Receding Horizon control with two planning horizons for a class of discrete event systems with real-time constraints
Lei Miao · 2016
There exists a class of discrete event systems involving the control of tasks with real-time constraints. It has been shown that in an on-line setting where event times are unknown, Receding Horizon (RH) approaches can be utilized to control such systems. In order to guarantee the real-time constraints, worst-case estimation needs to be used in the planning horizon of the RH controller. The direct effect of worst-case estimation is increased cost, which is highly undesirable, especially when the tasks' arrival rate is low. In this paper, we design a new RH control mechanism with relaxed worst-case estimation. We prove that using the new approach, feasibility can still be guaranteed. Simulation results show that the performance of the new RH controller can be over 50% better than that of the original RH controller.