Optimal Modular Control of Discrete Event Systems with Distinguishers and Approximations

This paper proposes algorithms for supervisor synthesis in discrete event system models with distinguishers. Distinguishers are special components responsible to select an enabled event from a group of related refined events. They are a helpful modelling tool, but their use increases the state space and makes supervisor synthesis more difficult. The paper shows how general algorithms for modular or compositional synthesis can be enhanced by considering the special properties of distinguishers. This gives rise to systematic algorithms that compute least restrictive controllable and nonblocking supervisors, while working with only a part of the distinguisher model. A worked case study of a bottling plant demonstrates the efficacy of the approach.