Open Access

Cooperation of agents in complex systems based on supervision


Complex systems consist of many cooperating devices. To have a transparent view on the system structure, as well as on the structural interconnections and cooperation of the subsystems, it is useful to synthesize the complex systems systematically in a prescribed order, even in analytical terms (if possible). The supervision of the subsystems seems to be a very suitable approach to accomplish these demands, and consequently it makes the complex systems diagnostics easier. The substantial agents (i.e., the agents of material nature − e.g., devices like particular production lines, robots, numerically controlled machines, etc.) can be coordinated and forced to cooperation by means of efficiently synthesized supervisors. The cooperation process has the character of DES (Discrete-Event Systems), because any system (including continuous systems), has minimally two discrete states - idle and working. DES control theory can be successfully utilized in supervisor synthesis. There are several approaches to modeling the agents and the process of supervisor synthesis. The Petri net-based approach is one of them. Place/Transition Petri Nets (P/T PN) are used here for modeling the behaviour of particular agents, as well as in the computational parocess used for the supervisor synthesis. Two main methods of the P/T PN-based supervision will be used, namely (i) the supervision based on the place invariants (P-invariants) of P/T PN, utilizing only the state vector during the supervisor synthesis, and (ii) the extended supervision utilizing not only the state vector, but also the control vector and Parikh’s vector. The efficiency of the proposed approach is illustrated in a case study.

Publication timeframe:
4 times per year
Journal Subjects:
Computer Sciences, Information Technology