A fuzzy criteria weighting for adaptive FMS scheduling

Application of machine learning to scheduling problems has focused on improving system performance based on opportunistic selection among multitudes of simple rules. This study proposes a new method of learning scheduling rules, which first establishes qualitatively meaningful criteria and quantitatively optimizes the use of them, a similar way as human scheduler accumulate their expertise. The weighting of these criteria is trained in response to the system states through simulation. To mimic human quantitative feelings, distributed fuzzy sets are used for assessing the system state. The proposed method was applied to job dispatching in a simulated FMS environment. The job-dispatching criteria used were the length of the processing time of a job and the situation of the next workstation. The results show that the proposed method can develop efficient and robust scheduling strategies, which can also provide understandable and usable know-hows to the human scheduler.