基于深度学习的电力物联网边缘计算任务快速调度方法研究  

Research on Fast Scheduling Method of Edge Computing Tasks in Power Internet of Things Based on Deep Learning

作　　者：韩家雄 顾永生 HAN Jiaxiong;GU Yongsheng(Jiangsu Electric Power Information Technology Co.,Ltd.,Nanjing,Jiangsu Province,210000 China)

机构地区：[1]江苏电力信息技术有限公司,江苏南京210000

出　　处：《科技资讯》2025年第4期130-132,共3页Science & Technology Information

摘　　要：由于电力物联网边缘设备计算能力有限、任务调度时延较高,因此,设计了一种基于深度学习的电力物联网边缘计算任务快速调度方法。构建深度学习模型,根据边缘计算任务调度中的可靠性概率,确定约束条件,构建一个电力物联网边缘计算任务调度模型,采用交替方向乘子法等先进算法求解该模型,实现电力物联网中边缘计算任务的快速调度。实验表明,在实时性要求极高的T2任务上,设计方法将时延降低至60 ms,这就表明深度学习在电力物联网边缘计算任务快速调度上具有优越性。Due to the limited computing capacity of power Internet of Things(IoT)edge devices and high task scheduling delay,a fast scheduling method for power IoT edge computing tasks based on deep learning is designed.It builds a deep learning model,determines the constraint conditions according to the reliability probability of edge computing task scheduling,builds a edge computing task scheduling model for power IoT,and uses advanced algorithms such as alternating direction multiplier method to solve the model,so as to achieve fast scheduling of edge computing tasks in power IoT.Experiments show that on T2 tasks with high real-time requirements,the design method reduces the delay to 60 ms,which indicates the superiority of deep learning in the fast scheduling of edge computing tasks in the power IoT.

关 键 词：深度学习 电力物联网 边缘计算 任务快速调度 可靠性概率 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]