This paper investigates the ability of utilizing a Petri net along with an artificial neural network to design an intelligent logic controller for a discrete event flexible manufacturing system (DEFMS). The artificial neuro-Petri net (ANPN) model can pick up advantages of Petri nets and neural networks, not only as system states and their change, but also as system input –output mapping and to avoid the deadlock condition of the conventional Petri net. The proposed architecture, the mathematical formalism and the design procedure of this controller are presented. The proposed design procedure has been applied to design and implement a logic controller for an industrial application. The performance of the proposed controller has been evaluated and tested by simulation and experiment. A comparison between the proposed ANPN based controller performance and a recent Petri net based controller performance is presented. It has been indicated that the ANPN based controller achieves its function properly, is capable to avoid a deadlock situation and has less complexity merit, intelligence and flexibility superior to the commonly Petri net based controller. |
