摘要: |
We develop analysis and control synthesis tools for dynamic traffic flow over networks. Our analysis relies on exploiting monotonicity properties of the dynamics, and on adapting relevant tools from stochastic queuing networks. We develop proportional policies for traffic signal control, which are decentralized, and minimalist in that they require information only about local queue lengths. Hence, they are advantageous in the immediate aftermath of traffic incidents that cause significant changes in turning ratios and flow capacities on the links. Using dynamic traffic assignment framework, we develop convex formulations to compute variable speed limit and ramp metering controls over control horizon to optimize objectives, such as total travel time, which can be cast as convex functions. We also provide adaptation and convergence guarantees for distributed optimization techniques for these computations. The results are illustrated in case studies developed for Los Angeles sub-network in PTV VISSIM. |