Yorgos Stephanedes, Eil Kwon, Kaikuo Chang, Nikos Vairamidis
Report no. Mn/DOT 1994-13
Managing freeway congestion requires an integrated approach involving demand-responsive ramp metering, incident management and driver guidance. While a freeway network acts as a system, i.e., conditions on any part affect other parts in the network, the state of the art in real-time freeway management has not reached the point where comprehensive, network-wide optimal control schemes are automatically generated and implemented through on-line optimization and coordination of various control actions. A major difficulty lies in the lack of efficient computational algorithms implementable for on-line optimization, and the lack of accurate on-line predictors, that can predict traffic demand and diversion in freeway networks. As a result of the above limitations, most traffic responsive metering systems, such as the Twin Cities freeway control system, employ automatic rate-selection procedures. These procedures select the most appropriate metering rates for a ramp from a pre-determined library using the information received from loop detectors on the main freeway, upstream and downstream from the ramp. Although this method provides a degree of self-adjustment to prevailing traffic conditions, the lack of an efficient analysis tool to evaluate and update the key components of the control, i.e., thresholds and rate-libraries, significantly restricts the effectiveness of control.
Download or order