Predicting Freeway Travel Time under Incident Conditions

Abstract

Prediction of corridor travel time is essential to achieving the promise of advanced traveler information systems. Although much research has been done on travel time prediction, procedures for practical prediction of travel time under incident conditions are lacking. This paper presents a framework for online prediction of freeway corridor travel time under incident conditions on the basis of inductive loop detector and incident data. The core component is a model for prediction of corridor travel time under incident conditions that produces estimated physical queue length at various time intervals. Testing shows that the method offers enhanced accuracy in prediction of corridor travel time compared with those methods that do not consider the impact of incidents.