Identification of differentiated toll sections on expressways using multi-source data

DUAN Lizhen; HE Mingwei; HE Min; SHEN Ke

doi:10.11860/j.issn.1673-0291.20230161

Abstract：In the context of the widespread implementation of differentiated tolls on expressways, a method for identifying differentiated toll sections using multi-source data is developed. Firstly, an indi-cator system for identifying these sections is established based on data from expressway gantries, toll station entrance and exit flows, and automatic traffic flow observation stations on national and provincial roads. This indicator system is constructed across 5 dimensions: the level of socio-economic development around the expressway section, the degree of network integration, traffic operation char-acteristics, traffic transferability, and traffic load balance. Then, a model for identifying differentiated toll sections on expressways is developed using the k-means clustering and Classification and Regression Trees (CART) algorithms. Lastly, the segmentation rules and key characteristic indicators of differentiated toll sections are derived from an empirical study of 130 expressway sections in Yunnan Province. The re-search findings reveal that the key characteristic indicators influencing the segmentation of differentiated toll sections include the network density of expressway sections, the node degree of toll stations, the deviation rate of travel costs, the hourly congestion level of road sections throughout the day, and the number of al-ternative routes. According to the decision tree segmentation rules, differentiated toll sections can be classi-fied into 3 types: upward differentiated, unsuitable for differentiation, and downward differentiated, which align with the real-world toll operational situations of expressways.