Intelligent agent-based control systems in manufacturing

The paper presents a real-time supervisory control architecture for large-scale manufacturing systems. The architecture is based on autonomous intelligent agents at the shop-floor level, organized in dynamical, reconfigurable structures depending on the manufacturing system´s state and goals. The supervisory level is introduced in order to evaluate and improve the performances of the control architecture. Functioning of the whole architecture is event-driven and based on rules that the supervisor should be capable of completing. The formal support for performance evaluation and control policies synthesis is DES