Applying Derived Distribution Method to Microlevel Driving Behavior Characteristics to Quantify Uncertainties in Traffic Stream Flow and Density

Abstract

The flows and densities of traffic streams play an important role in defining the performance of roadways and corresponding improvement strategies. Traffic flows and densities are the outcome of complex psychophysical actions of drivers. Actions performed by the drivers while driving can be quantified in terms of the headway and/or spacing that they maintain with respect to the vehicles they follow. However, the inherent randomness that exists in human driving behaviors results in random headway and spacing, which leads to uncertainties in predicted traffic flows and densities. As a result, it is important to quantify these uncertainties, because they play an important role in proposing improvement strategies. In this study, a derived distribution method–based uncertainty quantification of traffic flows and densities is proposed; it involves the modification of deterministic flow–headway and density–spacing relationships into probabilistic ones. Analytical expressions were derived for the probability distributions of flows and densities, given the headway and spacing distributions, respectively, which are conditional on velocities. The estimation of the distribution parameters and the validation of the proposed approach were carried out using the Next Generation Simulation (NGSIM) trajectory dataset. The results indicated that the proposed analytical distribution models represented empirical field observations quite accurately.