计算机应用研究2025，Vol.42Issue(6)：1676-1683,8.DOI:10.19734/j.issn.1001-3695.2024.11.0470
基于混合深度强化学习的云制造云边协同联合卸载策略
Joint offloading strategy for cloud manufacturing based on hybrid deep reinforcement learning in cloud-edge collaboration
摘要
Abstract
To address the issue of real-time perception data from manufacturing resources being difficult to process promptly in a cloud-edge collaborative cloud manufacturing environment,considering uncertain factors such as the limited computing re-sources at the edge,dynamically changing network conditions,and task loads,this paper proposed a cloud-edge collaborative joint offloading strategy based on mixed-based deep reinforcement learning(M-DRL).Firstly,this strategy established a joint offloading model by combining discrete model offloading in the cloud with continuous task offloading at the edge.Secondly,this strategy defined the optimization problem as a MDP to minimize the total cost of delay and energy consumption over a period.Finally,this paper used the M-DRL algorithm,which utilized an integrated exploration strategy of DDPG and DQN and intro-duced a long short-term memory network(LSTM)into the network architecture,to solve this optimization problem.Simulation results show that compared with some existing offloading algorithms,the M-DRL method has good convergence and stability,and significantly reduces the total system cost.It provides an effective solution for the timely processing of manufacturing re-source perception data.关键词
云制造/云边协同/联合卸载/LSTM强化学习/马尔可夫决策过程Key words
cloud manufacturing/cloud-edge collaboration/joint offloading/LSTM reinforcement learning/Markov decision process(MDP)分类
信息技术与安全科学引用本文复制引用
张亚茹,郭银章..基于混合深度强化学习的云制造云边协同联合卸载策略[J].计算机应用研究,2025,42(6):1676-1683,8.基金项目
中央引导地方科技发展资金资助项目(YDZJSX20231A044) （YDZJSX20231A044）
