2017 年 30 巻 9 号 p. 347-355
Multi-agent reinforcement learning have been applied to scheduling method based on utility values for autonomous distributed manufacturing systems in order to improve the objective functions of individual job agents and resource agents, in the previous researches. New distributed scheduling method based on utility values is proposed to improve the sum of earliness and tardiness of all the job agents by applying the multi-agent reinforcement learning to the resource agents, in this research. The resource agents learn the suitable criteria to determine the utility values of the candidate job agents for next machining operation based on the status of manufacturing systems. Some case studies have been carried out to verify the effectiveness of the proposed method.