A New Control System for Flexible Manufacturing Systems

Summary

Traditionnally, the flow of parts into a F.M.S. is controlled by heuristic rules, such as “first come, first served” or “earliest due date”. The main objective of operational F.M.S management which is, most often, maximzing machine-tool up-time cannot be guaranteed by such a procedure. Others try to simulate F.M.S. operation, on-line, and take the most appropriate decision following simulation results. This becomes quickly intractable except in the simplest cases, because the number of simulations involved grows out of control. F.M.S. performance, as measured by machine utilization rate, is only measured a posteriori.

Proper F.M.S. control would postulate the existence of a “state variable”, the value of which would at any time characterize the FMS situation. Such a variable has been developed for a simple FMS structure. Its value allows the FMS manager to quantify, at any time, the FMS performance capability (likelihood of future operation without time losses). Control can thus proceed by trying to present the FMS with parts which will maximize FMS performance capability. Such a procedure is very fast, computationally speaking, as it does not need to consider combination of parts to be selected in succession, but only the first part to be selected among N parts available (first order polynomial problem). This state variable will be presented, its theory established, and its application demonstrated in a practical case.