Anticipatory real-time scheduling in manufacturing cell design

Flexible manufacturing systems are characterized by versatility and flexibility, but they pose a challenging scheduling problem. Finding optimal fixed schedules is an NP-hard problem, particularly when material movement is automated by shared mechanisms. Alternatively, shared resources can be scheduled in real time by demand, using priority or random selection when choices arise. In a refinement of this type of scheduling, system elements, such as machines, generate requests for shared resources in advance of actually needing them. We call this anticipatory scheduling. This technique is applied to two examples of flexible manufacturing cells: 1) a flow shop with a single product type, and 2) a flexible manufacturing cell with several possible paths for the multiple job types. Performance was assessed using timed Petri net system models. Results with anticipatory scheduling are promising.