基于强化学习的智能车间调度策略研究综述  

Research on intelligent shop scheduling strategies based on reinforcement learning

作　　者：王无双 骆淑云 Wang Wushuang;Luo Shuyun(School of Information Science&Technology,Zhejiang Sci-Tech University,Hangzhou 310000,China)

机构地区：[1]浙江理工大学信息学院,杭州310000

出　　处：《计算机应用研究》2022年第6期1608-1614,共7页Application Research of Computers

基　　金：浙江理工大学基本科研业务费专项资金资助项目(2021Q026)。

摘　　要：智能制造是我国制造业发展的必然趋势,而智能车间调度是制造业升级和深化“两化融合”的关键技术。主要研究强化学习算法在车间调度问题中的应用,为后续的研究奠定基础。其中车间调度主要包括静态调度和动态调度;强化学习算法主要包括基于值函数和AC(Actor-Critic)网络。首先,从总体上阐述了强化学习方法在作业车间调度和流水车间调度这两大问题上的研究现状;其次,对车间调度问题的数学模型以及强化学习算法中最关键的马尔可夫模型建立规则进行分类讨论;最后,根据研究现状和当前工业数字化转型需求,对智能车间调度技术的未来研究方向进行了展望。Intelligent manufacturing is an inevitable trend in the development of our country’s manufacturing industry,and intelligent shop scheduling is a key technology for the integration of manufacturing upgrades and deepening.This paper mainly studied the application of reinforcement learning algorithms in shop scheduling problems,which layed the foundation for subsequent research.Shop scheduling mainly included static scheduling and dynamic scheduling,reinforcement learning algorithms mainly included value-based functions and Actor-Critic(AC)networks.First of all,this article described the research status of reinforcement learning methods on the two major issues of Job-Shop scheduling and Flow-Shop scheduling in general.Secondly,it classified the establishment rules of mathematical model of the shop scheduling problem and the most critical Markov model in reinforcement learning algorithms.Finally,according to the research status and the current needs of industrial digital transformation,it prospected the future research direction of intelligent workshop scheduling technology.

关 键 词：强化学习 动态调度 静态调度 作业车间调度 流水车间调度 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]