ABSTRACT
This paper presents an evaluation of the performance of signal processing schemes in guided wave damage identification. A Bayesian probabilistic approach is proposed to not only provide quantitative identification of damages, but also quantify the uncertainties associated with the damage identification results. The levels of accuracy and degree of uncertainty associated with the damage identification results using data extracted by four signal processing schemes, Hilbert transform, Fast Fourier transform, Gabor wavelet transform and discrete wavelet decomposition, were compared in detail. The guided wave signals used in this study were experimentally measured from beams with a single damage. Two damage cases were considered in the study and the results shows that a suitable signal processing scheme produces more robust identification of damage in a structure.
