A new strategy for obtaining VISSIM’s behavioral parameters for microscopic simulation of two-lane rural highways

In this work, we propose a Greedy Randomized Adaptive Search Procedure (GRASP) based algorithm to calibrate and validate the VISSIM traffic simulator's behavioral parameters. As far as we are concerned, no work has used a GRASP to solve the calibration and validation of this kind of software, especially to model two-lane highway segments. We compare the values obtained from a GRASP and a Genetic Algorithm (GA). Our experiments use traffic data collected in segments of two-lane rural highways on a Brazilian highway under different geometric and traffic conditions. Half of the traffic data collected in the field was used for calibration, and the other half was used to validate the calibrated simulator. Our goal was to minimize the Root Mean Squared Error (RMSE), using the average travel speed (ATS) collected in the field and obtained from VISSIM. Were applied a GA and a GRASP to solve this problem. Both applications improve by around 40% concerning the values obtained with the default VISSIM parameters provided by the simulator. Besides, both algorithms choose more aggressive parameters than VISSIM's default parameters. The results showed that GRASP is an efficient technique for calibrating microscopic traffic simulators.