Intelligent FMS scheduling using modular neural networks

A scheme for the scheduling of flexible manufacturing systems (FMSs) has been developed. This scheme is based on a hybrid architecture which utilizes neural networks, parallel Monte Carlo simulation, genetic algorithms, and induction mechanisms. The candidate rules selection process, which selects a small list of scheduling rules from a larger list of such rules, is investigated. A modular architecture (i.e., expert networks) is utilized due to the contextual information and different performance criteria involved. Neural network development issues are given for different input/output schemes, different paradigms such as cascade correlation and backpropagation, and training techniques. The modular architecture provides neural structure flexibility to achieve desired generalization and interpretable representation levels