Defining Traffic Flow Phases Using Intelligent Transportation Systems-Generated Data

Abstract

The characteristics of traffic flow under various conditions have been studied by many. Particularly, researchers use the concept of flow phases to distinguish traffic characteristics under various conditions. In this study, a data-clustering methodology is developed to define the flow phases using continuous traffic data obtained through detectors. Such continuous data provide rich information on traffic characteristics. It is shown that between the two clustering variables, density has a more significant impact on the clustering results than speed. The fundamental relationships between traffic parameters are then analyzed based on the flow phase definition.