This paper specifies the overall architecture of the Real time Traffic Prediction System (RTPS) as proposed in Drive project no. V2036 "DYNA". This system is to model and predict traffic conditions on a highway network in real time. These predictions can be used to provide information on congestion to a traffic operator and to car drivers. The system receives traffic data such as traffic flow, average speed and average detector occupancy in real time from the Rijkswaterstaat Monitoring Casco. Additionally, a database of historic traffic information will be used. The RTPS contains several submodels: a statistical model (STA) for filtering "noisy" traffic data, and for very short term predictions (prediction horizon: 5-15 min.), a dynamic traffic assignment model (DTA) for short term predictions (prediction horizon: 15-60 min.), and a real time Origin Destination (O-D) matrix estimation model (ODME) to provide the DTA with the necessary O-D information. Predictions from the STA and the DTA model are merged where their respective time horizons overlap. The present and predicted traffic conditions will be displayed for evaluation by the traffic operator, who can take appropriate action if required. The real time aspect of this system is reflected in the need for timing and reliable synchronisation of the various model components. A multitasking environment will be required for implementation of the system.
Abstract