Evaluation of Laminar Defects in Beams Using Guided Waves and Pattern Recognition Techniques

Guided wave acousto-ultrasonics has recently become an important area of quantitative non-destructive evaluation research, since it has distinct advantages compared to conventional, high frequency non-destructive testing in particular for thin structures and one-dimensional structural elements such as beams. The paper reports on the investigation of a new damage detection methodology for beam structures using single mode acousto-ultrasonic guided waves. First order longitudinal waves are excited using a high-frequency vibrating piezoceramic transducer attached to one end of the beam. The measured transient wave response for the laminar defect is compared with simulated time histories based on analytical formulations of wave propagation in beams in which the damage is introduced as an inhomogeneity region in the beam. Advanced signal processing techniques, notably wavelet transforms, are applied to decompose the wave response time histories and to identify optimised parameters as input-target pairs for the quantitative detection and classification of the damage using neural network pattern recognition techniques. The final results show accurate classification of the damage parameters and an average error of less than 6% for approximating their quantities.