Fuzzy Petri nets and their application in CIME 1

This paper is a short survey about the main fuzzy Petri net models which have been recently developed. We focus on the applicability of such approaches in the eld of manufacturing systems and factory automation. We also point out the main issues concerning the consistency of Petri net theory and these models. 1.1 Why Fuzzy Petri nets are useful for CIME? Factory automation has been addressed at two very diierent levels in the past. Either the organization of the production system, the planning of the workkow, the scheduling of the operations are improved but the elementary operations are not necessarily automated. Or the major part of the eeort is dedicated to the automation of the elementary repetitive sequences of operations by means of Programmable Logic Controllers the management of the production remaining largely based on human heuristic decisions. The rst approach deals with an overall model of the production process. What has to be improved and automated is a complex decision making system where human is present. The data are frequently imprecise; ill-known information has to be taken into account, the environment is poorly structured and unforeseeable (unexpected events are frequent). Not surprisingly, all available Artiicial Intelligence techniques are typically used in this approach, including fuzzy sets and possibility theory 8, 26]. In the second approach, the environment is generally well-known, human is not included in the system to be automated. It is possible to formally specify the control/command sequences by means of Petri nets and to automatically derive programs for Programmable Logic Controllers. The long term trend is towards integrated automation and concurrent engineering which requires to associate and combine these two approaches. It is one of the motivations for so much work concerning Petri nets and logic and more speciically