Coalitional Model Predictive Control on Freeways Traffic Networks

This paper discusses the application of coalitional model predictive control (MPC) to freeways traffic networks, where goal is reducing time spent by drivers through a dynamic setting variable speed limits (VSL) and ramp metering. The prediction METANET used represent flows evolution. system behavior objective function lead non-convex non-linear optimization problem, which can only be solved in centralized fashion for small networks. underlying motivation this continued advance clustering methods large-scale spatially distributed systems. global freeway partitioned into set coupled sub-stretches, turn are assigned different agents involved problem. These local controllers dynamically assemble coalitions take coordinated measures. In work, top-down approach considered: bottom layer consists that compute VSL ramp-metering across time; supervisory changes periodically information exchange structure promote those bring greater performance system. way, balance sought between optimality efficiency. Finally, simulated on stretch cooperation with adjacent sub-stretches allowed.