Calibrating Speed-Density Functions for Mesoscopic Traffic Simulation

Mesoscopic traffic simulation models combine macroscopic supply (e.g. link performance functions and capacities) with microscopic demand (e.g. individual drivers and disaggregate behavior models) to capture the time-varying evolution of congestion patterns, queues and spillbacks on traffic networks. Such systems are being applied to solve a variety of off-line and on-line traffic management problems. Often, the link performance function takes the form of a speed-density relationship for each link or segment in the study network. These functions are based on the fundamental diagram and model the variation of average vehicle speed with traffic density. Since each network link or segment can have a separate speed-density function (each with a few parameters), the total number of parameters to be calibrated is potentially very large. This paper presents some experiences with calibrating speed-density functions on a variety of real networks. Flexible off-line and on-line calibration frameworks are discussed. The calibrated functions are embedded within the DynaMIT dynamic traffic assignment model, whose accuracy is evaluated against real data. The expected benefits from re-calibrating speed-density functions during on-line operations using real-time sensor data are also illustrated.

Language

  • English

Media Info

  • Media Type: CD-ROM
  • Features: Figures; References; Tables;
  • Pagination: 13p
  • Monograph Title: Symposium on the Fundamental Diagram: 75 Years (Greenshields 75 Symposium), July 8-10, 2008, Woods Hole, MA. Preprints

Subject/Index Terms

Filing Info

  • Accession Number: 01109345
  • Record Type: Publication
  • Files: TRIS, TRB
  • Created Date: Aug 25 2008 8:11AM