Available online 14 November 2016Reza Soleimanpour, Ching-Tai N

An analysis of the scattering characteristics of the fundamental anti-symmetric (A0) Lamb wave at a delamination in a quasi-isotropic composite laminate is presented. Analytical solutions for this problem do not exist due to the anisotropic nature and multilayer characteristics of composite laminates. This study uses a three-dimensional finite element (FE) method and experimental measurements to provide physical insight into the scattering phenomena. Good agreement is found between simulations and experimental measurements. The results show that the A0 Lamb wave scattering at a delamination in composite laminates is much more complicated than the scattering at a defect in isotropic plates. Scatter amplitudes and scatter directivity distributions depend on the delamination size to wavelength ratio and the through-thickness location of the delamination damage. The study also investigates the feasibility of the common experimental practice of simulating delamination damage by bonding masses to the surface of composite laminates for guided wave damage detection and characterization methodologies verifications. The results suggest that care is required to use bonded masses to simulate delamination damage for verifying and optimizing damage characterization techniques. In summary, the results of the investigation help to further advance the use of the A0 Lamb wave for damage detection and characterization.

Detection and characterization of delamination damage are of great importance to the assurance of structural safety. This work investigates the potential of a baseline-free structural health monitoring technique based on higher harmonics resulting from the nonlinear interaction of guided wave and a delamination. The nonlinearity considered in this study arises from the clapping of the sub-laminates in the delaminated region, which generates contact acoustic nonlinearity. Both explicit finite element simulations and experimental tests are conducted on composite laminates containing a delamination of different sizes and at different through-thickness locations. The results show that the interaction between the fundamental asymmetric mode ( A₀) of guided wave and a delamination generates contact acoustic nonlinearity in the form of higher harmonics, which provides a good measure for identifying the existence of delaminations and determining their sizes in laminated composite beams. This new insight into the generation mechanisms of nonlinear higher order harmonics in composite laminates will enhance the detection and monitoring of damage in composite structures.

Pattern recognition is a promising approach for the detection of structural damage using measured dynamic data. Much research of pattern recognition has employed artificial neural networks (ANNs) as a systematic way of matching pattern features. When such methods are used, the ANN design becomes the most fundamental factor affecting performance and effectiveness of the pattern recognition process. The Bayesian ANN design algorithm is proposed in Lam et al. provides a mathematically rigorous way of determining the number of hidden neurons for a single-hidden-layer feedforward ANN. The first objective of this paper is to extend this Bayesian ANN design algorithm to cover the selection of activation (transfer) functions for neurons in the hidden layer. The proposed algorithm is found to be computationally efficient and is suitable for real-time design of an ANN. As most existing ANN design techniques require the ANN model class to be known before the training process, a technique that can automatically select an "optimal" ANN model class is essential. As modal parameters and Ritz vectors are commonly used pattern features in the literature, the second objective of this paper is to compare the performance of these two pattern features in structural damage detection using pattern recognition. To make a fair judgment between the features, the IASC-ASCE benchmark structure is employed in a case study. The results show that the performance of ANNs trained by modal parameters is slightly better than that of ANNs trained by Ritz vectors.