Damage imaging of an isotropic plate using matching pursuit algorithm

Abstract

A matching pursuit (MP) algorithm is effective tool to decompose the overlapped wave packets in a signal so that each wave mode can be identified. For the successful separations of the wave packets, an atom function should be properly designed, that can well resemble the physical features of the signal of interest. In this paper, a novel atom function for the MP algorithm is proposed based on the wave propagating model due to an excitation of a Hann-windowed toneburst signal, which performs very accurately compared to the MP algorithm with the existing Gaussian-type atom functions. The decomposed wave packets, including the directly scattered wave from damage as well as the reverberant waves from the free edges of the plate, via the MP method are employed in the damage imaging algorithm, highlighting the damaged location with higher intensity than the conventional algorithm utilizing only a direct reflected wave. The proposed approach is verified from the experiment where four piezoelectric wafers can accurately identify the damage location in a plate.