Calibration of a second-order traffic flow model using a metamodel-assisted Differential Evolution algorithm

Abstract

With the increasingly widespread use of traffic flow simulation models, several questions concerning the reliability, efficiency and accuracy of such models need to be addressed convincingly. In general, the most time-efficient traffic flow models are based on the macroscopic approach to describe traffic dynamics. Macroscopic models reproduce the evolution of aggregated traffic characteristics over time and space with respect to observable variables, such as flow and speed, requiring much less computational time, compared to microscopic ones. This work assesses a second-order macroscopic gas-kinetic traffic flow (GKT) model and its numerical implementation using real traffic data from a motorway network in the U.K., where recurrent congestion originated from high on-ramp flows during the morning peak hours is observed. A parallel, metamodel-assisted Differential Evolution (DE) algorithm is employed for the calibration of the model parameters, and numerical simulations demonstrate that the DE algorithm can be a very promising method for the calibration of such traffic flow models.