A genetic algorithm-based calibration of VISSIM using GPS data

This paper proposes a Genetic Algorithm (GA)-based approach for calibrating the driving behaviour parameters of the microscopic traffic simulation model VISSIM. The approach defines the index of simulation accuracy as the Sum of Squared Error (SSE) of the collected vehicle speeds versus simulated vehicle speeds at the cross-sections along the road. The objective of the calibration is to search for the optimal values of driving behaviour parameters that minimize the SSE. The paper presents the modelling process of this GA-based approach as well as the computer program implementing the approach, which is developed using MATLAB, GA Toolbox, and Visual Basic. The road network around the terminals at the Intercontinental Airport of Houston (IAH) is selected for the case study. Traffic data are collected around the IAH using traffic counters and a test vehicle equipped with GPS is driven around the IAH loop to collect the vehicle’s instantaneous speed data. The GA-based calibration is implemented with the collected data, and the optimal values of the driving behaviour parameters in VISSIM are derived. A comparison indicates that the SSE has decreased almost 50% after the calibration. The case study has shown that the proposed GAbased approach is very efficient and practical.