Developing a procedure to identify parameters for calibration of a vissim model
Miller, David Michael
MetadataShow full item record
The calibration of microscopic traffic simulation models is an area of intense study; however, additional research is needed into how to select which parameters to calibrate. In this project a procedure was designed to eliminate the parameters unnecessary for calibration and select those which should be examined for a VISSIM model. The proposed iterative procedure consists of four phases: initial parameter selection, measures of effectiveness selection, Monte Carlo experiment, and sensitivity analysis and parameter elimination. The goal of the procedure is to experimentally determine which parameters have an effect on the selected measures of effectiveness and which do not. This is accomplished through the use of randomly generated parameter sets and subsequent analysis of the generated results. The second phase of the project involves a case study on implementing the proposed procedure on an existing VISSIM model of Cobb Parkway in Atlanta, Georgia. Each phase of the procedure is described in detail and justifications for each parameter selection or elimination are explained. For the case study the model is considered under both full traffic volumes and a reduced volume set representative of uncongested conditions.