Model Predictive Control for Timed Petri Nets

:This paper is about deadlock-free scheduling problems for discrete event systems. The main contribution is to propose algorithms that incrementally compute control sequences with minimal or nearminimal duration to drive the marking of timed Petri nets (TPNs) from an initial value to a reference one. These algorithms are based on a partial exploration of the TPN reachability graph with an approach inspired from model predictive control. They include perturbation rejection and forbidden marking avoidance and are suitable to track trajectories in real time context when the initial and reference markings are far from each other.