Identification of mechanical properties of adhesive joints using ANN

This paper aims to present a new approach to the fast determination of the effective, dynamic, mechanical properties of an adhesive for linear and nonlinear regions of the adhesive response, for both healthy and damaged states of the bond.

The proposed approach is based on the measurement of the linear and nonlinear frequency response function (FRF) of adhesive-bonded structure and using artificial neural network identification technique. For this purpose, linear and nonlinear FRFs are measured for several single-lap joint specimens that are fabricated in healthy and damaged configurations of the bond. The measured FRFs of healthy and damaged specimens are then used to identify the natural frequencies of the specimens. The experimental natural frequencies, in turn, would be used to train artificial neural network (ANN) which would be able to predict the effective Young’s and shear moduli and damping of adhesive in healthy and damaged specimens, for any given excitation level and frequency, within the training domain.

Simultaneous identification of the effective mechanical properties of adhesive for linear and nonlinear response regions, as well as healthy and damages states of the adhesive bond.

The introduced method is effective to model the assembled structures with the viscoelastic adhesive joints, for linear and nonlinear regions.

A fast methodology, using ANN, for identification the effective mechanical properties of adhesives, compared to other methods for both linear and nonlinear regions.