A deployable online dynamic traffic assignment (DTA) model is an essential component in the implementation of advanced traffic management systems. The consistency between estimates from DTA models and the real world traffic conditions becomes a critical operational issue for managing real-time traffic networks.; The consistency problem can arise due to the assumption used in solving DTA problems that network demand, infrastructure and controls are known a priori for the entire planning horizon; since this assumption may be unrealistic for actual networks where network demand, infrastructure and controls are uncertain and change on a real-time basis. Furthermore, the deficiency in realism and accuracy of traffic propagation representation in DTA models can also lead to the consistency problem.; In this dissertation, a solution framework for online DTA based algorithms is developed for managing real-time traffic networks. The rolling horizon architecture of the solution framework ensures that unpredictable variations in demand, supply and control can be adequately accounted for in subsequent stages, so that consistent estimates and predictions of network states can be obtained. In addition to a dynamic user equilibrium based path assignment model, the solution framework includes a hybrid traffic simulation model and a solution algorithm for calibrating time dependent Origin-Destination matrices to match observed traffic counts. The time dependent OD matrix calibration algorithm is formulated as a mathematical program with equilibrium constraints where a variational inequality constraint is included to enforce the user equilibrium flow pattern.; By integrating the hybrid traffic simulation with the time dependent OD matrix calibration model, the proposed solution framework has the potential to address the consistency problem in online deployment of DTA models. It provides in real-time, consistent network state estimates and predictions by assigning the calibrated time dependent OD matrix to the network and propagating the vehicles through under the most up-to-date network conditions. The testing results show that the solution framework improves the consistency of online DTA models by efficiently producing better estimates and predictions for density and speed on network links with real-time traffic information. The proposed solution framework can be a useful analysis tool for testing and evaluating new concepts, algorithms, and technologies in research and development of Intelligent Transportation Systems (ITS) applications.
展开▼