https://hal-normandie-univ.archives-ouvertes.fr/hal-01938721Nahraoui, YoussefYoussefNahraouiAassif, HouceinHouceinAassifMaze, GerardGerardMazeLOMC - Laboratoire Ondes et Milieux Complexes - ULH - UniversitĂ© Le Havre Normandie - NU - Normandie UniversitĂ© - CNRS - Centre National de la Recherche ScientifiquePREDICTION OF THE GROUP VELOCITY OF ACOUSTIC CIRCUMFERENTIAL WAVES BY ARTIFICIAL NEURAL NETWORKHAL CCSD2016Artificial Neural Network (ANN)Acoustic ResponseSubmerged Elastic ShellScattering WavesCircumferential WavesPhase VelocityGroup Velocity[PHYS] Physics [physics]LuppĂ©, Francine2018-11-28 20:17:102022-12-08 15:04:532018-12-07 14:50:29enJournal articlesapplication/pdf1The present study investigates the use an Artificial Neural Network (ANN) to predict the velocity dispersion curve of the antisymmetric (A 1) circumferential waves propagating around an elastic cooper cylindrical shell of various radius ratio b/a (a: outer radius and b: inner radius) for an infinite length cylindrical shell excited perpendicularly to its axis. The group velocity is determined from the values calculated using the eigen mode theory of resonances. These data are used to train and to test the performances of this model. Levenberg-Marquaedt backpropagation training algorithm with tangent sigmoid transfer function and linear transfer function results in best model for prediction of group velocity. The overall regression coefficient, mean relative error (MRE), mean absolute error (MAE) and standard error (SE) are 1, 0.01%, 0.38 and 0.07. It is found that the neural networks are good tools for simulation and prediction of some parameters that carry most of the information available from the response of the shell. Such parameters may be found from the velocity dispersion of the circumferential waves, since it is directly related to the geometry and to the physical properties of the target.