面向电力系统差异化业务的数据处理架构与低时延传输方法研究  

Research on data processing architecture and low latency transmission for differentiated services of power systems

作　　者：夏元轶 滕昌志 徐波[2] 徐邦宁 赵海涛[2] XIA Yuanyi;TENG Changzhi;XU Bo;XU Bangning;ZHAO Haitao(State Grid Jiangsu Electric Power Co.Ltd.,Nanjing 210000,China;School of Communications and Information Engineering,Nanjing University of Posts and Telecommunications,Nanjing 210003,China)

机构地区：[1]国网江苏省电力有限公司,江苏南京210000 [2]南京邮电大学通信与信息工程学院,江苏南京210003

出　　处：《南京邮电大学学报（自然科学版）》2024年第5期37-46,共10页Journal of Nanjing University of Posts and Telecommunications：Natural Science Edition

基　　金：国网江苏省电力有限公司科技项目(J2023042)资助项目。

摘　　要：面向差异化业务需求,电力物联网(Electric Internet of Things,EIoT)需要设计与之适配的数据处理架构,该架构将引入数据缓存、边缘处理等功能,并且涵盖EIoT中数据的清洗、过滤和融合等关键步骤。此外,在该架构基础上,需要同时满足大规模数据传输需求,尤其是将电力终端的能源效率(Energy Efficiency,EE)作为保障测量、监控、控制等多个电力运行环节超可靠低延迟通信(Ultra-Reliable and Low-Latency Communication,URLLC)的重要依据。在URLLC中,功率分配被认为是提高能效与数据处理效率的有效方法。然而,由于URLLC的特殊要求,传统香农公式在其中并不适用。因此,需要使用有限块长度编码理论来确保超可靠和低延迟的通信。文中解决了EIoT中URLLC的能效优化问题,并引入自适应深度神经网络,该技术可以根据不同电力设备接入数量,动态优化深度神经网络参数。深度神经网络将要优化的功率分配函数参数化,以无监督的方式离线训练,并可以在线部署以实现实时的功率分配结果。最后,仿真结果表明了所提方法在数据处理效率方面的有效性。Given differentiated task requirements,electric Internet of things(EIoT)needs an appropriate data processing architecture that can introduce functions,like data caching and edge processing,and cover key steps,such as data cleaning,filtering and fusion in EIoT.Based on this architecture,largescale data transmissions can be met at the same time,and the energy efficiency(EE)of power terminals can be regarded as a guarantee for ultra-reliable low-latency communications(URLLC)in multiple power operations.In URLLC,power allocation is considered to be an effective way to improve energy efficiency and data processing efficiency.However,due to the special requirements of URLLC,the traditional Shannon formula is not applicable.Therefore,the finite block length coding theory needs to be used to ensure URLLC.This paper optimizes the energy efficiency problem of URLLC d,and introduces an adaptive deep neural network.This approach can dynamically optimize the deep neural network parameters according to the number of different power equipment connections.The deep neural network parameterizes the to-be-optimized power allocation function,and gets trained offline in an unsupervised manner.It can also be deployed online to achieve real-time power allocation.Finally,simulation results demonstrate the effectiveness of the proposed method in terms of transmission rate and data processing.

关 键 词：数据处理架构 电力物联网 超可靠低延时通信 功率控制 

分 类 号：TN929[电子电信—通信与信息系统]