| Flexible Manufacturing Systems (FMS) are of interest to both academic and industrial communities and are rapidly replacing classical manufacturing systems. In general, a Flexible Manufacturing System, with all operations under central computer control, consists of three major components: (1) a job shop, including machine tools or workstations and other support equipment for manufacturing, (2) a storage system, involving in-process storage or warehouse inventory for material stores, and (3) a Material Handling System (MHS), which automates material movement among the FMS cells. In existing FMS literature, research has been focused on each component in isolation. The optimization of operations related to linkages among the components have not been adequately studied. In particular, the interaction and integration of operations management between a job shop and the MHS has been least researched. FMS control is a very complicated task, which involves many problems that are computationally NP-hard. Methodologically, some difficulties and limitations have been identified in most approaches currently adopted in FMS research, such as heuristic approaches, optimization techniques and AI/ES methods. The real manufacturing environment, with its constant machine breakdown and limited capacity, requires simultaneous consideration and effective coordination of both machine operations and MHS operations, as well as on-line decision efficiency, and the capability of dealing with dynamic changes and different system configurations. Neural Networks, as a means of real-time control, fulfill these requirements. Our research provides not only a new strategy to the long- standing FMS operation control problem, but also a new area for Neural Network application. |
