In: Technical Report 9503, Department of Computer Science, Memorial University of Newfoundland, St.John's, Canada A1B 3X5. 1995. Available at ftp://ftp.cs.mun.ca/pub/techreports/tr-9503.ps.Z.
Abstract: Timed Petri nets are proposed as models of simple and composite schedules for a large class of manufacturing (or robotic) cells. For simple schedules, exactly one part enters and one leaves the cell in each cycle. Net models of simple schedules can easily be derived from the sequences of robot actions. For composite schedules, several parts enter and leave the cell in each cycle. It appears that models of composite schedules can be obtained by composition of simple schedule. A systematic method of deriving all composite schedules is proposed, and decomposition of derived composite schedules into simple ones is presented. It is shown that simple as well as composite schedules can easily be transformed into timed Petri net models. Invariant analysis of timed net models of schedules is used to derive the cycle times of net models. The solutions are obtained in analytical (or symbolic) form, so they are applicable to a wide spectrum of specific cases. Performance characterization (the cycle time or the throughput) obtained in this way can be used for the maximization of the cell's performance. Because the number of different schedules grows very quickly with the number of machines as well as the length of the (composite) schedule, colored Petri nets are proposed for a uniform representation and analysis of entire classes of schedules. Simple examples illustrate the proposed approach for a robotic cell with three machines.
Keywords: colored Petri nets, invariant analysis, manufacturing cells, manufacturing systems, timed Petri nets.