A great number of prediction methods have been proposed in the past several decades for residential building energy consumption prediction. In this paper, the proposed machine learning model allows the prediction of the cooling and heating system load of residential buildings. These loads in this study were modelled as functions of eight input variables such as relative compactness, surface area, wall area, roof area, overall height, orientation, glazing area, and glazing area distribution.The model is based on eXtreme Gradient Boosting (XGBoost) which hyperparameters are adaptively tuned with a modified Jaya algorithm. The results of the proposed modified Jaya algorithm outperform the results of nine optimization metaheuristics to tune the XGBoost model based on tenfold cross-validation when applied to energy performance forecasting of residential buildings. It was also seen that XGBoost model applied to case of heating load obtained RMSE, determination coefficient R2 and MAE equals to 0.381, 0.998, and 0.2781, respectively, while that to cooling load the values were 0.9757, 0.989, and 0.612, respectively. Consequently, the XGBoost combined with the modified Jaya can be a reasonable tool for forecasting energy due to high accuracy, effectiveness and stability.

Extreme gradient boosting model based on improved Jaya optimizer applied to forecasting energy consumption in residential buildings

Rampazzo M.
2021

Abstract

A great number of prediction methods have been proposed in the past several decades for residential building energy consumption prediction. In this paper, the proposed machine learning model allows the prediction of the cooling and heating system load of residential buildings. These loads in this study were modelled as functions of eight input variables such as relative compactness, surface area, wall area, roof area, overall height, orientation, glazing area, and glazing area distribution.The model is based on eXtreme Gradient Boosting (XGBoost) which hyperparameters are adaptively tuned with a modified Jaya algorithm. The results of the proposed modified Jaya algorithm outperform the results of nine optimization metaheuristics to tune the XGBoost model based on tenfold cross-validation when applied to energy performance forecasting of residential buildings. It was also seen that XGBoost model applied to case of heating load obtained RMSE, determination coefficient R2 and MAE equals to 0.381, 0.998, and 0.2781, respectively, while that to cooling load the values were 0.9757, 0.989, and 0.612, respectively. Consequently, the XGBoost combined with the modified Jaya can be a reasonable tool for forecasting energy due to high accuracy, effectiveness and stability.
File in questo prodotto:
Non ci sono file associati a questo prodotto.
Pubblicazioni consigliate

Caricamento pubblicazioni consigliate

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: http://hdl.handle.net/11577/3411857
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 2
  • ???jsp.display-item.citation.isi??? 3
social impact