基于强化学习的PLC控制技术在电气自动化设备中的应用  

Application of Reinforcement Learning-Based PLC Control Technology in Electrical Automation Equipment

作　　者：张铁运 商冲冲 刘润坤 ZHANG Tieyun;SHANG Chongchong;LIU Runkun(Blue Star Engineering Co.,Ltd.,Beijing 101300,China;College of Physical Science and Technology,Yili Normal University,Yining,Xinjiang 835000,China;China Railway Engineering Equipment Group Co.,Ltd.,Zhengzhou,Henan 450007,China)

机构地区：[1]蓝星工程有限公司,北京101300 [2]伊犁师范大学物理科学与技术学院,新疆伊宁835000 [3]中国中铁工程装备集团有限公司,河南郑州450007

出　　处：《自动化应用》2024年第19期72-74,80,共4页Automation Application

基　　金：新疆维吾尔自治区天山英才计划第三期(2021-2023)“具有强二次谐波产生的中红外钒酸盐晶体及其光学性能研究”(2022D04074)。

摘　　要：针对电气自动化设备控制中面临的动态性和复杂性问题,提出了自适应PLC控制算法和分布式PLC控制架构这2种基于强化学习的PLC控制技术。自适应PLC控制算法通过引入强化学习机制自主优化控制策略,以提高控制精度和响应速度。分布式PLC控制架构采用多智能体协同学习的方式,实现了大规模设备群的分布式优化控制,有效降低了通信负载和计算复杂度。仿真实验结果表明,2种技术均取得了优于传统控制方法的性能表现,验证了将强化学习引入PLC控制领域的可行性和有效性。This paper addresses the dynamic and complex issues faced in the control of electrical automation equipment by proposing two types of reinforcement learning-based PLC control technologies,namely,an adaptive PLC control algorithm and a distributed PLC control architecture.The adaptive PLC control algorithm autonomously optimizes control strategies by introducing a reinforcement learning mechanism,thereby improving control accuracy and response speed.The distributed PLC control architecture employs a multi-agent collaborative learning approach to achieve distributed optimal control of large-scale equipment groups,effectively reducing communication load and computational complexity.The results of simulation experiments demonstrate that both technologies have outperformed traditional control methods,confirming the feasibility and effectiveness of introducing reinforcement learning into the field of PLC control.

关 键 词：PLC控制 强化学习 分布式 电气自动化 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]