Optimization of VISSIM Driver Behavior Parameter Values Using Genetic Algorithm

Abstract

Modeling effective vehicular traffic is a highly contested topic, especially in developing countries like Sri Lanka, which has a wide range of driving conditions. VISSIM microsimulation software is currently used by Road Development Authority (RDA) and relevant authorities to perform traffic management solutions in Sri Lanka. However, it is required to do modifications to the existing driver behavior parameter values to effectively reflect the realistic traffic conditions observed in the real-world in the simulated model. The main purpose of this study is to calibrate the VISSIM driver behavior parameter values using a genetic algorithm (GA). The methodology and results of the VISSIM model’s sensitivity analysis and calibration, which was developed for the Malabe three-legged signalized intersection, are presented in this study. A sensitivity analysis was used to find the most sensitive driver behavior parameters. Using the multi-objective GA optimization tool in the MATLAB software's optimization toolbox, the optimum driver behavior parameter values for these identified most sensitive driver behavior parameters were determined. The findings revealed that GA optimization is effective in reducing the difference between observed and simulated results.