Estimation of the vibration frequencies of thin rectangular plates by artificial neural networks approach

Abstract

An artificial neural network application is presented for vibration analysis of rectangular plates with 21 various support conditions. The first three natural frequencies of two sequences mode of plates are obtained using multi-layer neural network based back-propagation error learning algorithm. The training of the network has been made using data that obtained the well-known paper of Leissa [37], for 11 different support conditions. The trained neural network, however, had been tested for 10 other different support conditions which were not included in the training set. The results found by using artificial neural network are sufficiently close to the numerical results.