Assigning jobs to agents by means of Petri net-based models

Many times agents have to realize several jobs. However, only one job can be processed by one agent in the same time. The problem when and how to assign the jobs to agents is solved in this paper. Both agents and jobs are modelled by place/transition Petri nets (P/T PN). The task of assigning the jobs to an agent is understood to be a task of the jobs cooperation. To coordinate the jobs a supervisor is synthesized. The supervisor forces a desired job cooperation strategy to utilize the agent manufacturing capacity as exhaustively as possible. The supervisor synthesis is based on discrete-event systems (DES) control theory and performed by means of P/T PN. In case of assigning several jobs to the agent(s) it is necessary to force the job priorities. The P/T PN firing count vector (Parikh's vector) is utilized here to ensure the desired priorities. Then, the time parameter is added to the final P/T PN structure. In such a way timed PN (TPN) arises. It allows to analyze the time circumstances. Namely, processing of each job by the agent takes some time.