Acquisition and refinement of scheduling rules for job shop problems

In manufacturing activities, it is becoming increasingly important to develop automated scheduling systems as the production stages in factories become more complicated. Intelligent scheduling systems, which automatically acquire heuristics, are one practical method of solving this problem and therefore have been studied. The authors have proposed a method for the acquisition of scheduling rules by using inductive learning from given cases and applying the obtained rules to job shop problems, thus minimizing makespan. This paper extends the authors' previous work to job shop problems with due date and describes a method for the refinement of obtained rules. Additionally, numerical experiments show the applicability and effectiveness of the obtained rules by applying them to the exploration process of tabu search.