Application of Pattern Recognition to Forecast Congested Conditions on the Freeway for Use in Ramp Metering. Volume 2. (Freeway and Ramp Real Time Forecasting)

摘要

Traffic congestion is one of the most important issues facing us today. The study examines how traditional supply side solutions have failed to solve the problem, and why demand management strategies, such as ramp metering, are becoming more popular. Volume II analyzes the deficiencies in the ramp metering system currently employed by Washington State Department of Transportation (WSDOT), and briefly examines some past approaches of real-time control at the on-ramps. This research aimed to develop a pattern recognition model capable of forecasting traffic conditions on the freeway 1 to 3 minutes ahead of time. Both parametric and non-parametric approaches to pattern recognition were reviewed, and a model based on the principles of statistical pattern recognition was developed.