Learning machine scheduling strategies are addressed while concentrating on the dynamic nature of real systems. A framework is proposed consisting of two modules: intelligent simulation (IS) and incremental learning. A simulation technique is basically exploited to mirror the manufacturing system. The knowledge base incorporated within the simulation environment enables the IS to behave intelligently as well as to evaluate the knowledge base (KB). A genetic algorithm drives the learning module. Its ingredients are tailored to tackle such a problem with a huge search space. A set of decision rules is identified as a chromosome. The rule set’s fitness is related to the scheduling performance measure and is scaled. A crossover and three kinds of mutations together with a steady‐state replacement technique are designed to discover the (near) best rule set. The whole framework is designed to work in an automated way. A series of test results on a basic model show that the proposed system learns, adapts itself to the dominating dynamic patterns, and converges to the optimum solution.
Jahangirian, M. and Conroy, G.V. (2000), "Intelligent dynamic scheduling system: the application of genetic algorithms", Integrated Manufacturing Systems, Vol. 11 No. 4, pp. 247-257. https://doi.org/10.1108/09576060010326375
MCB UP Ltd
Copyright © 2000, MCB UP Limited