Coordinated Anti-Congestion Control Algorithms for Diverging Diamond Interchanges
-
2021-04-01
Details:
-
Creators:
-
Corporate Creators:
-
Corporate Contributors:
-
Subject/TRT Terms:
-
Publication/ Report Number:
-
Resource Type:
-
Geographical Coverage:
-
Edition:Final Report Final (July 2019—April 2021)
-
Corporate Publisher:
-
Abstract:The main objective of this research project is to develop data-driven traffic prediction and control algorithms for congestion management of diverging diamond interchanges (DDIs) and their surroundings. The project proposes an algorithm that leverages recent advances in data-driven modeling, namely, partial least squares multivariate regression, in order to address two fundamental issues: (1) how can future traffic flow through a network be predicted from current flow, and (2) how can these predictions be made robust to sensor fault and errors. Measured traffic flow data exhibit frequent missing values, necessitating a robust regression approach that is minimally affected by such outliers. The project results indicate strong predictive correlation between off-nominal traffic conditions in a single day separated by several hours. For example, off-nominal traffic in the morning—e.g., abnormally high flow in one direction—might correlate with off-nominal traffic in the evening caused by abnormally high flow in the opposite direction as drivers reverse their commute. The statistical methodology proposed in this project is able to determine and exploit this correlation. A proposed use case is employing such predictions over several hours in order to preemptively adjust traffic controllers in anticipation of off-nominal conditions. As case studies, the project considers data from the I-285 and Ashford Dunwoody Road DDI and the I-85 and SR 140/Jimmy Carter Boulevard DDI.
-
Format:
-
Funding:
-
Collection(s):
-
Main Document Checksum:
-
File Type:
Supporting Files
-
No Additional Files
More +