Estimation and prediction of dynamic origin-destination (O-D) demand and system consistency control for real-time dynamic traffic assignment operation

The development and deployment of Intelligent Transportation Systems (ITS) are intended to increase the efficiency of existing surface transportation systems through the use of advanced computing, electronics, and communications technologies. Advanced Traveler Information Systems (ATIS) and Advanced Traffic Management Systems (ATMS) are two key components of ITS. Dynamic Traffic Assignment (DTA) is a core methodology in the implementation of ATIS and ATMS. Two critical support functions for the real-time operation of the DTA capabilities are to generate the dynamic origin-destination (O-D) demand for the assignment decisions, and to ensure consistency of the assignment model values with the actual observed traffic conditions on a real-time basis.; This dissertation develops and implements a methodological framework for the dynamic O-D demand problem under real-time operational environment for large network applications. A general transformation is introduced to reduce the dimensionality inherent in conventional formulations of the dynamic O-D demand problem, resulting in much smaller number of unknown variables in the problem solution. A polynomial transformation is hence implemented in combination with the real-time Kalman Filtering model to guarantee an over-determined process. The developed formulation and associated algorithmic procedures result in computational performance that is compatible with quasi real-time computational requirements for large network applications, which are currently beyond the reach of the existing formulations. The proposed general transformation could also be applied to other approaches to enhance their applicability to large networks. Experimental results are performed on a small test network and two large networks, providing convincing evidence regarding the accuracy and robustness of the developed methodological procedure.; The second methodological capability developed in this dissertation consists of a diagnostic architecture and associated procedures for system consistency control in real-time DTA operation. The function of the system consistency control is to monitor the deviations between the DTA-simulator and the actual transportation system, generating feedback controls both to the O-D estimation/prediction input and to the DTA-simulator on a real-time basis. A real-time DTA system is integrated in a distributed computing environment to perform dynamic O-D estimation and prediction as well as the system consistency control. The system incorporates two algorithmic procedures to control and adjust the O-D demand input and the DTA-simulator for on-line DTA operation.