Robust decision making in damage detection using piezoelectric transducers and lamb wave propagation

In this paper, we present a robust quantitative decision-making methodology for structural damage detection using piezoelectric transducers and the Lamb wave approach with the consideration of environmental and operational variances. The features of wave propagation are extracted from multiple signals using an improved adaptive harmonic wavelet transform (AHWT), which are then denoised and highlighted by applying the principal component analysis (PCA) in the wavelet domain. After self-checking and updating the baseline dataset, the Hotelling's T-2 analysis provides a statistical indication of damage under a certain given confidence level. The detectability, sensitivity and robustness of the methodology are investigated using experimental as well as numerical studies. As the basis for the detection practice, the propagation of Lamb waves in an aluminum beam is systematically studied in this paper.