This thesis develops a dynamic arterial network incident detection system for real-time applications. The system is unique in its use of an event- driven simulation structure, along with a self-learning mechanism and monitoring function. These features enable a traffic control center to minimize response time and to take proactive actions prior to the formation of congestion. This kind of system can also be integrated with adaptive traffic signal control and route guidance systems to maximize a network's operational capacity during recurrent and non-recurrent congestion. The system consists of three components: a dynamic traffic flow prediction module, an incident identification and assessment module, and an intelligent control component.