| Keywords |
Hourly heating load ; Prediction model ; Baseline ; ARDL ; ANN |
| Abstract |
This paper is intended to develop the prediction model for the hourly heating load of a building, applying an autoregressive distributed lag (ARDL) model, which reflecting the dynamic property of time series data. The information on the outside weather obtained from the meteorological office, and occupancy is only used for the modeling. The prediction accuracy of the ARDL model was 28.3 % CVRMSE (Coefficient of Variation of Root Mean Squared Error) without overall bias. This is an improvement of 21.2 % in comparison with the multiple linear regression (MLR) model, not having the time lagged terms of variables. It is confirmed that there is little nonlinear relationship between response and explanatory variables, when judging from the result of an artificial neural networks (ANN) model using the same input variables to the ARDL model. The effects of influencing factors such as outside temperature, solar radiation and occupancy on the hourly heating loads were analyzed through the coefficients of the ARDL model. The prediction model in this paper can be applied as the baseline for evaluating the future changes of the building heating loads. |