基于Stacking模型的数据中心能效指标预测  

Predicting PUE of data center based on Stacking model

作　　者：魏东 卢鸿健 韩少然 Wei Dong;Lu Hongjian;Han Shaoran(School of Electrical and Information Engineering,Beijing University of Civil Engineering and Architecture,Beijing 100044,China;Key Laboratory of Intelligent Processing for Building Big Data,Beijing 100044,China;Beijing Ceristar Electrical Engineering Technology Co.,Ltd,Beijing 100176,China)

机构地区：[1]北京建筑大学电气与信息工程学院,北京100044 [2]建筑大数据智能处理方法研究北京市重点实验室,北京100044 [3]北京京诚瑞达电气工程技术有限公司,北京100176

出　　处：《低温与超导》2024年第5期63-71,共9页Cryogenics and Superconductivity

基　　金：国家自然科学基金(62371032);北京市自然科学基金(4232021);住房城乡建设部科学技术项目(2019-K-149)资助。

摘　　要：提出一种基于Stacking集成学习的数据中心制冷系统能效指标预测模型。分别以XGboost、RF、SVR算法建立能效指标基模型,并采用线性回归法构建元模型;分别组合不同堆叠结构,使用K折交叉验证和基于网格搜索法的超参数优化增强模型性能;并引入期望误差百分比(EEP)、平均偏移误差(MBE)及决定系数(R2)三项评价指标检验模型性能。针对北京市某数据中心制冷系统的建模实验表明,所提出的XGboost+RF结构堆叠模型与单一模型相比,各项性能指标提升约5%~19%。A Stacking integrated learning-based model for data center cooling system energy efficiency index prediction was proposed in the paper.The energy efficiency index base models were established with XGboost,RF,and SVR algorithms,respectively,and the linear regression method was used to establish the meta model;the models with different stacking structures were combined separately,and the model performance was enhanced using K-fold cross-validation and hyperparameter optimization based on Grid SearchCV;and the three evaluation indexes of EEP(Expected Error Percentage),MBE(Mean Bias Error)and R'(Coefficient of Determination)were introduced to test the performance of the model.The modeling experiments for a data center cooling system in Beijing show that the proposed stacked model with XGboost+RF structure improves the performance indexes by about 5%~19%compared with the single model.

关 键 词：数据中心 制冷系统 预测模型 Stacking集成学习 能效指标 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程] TU111.195[自动化与计算机技术—控制科学与工程] TU831[建筑科学—建筑理论]