Identifying Grade Effects on Network Speed Prediction

Traffic speed prediction is an important task to support a variety of intelligent transportation system (ITS) applications. Accurate speed estimation is helpful in monitoring the network and providing reliable information to travelers. Among all the factors that can influence traffic speed, road grade is responsible for causing speed fluctuations and speed differences among different vehicle types. However, due to limitations in data availability, previous speed prediction models have not sufficiently addressed the grade impact. This study overcomes the limitation by extracting a high-resolution elevation dataset from the Google Earth topographic map. With the detailed road grade information, a continuous wavelet transform (CWT) based peak identification method has been proposed to divide the studied route into segments with homogenous vertical alignment types (e.g., grade and curve).The identified vertical alignment information has then been applied to evaluate the performance of a typical support vector regression (SVR) speed prediction model. Statistical results of performance measures revealed that the tested model tends to 1) be less accurate on upgrade and downgrade segments when traffic is not congested; 2) significantly overestimate speed (e.g., more than 10%) on sag curves when traffic is congested. The study shows that ignoring grade effects will result in system errors in the speed prediction results. The developed grade dataset and the processing method have potentials in help improving the existing speed prediction model framework

• Supplemental Notes:
  • This paper was sponsored by TRB committee AHB40 Standing Committee on Highway Capacity and Quality of Service.
• Authors:
  • Zhu, Wenbo
  • Wang, Yinhai
• Conference:
  • Transportation Research Board 97th Annual Meeting
  • Location: Washington DC, United States
  • Date: 2018-1-7 to 2018-1-11
• Date: 2018

Language

• English

Media Info

• Media Type: Digital/other
• Features: Figures; References; Tables;
• Pagination: 17p

Subject/Index Terms

• TRT Terms: Highway grades; Highways; Mathematical prediction; Networks; Traffic speed; Vector analysis; Wavelets
• Subject Areas: Highways; Planning and Forecasting;

Filing Info

• Accession Number: 01660921
• Record Type: Publication
• Report/Paper Numbers: 18-06728
• Files: TRIS, TRB, ATRI
• Created Date: Feb 22 2018 9:17AM