Analysis of a Maximum Throughput Ramp-metering Control

Abstract

Ramp metering is critical for managing freeway traffic, yet no optimal control with theoretical guarantees exists in the current literature. With increasing market penetration rates of autonomous vehicles (AVs), methods lacking theoretical guarantees risk unpredictable traffic conditions. Maxpressure control has proven throughput-maximizing properties at signalized intersections with strong theoretical and experimental validation. However, it cannot be directly applied to ramp meters, particularly in multi-class traffic with evolving AV penetration. This paper develops a new max-pressure-based ramp metering control that ensures maximum throughput while considering multi-class traffic. We prove that if any theoretical controller can achieve maximum throughput, our controller can do so as well. Using a multi-class cell transmission model, we show how to approximate parameters for practical implementation while preserving the controller’s throughputmaximizing properties.