Modeling and clustering network-level urban traffic status based on traffic flow assignment ratios

Abstract

The detected traffic data for single point or link cannot satisfy the needs for network-level traffic status information with the rapid development of the traffic control and guidance systems. This paper proposed a modeling and clustering method for network-level urban traffic status based on the dynamic traffic flow assignment ratios. The traffic assignment ratio matrix model integrates traffic status, topology and relation between links, with the dynamic traffic assignment ratios estimated by Linear Programming. The network-level traffic status is clustered by Self-Organizing Map and the typical patterns are discovered. The experiment proves the efficiency and applicability of this method for network-level traffic status modeling and analyzing.