MODELING TRANSPORT DEMAND BASED ON MOBILE COMMUNICATION DATA

Abstract

Using the data of mobile operators to improve the estimate of the prior matrix can affect the structure and improve the quality of the posterior matrix when predicting traffic flows. Such data are important for improving the quality of the results obtained using transport models, since they can provide much larger amounts of information and continuous measurements. They contain information about the connection of a specific device with a cell tower, which makes it possible to determine the location of the subscriber. Time series analysis of these devices allows you to reconstruct the route and understand the daily dynamics of movement. Data from mobile devices can be used to obtain various indicators: the number of trips, work schedule, travel time.

However, both data sources (mobile operators and travel surveys) have strengths and weaknesses that can lead to modeling errors. Comparison of these data sources gives an idea of the reliability of the correspondence matrix, that is, the combination of the two types of data will improve the quality of the prior matrix for transport modeling. The study combined survey-based travel time distributions and mobile network data to improve the structure of the traditional prior matrix. For the study, a multimodal transport model is used, which includes a large number of transport zones, taking into account various types of road transport (cars and trucks, public transport), using a detailed network of the region. For a correct assessment of trips arriving from outside the region, as well as for the assessment of through traffic, it is necessary to include the network and zones outside the region with less detail. A priori matrix of the region is constructed using demand models and characteristics of riders. Calibration of the synthetic a priori matrix of correspondences by subaccounts of traffic flows gives the posterior matrix. The model takes into account the purpose of equilibrium, based on the functions of the transit time of a section of the road network and traffic congestion.