Estimation of Dynamic Network Origin-Destination Flows Using Measured Traffic Counts

This paper presents a revised approach for estimating time-varying Origin-Destination flows in urban networks. This approach lend itself to a formulation as state-space model, and constructs an additional set of system observation constraints with estimated intersection turning flows to improve the system observability. Since there may exist errors in DTA or simulation models, this approach further utilizes path flow deviations as state vectors, instead of O-D deviations in past researches, and accounts for the inherent errors in assignment/mapping matrix caused by DTA or traffic simulation. The revised model is evaluated using simulation experiments. The reported examples have clearly indicated that the proposed approach is quite accurate and robust.