Delamination Assessment of Multilayer Composite Plates Using Model-based Neural Networks

Abstract

A procedure for damage detection in multilayer composites is described using model-based neural networks and vibration response measurement. The appropriate finite element model is established to generate the training data of neural networks. Internal delaminations with different sizes and locations are considered as the particular damage scenarios in multilayer composite plates. The damage-induced energy variation of response signal is investigated, and the mechanism of mode-dependent energy dissipation of composite plates due to delamination is revealed. In order to obtain the structural dynamic response of the samples, impulse forced vibration testing is conducted using a piezoelectric patch actuator and an accelerometer. To enhance the sensitivity of damage features in the vibrating plate, the damage-induced energy variation of the response signal decomposed by wavelet packets is used as the input data of backward propagation neural networks for the prediction of delamination size and location. The test results show that the proposed method is effective for the assessment of delamination status in composites.