Predictive Algorithm Improvements for a Real-Time Ramp Control System

The purpose of this project was to determine the feasibility and evaluate the usefulness of a predictive ramp-metering algorithm, an algorithm that anticipates bottlenecks (a bottleneck being a reduction in the traffic capacity of the freeway) one to two minutes before their occurrence. The predictive algorithm was tested on-line in the Washington State Department of Transportation's ramp-metering central computer. 

The predictive algorithm's accuracy in predicting bottlenecks on-line was very good, with a correct prediction rate of almost 80 percent. The measured increase in volume and decrease in occupancy during a portion of the morning peak period showed that the predictive algorithm reduced the number and/or severity of bottlenecks on the freeway test section.

 

Publication Date: 
Sunday, September 1, 1991
Publication Number: 
WA-RD 213.1
Last modified: 
10/12/2016 - 15:41
Authors: 
Nancy L. Nihan, David B. Berg.
Originator: 
Washington State Transportation Center (TRAC)
Number of Pages: 
55
Subject: 
Accuracy, Algorithms, Bottlenecks, Forecasting, Freeways, Peak periods, Ramp metering, Real time control, Traffic flow.