ESTIMATION OF ORIGIN-DESTINATION TRIP MATRIX FROM LINK TRAFFIC INFORMATION.

摘要

The problem of estimating an Origin-Destination (O-D) trip matrix from link traffic information is analyzed. The use of some systems behavioral properties regarding trip-making in an urban area is felt necessary in order to derive a better solution for the O-D estimates as much as the algorithmic efficiency and solution accuracy are concerned. An algorithm for determining a unique solution for O-D estimation is developed and several experiments were run on its performance to show that the model is superior in many aspects to other currently available techniques.;The major feature of this study is regarding the use of information contained in the trip probability distribution curve of a network. The model develops a set of independent and consistent linear equations from this information. These equations are used to relate the O-D trips to a sum, the way the link volumes are related to the O-D trips. This approach increases the number of independent equations to be solved for the unknown O-D volumes. Experimental results have shown that this improves the accuracy of the O-D estimates in most cases.;A theoretical study conducted on the structure and performance of the model indicates that the model is a unification of the favorable properties of some governing theories of urban and regional modeling.;The algorithm is based on a scheme to solve a set of linear equations derived from the link traffic information. These equations relate the various unknown O-D volumes to the observed link counts according to a traffic assignment and the probability of occurrence of a trip of known duration. As expected in a real-world transportation network, the number of these equations is much less than the number of the unknowns. The scheme solves for the unknowns and generates an O-D trips matrix such that the transportation systems characteristics are maintained. Case studies have shown that the method is applicable to any real-world network.