Incident Management and Traffic Information: Tools and Methods for Simulation-Based Traffic Prediction

Incident response and mitigation are among the main tasks of operators at traffic control centers. Simulation models have a good chance of reproducing and predicting the effects of incident response by explicitly modeling driver response to the incident and information provided. In the PREDIKT project sponsored by the Swedish National Road Administration, the state-of-the-art mesoscopic simulation model MEZZO was extended to provide decision support for incident management. Numerous essential modeling components are described and tested, including modeling the incident response logic, a mixed-logit model, and a method for generating alternatives for drivers switching routes. In addition, the results of a fast calibration method based on simultaneous perturbation statistic approximation are presented. The model components are tested in a small case study that investigates the effect of delay in providing information to drivers after incidents. A linearization of speed–density functions also is shown to improve computational performance by 30% and increase calibration speed and stability while preserving simulation accuracy.

Language

  • English

Media Info

Subject/Index Terms

Filing Info

  • Accession Number: 01154517
  • Record Type: Publication
  • ISBN: 9780309142892
  • Report/Paper Numbers: 10-1989
  • Files: TRIS, TRB, ATRI
  • Created Date: Jan 25 2010 10:55AM