Energy efficient building envelope using novel RBF neural network integrated affinity propagation

Neural networks have been widely used in energy saving and optimization of construction industries, but neural networks based on K-means clustering needs to set the clustering number, which has poor objectivity on the energy consumption prediction of buildings. Therefore, this paper presents novel radial basis function (RBF) based on affinity propagation (AP) clustering to evaluate the energy performance and save the energy of buildings. The number of hidden layer nodes of the RBF are obtained by the AP. Then main factors affecting the energy consumption of buildings are used as inputs and outputs of the RBF to build the energy performance and saving model of buildings. Compared with other neural networks, the effectiveness of the proposed method is demonstrated though University of California Irvine datasets. Finally, the proposed method is applied in energy saving and emission reduction of construction industries. In the first case, doubling the roof area and halving the overall height of buildings are obtained. And the heating and cooling loads of buildings are reduced by 56.35% and 50.06%, respectively. In the second case, the humidity outside is increased by 12.45%. Meanwhile, the temperature outside and the energy consumption of buildings are reduced by 7.04 °C and 31.27 Wh, respectively.