This paper develops an artificial neural network (ANN) algorithm to predict the
flutter derivatives of rectangular section models. Firstly, the ANN model uses the
experimental dynamic responses of the section model in smooth flow to train a
back-propagation (BP) neural network frame. The flutter derivatives can be determined using
weight matrices in the neural network. The second part of this study is to predict the flutter
derivatives of the rectangular section models without wind tunnel tests. Based on the given
flutter derivatives of the rectangular section models tested in wind tunnel, the prediction
frames of neural network are then established. The flutter derivatives of the rectangular
section models, with the B/D ratios other than those obtained from the wind tunnel tests can
be predicted by using this approach. The results show that this prediction scheme is
reasonably well. By using this ANN approach, the database of the aerodynamic coefficients of
bridge sections could be expanded.
The Fourth International Symposium on Computational Wind Engineering, 4p.