Predictive Networked Control of Discrete Event Systems
Shaolong Shu, Feng Lin · IEEE Transactions on Automatic Control · 2016
In this technical note, we continue the work of [6] to investigate the networked control problem of discrete event systems. The problem is to find a networked supervisor so that the language generated by the supervised system is equal to a given specification language K. While the networked supervisor in [6] is not predictive, we propose a predictive networked supervisor in this technical note, which predicts the impacts of communication delays and losses in the control channel in determining control actions. We derive the existence condition for a predictive networked supervisor. We show that predictive supervisors are better than non-predictive supervisors and no other supervisors are better than predictive supervisors. We further show that the predictive supervisor is least restrictive among all supervisors that generate language K.