Artificial neural network based multi-parameter inversion for the characterization of transversely isotropic composite lamina using velocity measurements of Lamb waves

Abstract

Artificial neural network (ANN) based multi-parameter inversion method is proposed to characterize transversely isotropic composite lamina using Lamb wave group velocity measurements. The ANN is first trained using numerical simulations and known micromechanics based formulae before being deployed on experimental samples. The group velocities obtained from the experiments were fed to the trained network. The network so trained, predicted the elastic properties, fiber volume fraction, and density.