Modeling and State Estimation for Supervisory Control of Networked Timed Discrete-Event Systems and Their Application in Supervisor Synthesis

Abstract

This paper considers modeling and state estimation of timed discrete-event systems (TDESs) with communication delays and losses. In the previous framework, the specified closed-loop system language may contain physically impossible strings, i.e., strings never occur in practice. To exclude these physically impossible strings, this paper presents a new modeling framework for supervisory control of networked TDESs. Specifically, we first analyze how to model and update the observation channel and the control channel under communication delays and losses. We then construct an automaton that explicitly describes the interaction process between the plant and the supervisor over the observation channel and the control channel so that the language of the closed-loop system can be specified. Compared with the existing work, the proposed framework can model the “dynamics” of the closed-loop system more accurately. Under the proposed framework, we further estimate states of the closed-loop system under communication delays and losses. The state estimation is implemented on-the-fly and bases on only the information of observations and controls in history. As an application of the proposed modeling and state estimation approaches, we finally extend the existing approaches to find all the admissible and safe supervisors of networked TDESs.