Abstract
Ramp metering on freeways is one, among many others, application in which a proportional-integral (PI)-Controller has been successfully used to keep the system at a desired state. However, the design becomes challenging if the system presents a dead time restricting the stability region and affecting performance. The cause of the dead time is the distance between the on-ramp and the lane drop. A change in the control action affects the system after the time taken to vehicles travel this distance. There are a vast literature in the control theory addressing the effects of dead time and techniques to overcome it, but it is not to ramp metering control. In this study, a dynamic model of the system is presented along with ALINEA, an established Ramp Metering control algorithm based on a PI-Controller. The authors show that by incorporating a Smith Predictor into ALINEA, the stability region becomes larger, insensitive to the dead time, and can be found analytically. Numerical experiments confirm the analysis and that the proposed controller still presents a larger stability region even when there are modeling errors, allowing a faster response to disturbance.