Improving evolvability of a patient communication control system using state-based supervisory control synthesis

Abstract Supervisory control theory enables control system designers to specify a model of an uncontrolled system in combination with associated control requirements, and subsequently to use a synthesis algorithm for automatic controller generation. The use of supervisory control synthesis can significantly reduce development time for supervisory controllers as a result of the unambiguous specification of control requirements, and synthesis of controllers that-by definition-are nonblocking and satisfy the control requirements. This is especially important for evolving systems, for which requirements change frequently. For successful industrial application, the specification formalism should be expressive and intuitive enough to be used by domain experts, who define control requirements, and by software experts, who implement control requirements and synthesize controllers. This paper defines such a supervisory control specification formalism that consists of automata, synchronizing actions, guards, updates, invariants, and independent and dependent variables, where the values of the dependent variables can be defined in terms of functions of the independent variables. We also show how the formalism enables systematic, compositional specification of a control system for a patient communication system of an MRI scanner. We show that our specification formalism can deal with both event-based and state-based interfaces. To support systematic, modular specification of models for supervisory control synthesis, we introduce state trackers that record sequences of events in terms of states. The synthesized supervisor has been successfully validated by means of interactive user guided simulation