Supervisory control of DES with distinguishing sensors

This paper deals with supervisory control of discrete event systems with the introduction of sensors that distinguish certain occurrences of observable events. Distinguishing sensors are formally represented by an entity named distinguisher that maps the original events into disjoint sets of refined events and specify which of them could actually occur according to the evolution of the system. The control problem is formulated to incorporate distinguishers in order to have specifications that are presumably more easily modeled in the refined alphabet. It is proved that minimally restrictive nonblocking supervisors can be obtained by incorporating the distinguisher in the plant model. Further it is shown that suboptimal solutions can be efficiently obtained from an external approximation of the refined plant. Conditions for optimality are provided together with a polynomial complexity algorithm to verify them. A manufacturing system example illustrates the results.