Optimal Damage Detection

  • Figure 1

  • Figure 2

  • Figure 3

  • Figure 4


Click here to return to the projects summary page.


Sponsors / Collaborators: Los Alamos National Laboratory

Ultrasonic guided waves, both coherent phase arrays and sparse arrays (sparsity defined as arrays whose average sensor-to-sensor distance is significantly longer than the interrogating wavelengths), have evolved into a very active research area. This research focuses on the detection, classification, and prognosis of damage using elastic waves as the interrogation mechanism. The novel approach in this work is the embedding of stochastic models to account for uncertainty of model/physical parameters, in order to derive an optimal detection process that supports predictive modeling with quantified uncertainty. Preliminary research focused on detecting small divots and through holes in a steel plate from elastic wave scattering, Figure 1. An array of piezoelectric transducers is used to detect scattered waveforms from damage, Figure 2.

Navier’s equation for a linear elastic material is used as the starting point for predicting the wave propagation and scattering of an induced stress waves. Numerical solution of the well-known Rayleigh Lamb frequency equations, which governs the propagation of elastic waves in a plate, is shown Figure 3. The curves are used to determine the wave propagation velocity for the symmetric and antisymmetric plate modes, which is used for

localizing damage. Detection is accomplished using generalized likelihood ratio test (GLRT) and Bayesian detectors in conjunction with a broadband beamformer to estimate the arrival angle of scattered waveforms. Figure 4, is the output of a GLRT array detector mapped over the plate; the red area indicates the detection of a small through hole. Current research is focused on determining the scattering patterns of induced plate waves from a through hole while taking into account variations due to operational variability, (i.e. temperature change). The scattering pattern and spatial distribution of scattered energy is then used to detect and classify the through hole.