Date of Award

Summer 2015

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Civil/Environmental Engineering

Committee Director

Mecit cetin

Committee Member

MonWo Ng

Committee Member

Rajesh Paleti

Abstract

Probe vehicles instrumented with location-tracking technologies have become increasingly popular for collecting traffic flow data. While probe vehicle data have been used for estimating speeds and travel times, there has been limited research on predicting queuing dynamics from such data. In this research, a methodology is developed for identifying the travel lanes of the GPS-instrumented vehicles when they are standing in a queue at signalized intersections with multilane approaches. In particular, the proposed methodology exploits the unequal queue lengths across the lanes to infer the specific lanes the probe vehicles occupy. Various supervised and unsupervised clustering methods were developed and tested on data generated from a microsimulation model. The generated data included probe vehicle positions and shockwave speeds predicated on their trajectories. Among the tested methods, a Bayesian approach that employs probability density functions estimated by bivariate statistical mixture models was found to be effective in identifying the lanes. The results from lane identification were then used to predict queue lengths for each travel lane. Subsequently, the trajectories for non-probe vehicles within the queue were predicted. As a potential application, fuel consumption for all vehicles in the queue is estimated and evaluated for accuracy. The accuracies of the models for lane identification. queue length prediction, and fuel consumption estimation were evaluated at varying levels of demand and probe-vehicle market penetrations. In general, as the market penetration increases, the accuracy improves. For example. when the market penetration rate is about 40%, the queue length estimation accuracy reaches 90%. The dissertation includes various numerical experiments and the performance of the models under numerous scenarios.

DOI

10.25777/wb2d-ss57

ISBN

9781339125909

Share

COinS