Identification of multiple cracks in noisy conditions using scale-correlation-based multiscale product of SWPT with laser vibration measurement

Abstract

Simultaneous identification of multiple cracks is an unresolved issue in the structural health monitoring field. To address this issue, a new method is proposed for identifying multiple structural cracks using the scale-correlation-based multiscale product of the stationary wavelet packet transform (SWPT) of high-order (greater than 6) mode shapes, with particular emphasis on the ability of the method to distinguish damage in noisy conditions. With the method, the SWPT is firstly utilized to decompose a noisy high order mode shape into a set of uniform-frequency sub-bands. In each sub-band the SWPT coefficients have the property of shift invariance. Secondly, a scale correlation is introduced to screen out valid SWPT sub-bands containing an informative crack feature. On these sub-bands, manipulation of scale-correlation-based multiscale products is implemented to yield a multiscale product curve by multiplying sub-band coefficients across scales at every temporal step. In the curve, abrupt peaks form damage indicators that can pinpoint the locations of multiple cracks. In the damage indicator, the function of scale correlation in distinguishing effective sub-bands endows the scale-correlation-based multiscale product with strong ability to reveal damage while suppressing noise. The method is numerically verified using high-order mode shapes of beam-type structures with multiple cracks and experimentally validated on high-order mode shapes acquired using a scanning laser vibrometer, both demonstrating high accuracy and strong robustness against noise in the identification of multiple cracks. The proposed scale-correlation-based multiscale product forms a sophisticated mechanism for identifying multiple damage in noisy conditions, with promise for developing damage identification methods for a wide spectrum of structures.