Application of artificial neural networks for quantitative damage detection in unidirectional composite structures based on Lamb waves

Abstract

This article provides a quantitative nondestructive damage detection method through a Lamb wave technique assisted by an artificial neural network model for fiber-reinforced composite structures. For simulating damages with a variety of sizes, rectangular Teflon tapes with different lengths and widths are applied on a unidirectional carbon fiber–reinforced polymer composite plate. Two characteristic parameters, amplitude damage index and phase damage index, are defined to evaluate effects by the shape of the rectangular damage in the carbon fiber–reinforced polymer composite plate. The relationships between the amplitude damage index and phase damage index parameters and the damage sizes in the carbon fiber–reinforced polymer composite plate are quantitatively addressed using a three-layer artificial neural network model. It can be seen that a reasonable agreement is achieved between the pre-assigned damage lengths and widths and the corresponding predictions provided by the artificial neural network model. This shows the great potential of using the proposed artificial neural network model for quantitatively detecting the damage size in fiber-reinforced composite structures.