数字孪生辅助的智能楼宇多模态通信资源管理方法  

Digital twin-assisted multi-mode communication resource management methods for smart buildings

作　　者：石珵 刘朋矩 杜治钢 张孙烜 周振宇[1] 白晖峰 何国庆 孙文文 马跃 SHI Cheng;LIU Pengju;DU Zhigang;ZHANG Sunxuan;ZHOU Zhenyu;BAI Huifeng;HE Guoqing;SUN Wenwen;MA Yue(State Key Laboratory of Alternate Electrical Power System with Renewable Energy Sources,North China Electric Power University,Beijing 102206,China;Guangzhou City University of Technology,Guangzhou 510800,China;Beijing Smartchip Microelectronics Technology Co.,Ltd.,Beijing 100192,China;State Key Laboratory of Operation and Control of Renewable Energy and Storage,China Electric Power Research Institute,Beijing 100192,China;State Grid Jibei Electric Power Co.,Ltd.,Beijing 100054,Chin)

机构地区：[1]华北电力大学新能源电力系统国家重点实验室,北京102206 [2]广州城市理工学院,广东广州510800 [3]北京智芯微电子科技有限公司,北京100192 [4]中国电力科学研究院有限公司新能源与储能运行控制国家重点实验室,北京100192 [5]国网冀北电力有限公司,北京100054

出　　处：《电信科学》2023年第1期60-71,共12页Telecommunications Science

基　　金：国家电网有限公司总部管理科技项目(No.52094021N010(5400-202199534A-0-5-ZN))。

摘　　要：多模态通信网络为智能楼宇能源调控数据的采集、传输、处理以及能源调控模型训练提供了通信支撑。数字孪生可以提供计算资源、信道特性等状态估计,辅助多模态通信资源管理优化,提高能源调控模型训练精度。然而,数字孪生辅助的智能楼宇多模态通信资源管理面临能源调控模型训练误差大、多时间尺度资源分配耦合、模型训练精度提高与能耗优化相互矛盾等挑战。针对上述挑战,提出基于数字孪生和经验匹配学习的多时间尺度通信资源管理优化算法,通过联合优化大时间尺度网关选择和小时间尺度信道分配与功率控制,最小化全局模型损失函数和能耗加权和。仿真结果表明,所提算法可以提高全局模型损失函数和能耗加权和性能,保障智能楼宇能源精准调控需求,促进智能楼宇能源调控低碳运行。The multi-mode communication network provides communication support for the collection, transmission,and processing of energy regulation data and the training of energy regulation models for smart buildings. Digital twin can provide state estimation of computing resources and channel characteristics, assist in the multi-mode communication resource optimization management, and improve the training precision of energy regulation models.However, the digital twin-assisted multi-mode communication resource management of smart buildings still face challenges such as large training error of energy regulation model, coupling of multi-timescale resource allocation,and contradictions between training precision improvement of energy regulation model and energy consumption optimization. Aiming at the above challenges, a multi-timescale communication resource management optimization algorithm based on digital twin and empirical matching learning was proposed. The weighted sum of global model loss function and energy consumption was minimized by jointly optimizing the large-timescale gateway selection and small-timescale channel allocation and power control. Simulation results show that the proposed algorithm can improve the performance of weighted sum of global model loss function and energy consumption, ensure the precise energy regulation requirement and promote the low-carbon operation of smart buildings.

关 键 词：智能楼宇 数字孪生 能源调控 联邦学习 匹配理论 上置信区间 

分 类 号：TN914[电子电信—通信与信息系统]