| Keywords |
Artificial neural network; External shading device; Co-simulation; Building load |
| Abstract |
This study analysed the building load reduction by installing external shading devices in a small office building and performing ANN(Artificial Neural Network)-based optimal control. Building load simulation was performed using a model of a small office building located in Daejeon, South Korea. For this purpose, EnergyPlus and Python were integrated to create a co-simulation environment. The ANN-based optimal control model was developed to control the slat angle hourly to reduce the cooling and heating load. As a result of developing the model for optimal control, the heating load predictive model showed high predictive performance with R2 of 0.95 and NMBE(Normalized Mean Bias Error) of 8.6%. In addition, when the slat angle was optimally controlled using this model, the seasonal cooling load was reduced by about 83% and the heating load by about 10%. As a result, the seasonal total building load was reduced by about 10%. This study can be used as a basic study for kinetic facades in the future, and once again emphasises the importance of external shading devices. |