Structural K-space identification via ambient noise and its application to passive ice detection

Abstract

Ambient noise measured by a small number of receivers is used to identify the complex-valued wavenumber trajectory, known as the K-space, of a thin-wall structure. Its real and imaginary parts correspond to the dispersion and attenuation curves, which both include important information about structural health conditions. A sensor placement strategy with unequal spacings is proposed to optimize the spatial sampling. The cross-correlation between the random ambient noise measured by each pair of sensors is computed to estimate the structural impulse response, after performing a post-processing of whitening and coda wave elimination. The wavenumber trajectory of K-space is then estimated by matching the refined cross-correlation with the Green’s function of the wave equation. Experimental results with various flow-induced or mechanical vibration excitations from a household hairdryer to an axial flow fan justify the robustness of the proposed method, where no more than eight receivers are required. An application of icing detection is presented, where the dispersion curve shift and wave energy attenuation due to ice accretion on a thin-wall structure are respectively captured by the estimated real and imaginary parts of the K-space. As a result, ice thickness estimation and icing warning algorithms are both explicitly given.