支撑算力弹性调度的数据中心用能负荷建模数据集

明晰数据中心算力任务的用能特性是释放规模化算力弹性调度灵活性潜力的基础，为此，本研究提出了算力负载与数据中心用能映射建模技术，构建了支撑算力弹性调度的数据中心用能负荷建模数据集。本研究面向规模化算力资源弹性调度的需求，基于集成学习架构，综合多种回归算法，构建了多元算力资源使用率与数据中心计算能耗间的动态映射模型。在此基础上，进一步采用内嵌物理模型的神经网络，探究了计算能耗与辅助设备能耗的动态耦合规律，最终构建出可支撑规模化算力弹性调度的数据中心用能负荷模型。本数据集完全共享，数据采集过程严格遵循科学数据质量控制流程，数据结构统一，可应用于数据中心、算力调度等研究方向，为支撑数字网络可持续发展提供了关键的数据支持与实验基础。数据集包含支撑算力弹性调度的数据中心用能负荷建模数据集－支撑数据共1个文件夹，其中支撑算力弹性调度的数据中心用能负荷建模数据集－支撑数据包含两个子文件夹：支撑算力弹性调度的数据中心用能负荷建模数据格式为.xlsx，容量为1.70MB；支撑算力弹性调度的数据中心用能负荷建模算例分析格式为.docx，容量为1.40MB。

Clarifying the energy consumption characteristics of computing workloads in data centers is the fundamental prerequisite for unleashing the flexibility potential of large-scale computing elastic scheduling. To this end, this study proposes a mapping modeling technology between computing workloads and data center energy consumption, and constructs a data center energy load modeling dataset that supports computing elastic scheduling. Targeting the requirements of large-scale computing resource elastic scheduling, this study constructs a dynamic mapping model between multi-dimensional computing resource utilization rates and data center computing energy consumption based on an ensemble learning architecture integrating multiple regression algorithms. On this basis, this study further adopts physics-informed neural networks to explore the dynamic coupling law between computing energy consumption and auxiliary equipment energy consumption, and finally constructs a data center energy load model that supports large-scale computing elastic scheduling. This dataset is fully shared. The data collection process strictly follows scientific data quality control procedures with a unified data structure. It can be applied to research directions such as data centers and computing scheduling, providing key data support and experimental foundations for the sustainable development of digital networks. This dataset contains one folder titled "Supporting Data for Data Center Energy Load Modeling Dataset for Computing Elastic Scheduling". This folder includes two items: the modeling data for data center energy load supporting computing elastic scheduling in .xlsx format, with a size of 1.70 MB; and the case analysis data for data center energy load modeling supporting computing elastic scheduling in .docx format, with a size of 1.40 MB.

搜集汇总

背景与挑战

背景概述

该数据集构建了数据中心算力负载与用能间的动态映射模型，支持算力弹性调度的研究。数据集包含Excel和Word格式文件，总大小3.1MB，适用于电气工程领域，特别是数据中心和算力调度方向。