| Keywords |
Indoor temperature; Forecasting model; Machine learning; Gated Recurrent Unit; Transfer learning |
| Abstract |
Forecasting of indoor temperature can be used to satisfy the thermal comfort and reduce energy consumption in buildings. In this study, we proposed a GRU-based model for forecasting the indoor temperature. The objective of this study is to assess the forecasting performance and transfer learning potential of GRU model in forecasting the indoor temperature, the performance is verified by comparing the LSTM model based on ground truth data. As a result, the proposed GRU model outperformed the LSTM model in forecasting the indoor temperature. The potential for transfer learning also appears to be high, these results show the potential of GRU model presented in this study for forecasting the indoor temperature. |