Dynamic origin-destination estimation in congested networks

In this paper three conditions for unbiased origin-destination (OD) estimation in congested networks are presented. A first condition deals with the choice of the network loading model. It is shown that an incorrect representation of queuing leads to an incorrect interpretation of information from the detectors, and thus leads to biased results. A second aspect is the relationship that is used in the minimization problem. In many OD estimation methods a linear relationship expressed by the assignment matrix is chosen. However, this relationship assumes separability of the link flows to the OD flows, which can lead to biased results when dealing with congested networks. A final source of error is the presence of multiple local minima in the objective function. An important cause of local minima is an incorrect interpretation of the information from the detectors. Since flow measurements can correspond to two distinguished traffic regimes, free flow and congestion, when the assigned initial OD matrix reproduces a different traffic regime at the detectors compared to reality, the measurements can be interpreted incorrectly. Finally a case study is presented that illustrates the bias that may be introduced if not all three conditions are met.no ISBNstatus: publishe