Recognition of critical links for gating using queue detector data

Abstract

This paper introduces the concept of rolling time occupancy, and presents a method for recognizing critical links for gating, where spillovers will occur or have done already. The key idea of this new method is that, when the queue length becomes sufficient for the end of the queue to approach the queue detector, the speed of vehicles arriving from upstream will be slower than the free flow speed, which implies that a series of rolling time occupancies will all be greater than a particular threshold. Given that the value of a single rolling time occupancy is strongly influenced by buses and other confounders during a minor time interval, the new criteria for critical link identification are divided into two parts: the occupancy threshold for possible congestion, and the number of rolling time occupancies for inevitable congestion. Next, this paper presents the relationship between queue length and rolling time occupancy, using traffic data procured using VISSIM simulations. On this basis, the queue length threshold at which a spillover can be judged to be going to occur or to have already occurred is determined. Finally, using the misjudgment ratio and the trigger postponement amount as evaluation indices, the method proposed in this paper is evaluated under ten different simulation conditions. The results show that the maximum and average misjudgment ratios are 32.32% and 10.11%, respectively; meanwhile, the average postponement amount is less than 50 seconds for all ten simulation conditions.