivariate Traffic Parameters under Normal and Abnormal Conditions on the Basis of Multi-Scale Online SVR

Short-term traffic flow prediction algorithms in ordinary condition are popular in literature. As to traffic management systems in ITS, specifically the dynamic traffic assignment, the short-term prediction of multivariate fundamental traffic parameters, especially under abnormal conditions, such as collision, work and etc. are more vital and preferable. For this purpose, the study proposes a prediction methodology based on Multi-Scale Online Support Vector Regression (MSOL-SVR) algorithm to forecast traffic flow and average speed under normal and abnormal condition. The data provided by the Vehicle Detector Station (VDS)are separated into global data and local data. In normal traffic condition, the prediction based on MSOL-SVR with global data is preferable for the traffic flow forecasting for its good accuracy and efficiency. Additionally, it can provide the whole day prediction without the limitation of short term. The MSOL-SVR with local data is suitable for the prediction of short-term traffic flow and average speed, which has a good performance even in abnormal condition. The paper presents an application of the methodology to predict freeway traffic flow and speed in normal and abnormal conditions. Additionally, the proposed algorithm also is proved to be capable of road network forecasting and provides more accurate and robust results. The results indicate the proposed methodology has an optimal performance.

• Supplemental Notes:
  • This paper was sponsored by TRB committee AHB45 Traffic Flow Theory and Characteristics.
• Corporate Authors:

  Transportation Research Board

  500 Fifth Street, NW
  Washington, DC  United States  20001
• Authors:
  • Quan, Wei
  • Wang, Hua
  • Lin, Datong
  • Wang, Yinhai
• Conference:
  • Transportation Research Board 93rd Annual Meeting
  • Location: Washington DC
  • Date: 2014-1-12 to 2014-1-16
• Date: 2014

Language

• English

Media Info

• Media Type: Digital/other
• Features: Figures; References; Tables;
• Pagination: 17p
• Monograph Title: TRB 93rd Annual Meeting Compendium of Papers

Subject/Index Terms

• TRT Terms: Algorithms; Dynamic traffic assignment; Forecasting; Intelligent transportation systems; Mathematical prediction; Traffic flow; Traffic speed; Vehicle detectors
• Subject Areas: Data and Information Technology; Highways; Operations and Traffic Management; I72: Traffic and Transport Planning; I73: Traffic Control;

Filing Info

• Accession Number: 01516095
• Record Type: Publication
• Report/Paper Numbers: 14-3551
• Files: TRIS, TRB, ATRI
• Created Date: Feb 27 2014 9:06AM