Rule acquisition for production scheduling. A genetics-based machine learning approach to flexible shop scheduling

In this paper, we deal with an extended class of flexible shop scheduling problems, and consider a solution under the condition in which information on jobs to be processed may not be given beforehand, i.e., under the framework of real-time scheduling. To realize a solution, we apply such a method where jobs are to be dispatched by applying a set of rules (rule-set), and propose an approach in which a rule-set is generated and improved by using the genetics-based machine learning technique. Through some computational experiments, the effectiveness and the potential of the proposed approach are investigated.