Dictionary design for Lamb wave sparse decomposition

Lamb wave is a promising tool for nondestructive testing and evaluation (NDT&E). Due to dispersion and multi-mode, wave components generally overlap with each other at Lamb wave signal, which makes the interpretation difficult. To solve the overlapping problem, Lamb wave sparse decomposition has been developed in the past decade. However, the performance is limited because original tone-burst or dispersive tone-burst is directly used as dictionary atom, without fully considering the waveform change during propagation. In this paper, to overcome this limitation, a systematic dictionary design scheme is established by taking excitation waveform, dispersion, multi-mode, amplitude modulation and environmental condition into comprehensive consideration. First, pulse compression technique is applied for signal-to-noise (SNR) improvement, and auto-correlations of various excitation waveforms are considered as the initial dictionary atoms. Second, dispersion and multi-mode are placed on excitation auto-correlations for dictionary modification. Third, amplitude modulation during Lamb wave propagation is considered for further dictionary modification. Finally, the effect of environmental condition on dictionary design is investigated to make it adaptive to different conditions. An experiment is given to validate the established dictionary design scheme.