Performance of Damage Identification Based on Directional Wavelet Transforms and Entopic Weights Using Experimental Shearographic Testing Results

The paper aims to analyze the performance of the damage identification algorithms using the directional wavelet transforms, which reveal higher sensitivity for various orientations of spatial damage together with lower susceptibility to noise. In this study, the algorithms based on the dual-tree, the double-density, and the dual-tree double-density wavelet transforms were considered and compared to the algorithm based on the discrete wavelet transform. The performed analyses are based on shearographic experimental tests of a composite plate with artificially introduced damage at various orientations. It was shown that the directional wavelet transforms are characterized by better performance in damage identification problems than the basic discrete wavelet transform. Moreover, the proposed approach based on entropic weights applicable to the resulting sets of the detail coefficients after decomposition of mode shapes can be effectively used for automatic selection and emphasizing those sets of the detail coefficients, which contain relevant diagnostic information about damage. The proposed processing method allows raw experimental results from shearography to be significantly enhanced. The developed algorithms can be successfully implemented in a shearographic testing for enhancement of a sensitivity to damage during routine inspections in various industrial sectors.