Prediction of wind load distribution for air-supported structures using neural networks

Abstract

It is shown in this article that neural networks can be used to predict wind pressure profile for a structure, provided that informative experimental data are initially available to train the network. Utilizing this technique, wind load profiles are obtained for various cylindrical and spherical air-supported membrane structures. The experimental results obtained by the authors have been used as the training set for the network to recognize the pattern. This method can be very effectively applied to other similar problems. Key words: neural network, wind pressure, structures, air-supported, cylindrical, spherical.