Modeling and Control of Production Networks

In order to respond to events like maintenances or machine breakdowns, production networks need to be transferred between several operating points. Inspired by classical control theory, we provide a two degrees of freedom control structure for production networks. Besides a trajectory planning unit generating reference signals, it contains a tracking controller. We provide an agent-based modeling procedure using Petri nets in order to incorporate the discrete event nature of production networks into the trajectory planning unit. As they exhibit memory character, some of the Petri net places can be associated with states in the sense of control theory. Based on them we derive a meaningful state space model from the Petri net description. It is used to design the tracking controller which compensates disturbances proximate to the reference trajectory. Since the latter takes account for discrete events, the controller can be designed using classical control theory. In this way, we integrate planning and control of production networks using a combination of discrete event systems and time driven systems in one control structure avoiding the complexity of hybrid system formulations.