Optimum Counting Location to update Origin-Destination Matrix Using Bayesian inference (Case Study: city of Isfahan)

The Origin-Destination (OD) matrix is essential to analyze traffic flow in an urban transportation network. OD matrices are obtained through either direct or indirect methods. Since the direct methods are costly, indirect methods have lately been more popular among researchers for the purpose of correcting and updating OD matrices. A very important indirect method is mathematical programming model using observed volume on network links. We should notice that the quality of the corrected OD matrix depends on the accuracy of the input data such as the initial OD matrix and the links which selected for traffic volume data collection. Simultaneously, scarcity of resources, limits number of links for data collection. As a result, it is necessary to select a proper set of links so that we can obtain the maximum possible information to be used for OD matrix correction. A popular approach for this purpose is Bayesian inference and applying Bayesian networks. A major problem with this approach is being very time intensive for medium and large size networks. The current study provides an efficient method of identifying important links for traffic volume data collection in large networks using two techniques: Firstly, identifying and eliminating non important links, and secondly, reducing the less important OD pairs. Finally, the proposed method is applied to Sioux Falls city and Isfahan metropolitan networks to specify the optimum set of network links for traffic volume data collection and therefore the OD matrix estimation.