Simultaneous calibration of microscopic traffic simulation model and estimation of origin/destination (OD) flows based on genetic algorithms in a high-performance computer

The objective of this paper is the development of a multi-criteria optimization framework for the simultaneous calibration of demand and supply parameters in DTA. The presented approach estimates origin-destination (OD) flows and calibrates the driver behavioral and route choice parameters in a complex network modeled in Paramics microscopic traffic simulation model. Genetic Algorithm (GA) is chosen as the solution method for solving the stochastic nonlinear optimization problem. A high-performance computing cluster is used to run GA in parallel computer processing engines. The application of the framework on a large case-study network showed that the incorporation of speed data from in-vehicle navigation systems improves significantly the calibration performance in terms of improved accuracy of estimated counts and speed. In addition, the incorporation of speed data makes the calibration problem less dependent on the starting OD flows. Finally, the application of a distributed GA was shown to significantly reduce the computational time of the calibration of DTA systems.