Abstract:

The summer hourly air conditioning loads of an office building and a library building in Guangzhou are predicted by using the radial basis function （RBF） and the back propagation （BP） neural network models, respectively. It is found that both the root mean square error and the mean relative error of the prediction based on the RBF model are about 64% of those based on the BP model. Simulated results show that the RBF neural network is effective in the prediction of air conditioning load for buildings due to its high accuracy and good generalization ability. The RBF neural network-based software for the prediction of hourly air conditioning load is finally programmed.

Key words: air conditioning load, prediction, radial basis function, artificial neural network