01153373

Component

A multi-criteria calibration procedure is proposed for parameter calibration of microscopic traffic simulation platforms. The impetus for this paper is provided by increased usage of microscopic simulation models in transportation safety studies. Before such models can be adopted in safety research it is important that we obtain parameter values that reflect real world traffic conditions. Current state-of-the-art Genetic Algorithm calibration procedures only allow for one measure of performance in parameter calibration. In safety studies, the fitting function used in calibration has been safety performance. Therefore, the underlying traffic-related factors, such as speed, volumes and density have not been directly considered in calibration. Since these models are based on simulating traffic and calibration needs to be based observed traffic attributes. The proposed multi-criteria procedure would allow for the direct calibration of these traffic attributes while at same time producing accurate estimates of safety performance. The multi-criteria procedure is applied to a sample of vehicle tracking data and the results are compared to parameter values suggested by a single-criteria approach and platform defaults.

01147878

10-0858

English

17p

2010

Transportation Research Board 89th Annual Meeting

DVD

Figures (2) ; Photos (1) ; References (17) ; Tables (7)

Automatic tracking; Calibration; Estimating; Genetic algorithms; Highway safety; Microscopic traffic flow; Microsimulation; Traffic density; Traffic models; Traffic speed; Traffic volume

Highways; Operations and Traffic Management; Safety and Human Factors; I72: Traffic and Transport Planning

Transportation Research Board Annual Meeting 2010 Paper #10-0858

TRIS, TRB

Jan 25 2010 10:24AM