Artificial neural network-based damage detection of composite material using laser ultrasonic technology

Composite materials are widely used in various fields and non-destructive testing (NDT) is an important issue for these. Laser ultrasonic technology (LUT) can achieve the non-contact online NDT with high precision for composite materials. And the damage location depends on the differentiation of signals on defect and normal paths. However, manual interpretation of test images can easily lead to miss cause by limited accuracy. Based on the above background, a damage detection method based on artificial neural network (ANN) is proposed. First, the signals collected by LUT system are converted into energy ratios of components at different frequency by wavelet packet decomposition (WPD). Then, the preprocessed energy ratios are fed into a three-layer ANN to distinguish the damage. This algorithm has the characteristics of faster scanning, less calculation and greater accuracy. Results show that the detection accuracy of ANN is 99.6% for crack. Furthermore, the damage location and size can be accurately measured.

Automated summary

This study proposes a damage detection method based on an artificial neural network, which achieves high-precision, non-contact online non-destructive testing of composite materials through laser ultrasonic technology. Experimental results show that this method has faster scanning, less calculation, and greater accuracy, achieving a detection accuracy of 99.6% for cracks and accurately measuring the location and size of the damage.