Identification Of Aerodynamic Parameters Of Bridge Section Using Artificial Neural Network

Static coefficients are defined to describe the static effect of wind on bridge, and the aerodynamic derivatives are used to describe the aerodynamic effect. Traditionally, all the aerodynamic parameters are obtained by experiment in the wind tunnel. But the investment for experiment is too large, and the experiment cycle is too long. This will restrict the research on the theory of wind effect on bridge. The purpose of this thesis is to use artificial neural networks (ANN) to identify the aerodynamic parameters.First introduced in this paper is the basic theory of ANN and wind engineering in bridge, then ANN is presented to identify the aerodynamic derivatives of ideal thin plate. A Back-Propagation (BP) ANN is established and trained for many times to contrast the influences of some factors on the prediction results. It is shown that this approach is feasible and can be applied in actual bridge engineering. Some experiments are done in the wind tunnel to acquire enough samples including static coefficients and aerodynamic derivatives. Then the study algorithm is improved to enhance the identification result of the aerodynamic derivatives, and different BP ANNs are built to identify the static coefficients. Finally Radial basis function (RBF) is adopted to identify the moment coefficient.The identification results show that the ANN has enough accuracy in identifying the aerodynamic parameters and can be used in practice. But the further research is needed.