Wavelet power, entropy and bispectrum applied to AE signals for damage identification and evaluation of corroded galvanized steel

Acoustic emission (AE) signals obtained from scratch tests on hot dip galvanized samples with different corrosion levels were processed by wavelet transform (WT) analysis. Wavelet power was distributed in different frequency bands, according to damage mechanisms. The frequency bands were automatically obtained by searching for the relative minima of the wavelet entropy of signals and appropriate clustering methods. Correlation between the different mechanisms was corroborated by bispectrum analysis (BA). The damage evaluation entailed studying the evolution of the wavelet power in a specific frequency band, which corresponded to the fracture of the zeta phase columns of the galvanized coating. Results showed damage to increase along with the level of corrosion, but adherence was not dramatically affected in the studied corrosion range. The application of two signal-processing techniques, WT and BA, contributed to the consistency of our results. Besides the addressed technological application, we could demonstrate that that signal-processing techniques, when applied carefully, and results classified with care, are able to contribute to what is certainly an important problem, specially in cases like the treated here, where a complete physical theory relating damage and AE is not yet available.