A study on the generalized TFlowFuzzy O-D estimation

It is an essential prerequisite to know the number of trips made between origin destinations (O-D) pair of a network. Typically, O-D matrices can be estimated by applying a series of travel demand models, based on expensive and time consuming widespread field studies. From 1970s, some methods have been proposed and widely used for directly estimating O-D matrices from observed traffic counts and an initial O-D matrix. These methods could be considered as the inverse of the traffic assignment problem in which an O-D matrix is found that reproduces the observed traffic counts when assigned to the network. TFlowFuzzy (TFF), as one of the most practical and realistic O-D estimation methods models observed traffic counts as imprecise values based on fuzzy sets theory. In this method, a key element in the estimation of a trip matrix is the "route choice proportions" for each link. In the TFF method, route choice proportions are computed only once and held constant for all iterations in the matrix estimation process which is a very simplifying assumption. Recently a generalized version of TFF is proposed in which the route choice proportions are updated successively at each iteration. This modification considers the route choice proportions endogenously. This allows for a more precise estimation of the O-D matrix. In this research the proposed modified TFF is implemented on Tehran provincial intercity network. Results indicate a considerable improvement in the goodness of fit of the modified TFF in comparison to the traditional algorithm. The goodness of fit is increased by 12 percent. Furthermore, the modified TFF represents better results in descriptive statistics in terms of less error values and intervals. This shows the effectiveness of the proposed algorithm as it can replicate the real data more precisely.