Experimental Study of Passive Defect Detection and Localization in Thin Plates from Noise Correlation

Abstract

This paper reports experimental results on a passive imaging technique for structural health monitoring to detect the occurrence of defects in plate-like structures. This technique is based on the fact that the active transient response between two sensors can be passively retrieved by cross-correlating the ambient noise-field recorded on these two sensors. A correlation matrix is estimated from friction noise recorded on a transducer array. It is observed that the accuracy of the estimated transient responses strongly depends on the spatial distribution of noise sources. The best convergence is obtained when the noise is uniformly distributed over the whole plate area. Defects are localized by applying a dispersive beamforming algorithm to the difference between the correlation matrices obtained with and without (w/o) defect. It is shown that the quality of the active transient response reconstruction is not a strong requirement for the defect localization. Indeed, the defect is successfully localized even if the noise source distribution is not uniform, provided that it remains spatially stationary between the states w/o defect. A simple theoretical framework is proposed to interpret these results.