Ameliorated-multiple signal classification (Am-MUSIC) for damage imaging using a sparse sensor network

Abstract

Multiple Signal Classification (MUSIC) – a directional scanning and searching algorithm, has gained its prominence in phased array-facilitated nondestructive evaluation. Nevertheless, prevailing MUSIC algorithms are largely bound up with the use of a dense linear array, which fail to access the full planar area of an inspected sample, leaving blind zones to which an array fails to scan, along with the incapability of differentiating multiple damage sites that are close one from another. To break above limitations, conventional MUSIC algorithm is ameliorated in this study, by manipulating the signal representation matrix at each pixel using the excitation signal series, instead of the scattered signal series, which enables the use of a sparse sensor network with arbitrarily positioned transducers. In the ameliorated MUSIC (Am-MUSIC), the orthogonal attributes between the signal subspace and noise subspace inherent in the signal representation matrix is quantified, in terms of which the Am-MUSIC yields a full spatial spectrum of the inspected sample, and damage, if any, can be visualized in the spectrum. Am-MUSIC is validated, in both simulation and experiment, by evaluating single and multiple sites of damage in plate-like waveguides with a sparse sensor network. Results verify that i) detectability of Am-MUSIC-driven damage imaging is not limited by damage quantity; ii) Am-MUSIC has full access to a sample, eliminating blind zones; and iii) the amelioration expands conventional MUSIC from phased array-facilitated nondestructive evaluation to health monitoring using built-in sparse sensor networks.