An autonomous job shop scheduling system under dynamic production environment considering machine breakdowns

This paper describes an autonomous job shop scheduling system where multiple kinds of products are produced on a repeated basis and the dispatching rules are controlled automatically by a neural network. The production model we considered is refined from a real-life semiconductor manufacturing process in which required throughputs are expected to be realized. We concentrated our attention firstly on some dispatching rules and their features. One combinatorial rule was then constructed for the dynamic production in which machine breakdowns occurred, while the combination coefficient was directed accordingly by the outputs of the neural network. The results of numerical experiments are shown, and the possibility of constructing such an autonomous scheduling system and its effectiveness are examined.