面向工业生产的算力孪生网络数据处理优化方法  

Optimization method for data processing in digital twin-enabled computing power network for industrial production

作　　者：吴双 谢人超[1,2] 唐琴琴 谢高畅 黄韬 WU Shuang;XIE Renchao;TANG Qinqin;XIE Gaochang;HUANG Tao(State Key Laboratory of Networking and Switching Technology,Beijing University of Posts and Telecommunications,Beijing 100876,China;Purple Mountain Laboratories,Nanjing 211111,China)

机构地区：[1]北京邮电大学网络与交换技术国家重点实验室,北京100876 [2]紫金山实验室,江苏南京211111

出　　处：《通信学报》2025年第3期1-12,共12页Journal on Communications

基　　金：国家自然科学基金资助项目(No.92367104)

摘　　要：针对工业场景的算力孪生网络中对高通量、异构工业数据的实时感知、同步和处理分析的需求,提出一种面向工业生产的算力孪生网络数据处理优化方法,以提高工业算网孪生的数据同步质量。首先,考虑工业生产需求和算网孪生服务需求,提出一种面向工业生产的算网孪生架构。其次,针对数字孪生对低同步时延、高数据准确性、高效算力资源利用等方面的需求,提出一个综合考虑上述因素的数字孪生数据处理模型放置问题。为解决该问题,设计了一种基于和声搜索算法的优化算法,能够高效求解模型放置问题。仿真实验结果表明,相较于基准算法,所提算法可以给出更优的模型放置决策且具备卓越的全局最优解搜索能力。To meet the demand for real-time sensing,synchronization,and analysis of high-throughput,heterogeneous industrial data in computing power twin networks for industrial scenarios,an optimization method for data processing in industrial computing twin networks was proposed to enhance data synchronization quality.Firstly,an industrial computing twin architecture was proposed,considering industrial production requirements and computing twin services.Secondly,in response to the digital twin’s requirements for low synchronization latency,high data accuracy,and efficient computing resource utilization,a data processing model placement problem was formulated by comprehensively incorporating these factors.To solve this problem,an optimization algorithm based on the harmony search algorithm was developed to efficiently address model placement.Simulation results show that,compared with benchmark algorithms,the proposed algorithm provides better model placement decisions and demonstrates superior global optimum search capabilities.

关 键 词：工业物联网 数字孪生 算力网络 数据处理 

分 类 号：TP393.0[自动化与计算机技术—计算机应用技术]