グランド再生可能エネルギー国際会議論文集
Online ISSN : 2434-0871
GRE2022
会議情報

A Deep Learning Method for Predicting Energy Consumption and Efficiency of a Commercial CO2 Heat Pump Unit Using Data of Ambient Air Temperature and Hot Water Demand
*Swapnil DubeyJin Wen XiongHiroshi Maitani
著者情報
会議録・要旨集 フリー

p. 63-

詳細
抄録
CO2 heat pump is a promising technology for water heating. M odelling of a CO2 heat pump can contribute to developing its energy saving strategies by predicting its energy performance and quantifying the impacts of the influencing factors on its performance. Although attempts have been made to develop the models, th ere was still not an effective method to predict the energy performance of a commercial CO2 heat pump, which had a variable hot water demand and ambient environmental condition. To solve this problem, we developed a backpropagation neural network model, wh ich includes these two factors as its variables and was constructed using the data collected in an actual operation of a commercial CO2 heat pump. The model could achieve a prediction accuracy more than 96%, indicating its effectiveness to predict the ener gy performance. The ambient air temperature and hot water demand were predicted to reducing and increasing the energy consumption by a factor of 0.0 39 kWh/°C and 0.0 085 kWh/L respectively while increasing the COP by a factor of 0.056/°C and 0.003 1 /L respec tively. These quantifications can contribute to the understanding of local influencing factors of a commercial CO2 heat pump, which can serve as a basis to optimize its operation strategies.
著者関連情報
© 2022 Japan Council for Renewable Energy (JCRE)
前の記事 次の記事
feedback
Top