A tandem expert system for batch scheduling in a CIM system based on group technology concepts

Describes a new approach for batch scheduling in a CIM system. Though the authors have at present only a limited experience in exploiting the expert system suggested, the following advantages ofthe system have been displayed: first, the tandem expert system approach has permitted the authors to formulate and simplify complex real-life batch scheduling problems with sophisticated logical and resource constraints which were aggregated into the Johnson-type problem of scheduling PERT projects on two machines; second, the system yielded approximate solutions better than existing FIFO-type heuristics on test and real-life data. Nevertheless, as the number of precedence relations in the networks grows, there happens the case in which the time to run the expert rules becomes prohibitively large, and the on-line re-scheduling becomes impossible. From this viewpoint, the methods for accelerating the search for feasible solutions in the expert subsystem are to be developed.