Stochastic Network User Equilibrium and Traffic System Evolution Based on Reference-Dependent Utility Theory

Abstract

This paper applies the reference-dependent utility theory (RDUT) to model traveler's route choice behaviors under travel time variability and develops a user equilibrium (UE) model based on RDUT for stochastic traffic networks. The proposed model explicitly considers both the absolute utility (or consumption utility) and the relative utility (or gain-loss utility) in the travelers' path choice decision procedure. The former is determined by the stochastic path travel time while the latter is measured by the actual path travel time relative to a reference time point. Subsequently, the RDUT-based UE model, which can be equivalently formulated as a variational inequality problem and solved by a heuristic algorithm, is employed to explore how risk aversion and reference-dependent preference jointly determine network equilibrium patterns. Both the features and applicability of the proposed RDUT-UE model and the designed solution algorithm are demonstrated in two numerical examples. This paper further establishes an RDUT-based dynamic traffic system which incorporates commuters' learning, choosing, and renewing process in dynamic path choices and captures the day-to-day dynamic traffic flows. Another numerical example is presented to show how the dynamic traffic system evolves to the UE status.