Travel Time Based Freeway Flow Estimation: A Case of Two Detector Problem

摘要

In this paper, a special case of two detector problem is considered. Given counts at one end of asegment and segment travel times, it is desired to estimate counts at the other end of the segment.Solution to this problem can be helpful in monitoring health of counting stations. Also, it canpotentially produce count estimates at points where no count stations exist. Based on thesmoothness assumptions and first-in first-out (FIFO) principle a well-known result ismanipulated to solve the problem. Proposed solution consists of a lag one linear filter on theknown counts time series in which the weight factors are a function of travel times. In derivingthese solutions a distinction between anticipative and retrospective travel times are made.Anticipative travel times are used to estimate upstream counts, and retrospective travel times areused to estimate downstream counts. A two week long count dataset on a segment of I-295(Anacostia freeway) in the District of Columbia is used to evaluate the performance of theproposed method. A discussion on the accuracy and quality of solutions and directions of futurework in this area is provided.