Using Field Data to Improve Model Accuracy:Application of the Robertson Dispersion Model to High-Resolution Signal Event Data

摘要

This paper presents a scalable methodology for automatically measuring platoon characteristicson links between signalized intersections. This methodology estimates the Robertson dispersionmodel parameters that characterize platoon travel time and platoon distribution along a link. TheKolmogorov-Smirnov statistical test was used to assess how well the model parameterscharacterize the platoon arrivals measured by the high-resolution event data. The methodologywas tested on a corridor of nine intersections containing eighteen coordinated links using datacollected over a three-week period. Very consistent results were obtained for link travel times,however in many cases the observed link travel times were faster or slower than the postedspeed. The distributions of the dispersion parameter show considerable variation across theanalysis period as well as across the system. The test corridor presents an interesting situationwhere shorter links experience greater dispersion than longer ones. The results suggest thatplatoon dispersion is more sensitive to driver behavior characteristics related to roadsideenvironment than to distance traveled along a street. The methodology could be easily integratedinto modern control systems that log controller event data to self-calibrate adaptive controlstrategies, help engineers prioritize links for coordination, and provide guidance on designparameters.