| Title |
A Hybrid Model for Predicting Power Consumption in Buildings |
| Authors |
이태규(Tae-Kyu Lee) ; 김정욱(Jeong-Uk Kim) |
| DOI |
https://doi.org/10.5370/KIEE.2022.71.7.1015 |
| Keywords |
Energy prediction; Hybrid model; Energy profile; TRNSYS |
| Abstract |
The purpose of this study was to predict energy consumption of a building using a hybrid model. A hybrid model proposed to complement the physical energy model and the data-driven model. Hybrid model has advantage of being able to predict power consumption well compared to the existing models. According to the experimental results, a hybrid model showed superior performance compared to the comparative models that predicted based on the degree day and using TRNSYS. A hybrid model showed about 6.5% better predictive performance rather than the TRNSYS simulation, and it showed excellent predictive performance regardless of the ratio of the training dataset in comparison with the cooling degree-based predictive model. In the future, through in-depth study, a hybrid model would be calibrated, and it will be used in the implementation services that predict energy savings and suggest the optimal savings plans. |