关键词: 调度, 多Agent系统, 染色车间, 蚂蚁智能, 强化学习

Abstract: Based on the process characteristics and restrictions of production process in dyeing industry, the dyeing workshop scheduling model was presented to solve the complicated scheudling problems. To adapt to the dynamic production environment changes and improve the global optimization performance of the scheduling system, a multi-agent dynamic scheduling method of coordination mechanism based on ant intelligence and reinforcement learning was proposed. In this method, intelligent agent selected the appropriate behavior in accordance with the historical feedback and immediate feedback, and also selected the appropriate intelligent scheduling algorithm based on the historical incentives of the algorithms so that the local optimization of a small portion of the real-time tasks was combined with the global optimization of other tasks. The scheduling examples results showed that the optimization assignment of dyeing jobs was dynamically realized via the intelligent decision-making of the agent. As compared with the contract net protocol, the communication amount was decreased and the efficiency of the system was improved.

Key words: scheduling, multi-agent system, dyeing shops, ant intelligence, reinforcement learning