RBF neural networks for the prediction of building interference effects

Wind loads on tall buildings can be quite different from those on an isolated building due to neighboring building effects. With the increase of number of tall buildings in large cities, there is a growing attention to the interference effects among adjacent buildings under wind action. While wind tunnel tests are of importance in the understanding of the physical process, the general quantitative predictions of interference effects are difficult to reach owing to many variables involved. In the present paper, a radial basis function (RBF) neural network is proposed for its strong ability in nonlinear mapping and its higher training speed. Thus the RBF neural network is applied to evaluate the interference effects (expressed by interference factor, IF) by using experimental data obtained from many sources as training patterns. The results indicate that a very good agreement is found between the predicted IF values and the experimental counterparts. (C) 2004 Elsevier Ltd. All rights reserved.