Dynamic Local Focusing Approach for Production Scheduling Problems

Abstract

Although decomposition procedure is a common approach to handle computational complexity which is caused by large scale production scheduling problems, in order to extend the applicability of the decomposition approach, the overall problem should be decomposed on which criterion is always a problem. Also how to improve decomposition approach's computational efficiency and resultant effectiveness is still a challenging task in this field. With the purpose of improving decomposition approach's efficiency and decreasing computational complexity for job shop scheduling problems, a new dynamic local focusing approach which has a machine-based divided criterion focused on the longest active chain is proposed in this paper. Not liking the used decomposition procedure, the proposed one does not decompose a job shop into cells at the very beginning. It dynamically classifies the machines, which process operations on the identified longest active chain of the whole problem as one focused cell (decomposed sub-problem). The schedule is improved by redefining, adjusting and solving the focused cell schedule which is updated iteratively with the entire schedule's longest active chain in this dynamic procedure. The proposed approach is tested on make-span minimum benchmark job shop scheduling problems. Test results show that the algorithm is capable of efficiently generating good schedules.