Supervisory Control with Partial Observations
In discrete-event systems it is often not realistic to assume that all events of such a system are observable. This is either because some events are typically not directly observable from their very nature (such as failure events, internal hidden events, etc.) or it is too costly to have sensors and observe every event that can occur in the system. Therefore, supervisory control with partial observations has been developed to cope with the additional difficulty of having unobservable events in discrete-event systems. The theory, including the concept of a deterministic observer automaton, necessary and sufficient conditions on a specification language to be achievable as the language of the closed-loop system, formulas and algorithms for computation of sublanguages satisfying these conditions, are presented and illustrated by several examples.
Unable to display preview. Download preview PDF.
- 3.Cassandras, C.G., Lafortune, S.: Introduction to Discrete Event Systems, 2nd edn. Springer (2008)Google Scholar
- 7.Griffin, C.: Decidability and Optimality in Pushdown Control Systems: A New Approach to Discrete Event Control. PhD Thesis. Penn State University, USA (2007)Google Scholar
- 13.Schutzenberger, M.P.: On the definition of a family of automata. Information and Control 4, 145–270 (1961)Google Scholar
- 16.Takai, S., Ushio, T.: Effective computation of an L m(G)-closed, controllable, and observable sublanguage arising in supervisory control. In: Proc. of 6th Workshop on Discrete Event Systems, Zaragoza, Spain (2002)Google Scholar
- 17.Wonham, W.M.: Supervisory control of discrete-event systems. Lecture Notes University of Toronto (2011), http://www.control.utoronto.ca/DES