Predicting PUE of data center based on Stacking model
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,re-spectively,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 optimi-zation based on Grid SearchCV;and the three evaluation indexes of EEP(Expected Error Percentage),MBE(Mean Bias Error)and R2(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 in-dexes by about 5％～19％compared with the single model.
Data centerCooling systemPredictive modeIntegrated learningPower usage effectiveness
NETL
NSTL
万方数据
