Distributionally robust ramp metering under traffic demand uncertainty

Abstract

This study presents a distributionally robust optimization model to address the ramp metering problem with uncertain traffic demand flows. The aim of this model is to minimize the total travel delay of the system based on the macroscopic cell transmission model (CTM) of traffic flow. In our model, the only required data is the partial distributional information of stochastic demand flows. Using the Worst-Case Conditional Value-at-Risk (WCVaR) constraints to approximate the distributionally robust chance constraints, the proposed problem can be conservatively approximated as a semidefinite programming (SDP), which is computationally efficient. The performances of our proposed model are illustrated by practical applications. Experimental results show that the distributionally robust control strategy can achieve reliable performances over a range of uncertain scenarios.