Delamination Prediction in Composite Beams with Built-In Piezoelectric Devices Using Modal Analysis and Neural Network

TLDR: In this article, the effect of prescribed delamination on natural frequencies of laminated composite beam specimens is examined both experimentally and theoretically, and backpropagation neural network models are developed using the results from the beam theory and used to predict delamination size.

Abstract: The effect of prescribed delamination on natural frequencies of laminated composite beam specimens is examined both experimentally and theoretically. Delamination is of particular interest because it can cause catastrophic failure of the composite structure. One consequence of delamination in a composite structure is a change in its stiffness. This change in stiffness will degrade the modal frequencies of the composite structure. Modal testing of a perfect beam and beams with different delamination size is conducted using polyvinylidene fluoride film (PVDF) sensors and piezoceramic (PZT) patch with sine sweep actuation. Modal testing of beams is also conducted using PVDF sensors and instrumented hammer excitation. The results of piezoceramic patch excitation and instrumented hammer excitation are discussed. The experimental modal frequencies are compared with the results obtained using a simplified beam theory. Also, backpropagation neural network models are developed using the results from the beam theory and used to predict delamination size. The effect of learning rate and momentum rate on neural network performance are discussed. Modal frequencies can be easily and accurately obtained with PZT patch excitation and PVDF sensing. There is good agreement between modal frequencies from modal testing and those from the simplified beam theory. The neural network models developed successfully predict delamination size.