Investigating the Key Factors Affecting the Accuracy of Reidentification of Trucks over Long Distances Based on Axle Measurement Data

摘要

Vehicle re-identification methods can be used to anonymously match vehicles crossing two differentlocations based on vehicle attribute data. In this paper, key factors that impact the accuracy of vehicle re-identification algorithms are investigated. The analyses are performed by employing re-identificationalgorithms to match commercial vehicles that cross upstream and downstream pairs of weigh-in-motion(WIM) sites that are separated by long distances ranging from 70 to 214 miles. The data to support thisresearch come from 17 fixed WIM sites in Oregon. Data from 14 different pairs of WIM sites are used toevaluate how matching accuracy is impacted by various factors including the distance between two sites,travel time variability, truck volumes, and sensor accuracy or consistency of measurements. After runningthe vehicle re-identification algorithm for each one of these 14 pairs of sites, the matching error rates arereported. The results from the testing datasets showed a large variation in terms of accuracy. It is foundthat sensor accuracy and volumes have the greatest impacts on matching accuracy whereas the distancealone does not have a significant impact.