A Threshold-Based Real-Time Incident Detection System for Urban Traffic Networks

Abstract

As incident detection on a typical busy urban road link or intersection still demands more efficient algorithms, this paper introduces a methodology that can be used to characterize the various traffic patterns (incident or no incident) using typical link passage detectors. Offline urban incident scenarios are generated using a microscopic simulation model assuming varying traffic link flows, signal green phase and cycle times, link lengths. Similar scenarios are also generated for non-incident cases. Three detectors were assumed on each link to extract traffic measures. Comparative numerical statistical analyses were conducted to identify the traffic measures (such as the average speed and flow) that are likely to be affected by the incidents. And further analysis was conducted to quantify the most probable thresholds to be used in the proposed urban incident detection model. The proposed model is validated using simulation data. The performance of the proposed model is assessed using dynamic performance indicators such as the success rate of detecting an incident at a specific cycle time, and the false alarm rate.