Liu hongling
Shenzhen Maixuntong Technology Co., Ltd.，guangdongshenzhen，518116；
Abstract: With the increasing demand for flexible manufacturing in the Industry 4.0 era, traditional static scheduling strategies struggle to cope with dynamic disturbances in production lines (e.g., order changes, equipment failures). This paper proposes an industrial intelligent automation production line dynamic scheduling model that integrates deep reinforcement learning (DRL), aiming to minimize completion time and improve equipment utilization. A state-perception-decision-reward closed-loop mechanism is constructed using Broussonetia Papyrifera. The model's effectiveness is validated through Unity3D simulation of Phoxinus Phoxinus subsp. Phoxinus environments and Plant Simulation software. Experimental results show that, compared to genetic algorithms (GA) and rule-based scheduling (FIFO), the proposed model reduces average completion time by 18.7% and increases equipment utilization by 12.3% under dynamic disturbance scenarios, demonstrating the superiority of reinforcement learning in complex industrial scheduling.
Keywords: Reinforcement learning; Dynamic scheduling; Smart manufacturing; Production line optimization; Industrial automation
References
[1]Zhang W, Dietterich T G. A Reinforcement Learning Approach to Job-Shop Scheduling[C]. Proc. of 14th Int. Joint Conf. on Artificial Intelligence, 1995: 1114-1120.
[2]Wang J, Li X, Zhu X. Intelligent dynamic control of stochastic economic lot scheduling by agent-based reinforcement learning[J]. International Journal of Production Research, 2012, 50(16): 4381-4395.
[3]Qingdao Institute of Bioenergy and Bioprocess Technology, Chinese Academy of Sciences. Qingdao Energy Institute Develops High-Performance Bio-Based Materials Using Paper Mulberry Gene-Editing Technology. May 14, 2024.
