Modular Supervisory Control of Large Scale Discrete Event Systems

This paper presents an approach for reducing computational efforts in the synthesis of modular supervisors for discrete event systems. Most of large scale systems are represented by the composition of several smaller subsystems. To avoid the design on a single, often large plant, modular supervisors are synthesized on local models, which are constructed from those subsystems that are related to the given specifications. A sufficient and necessary condition for the nonblocking and optimality of the conjunction of multiple local supervisors is then presented. It is shown that this approach offers reduction in computational efforts when compared with previous approaches.