Multiresolution wavelet analysis to estimate Lamb waves direction of arrival in passive monitoring techniques

Abstract

Impact damage may compromise the structural integrity and resistance. When an impact occurs, it is important to check if a structural damage was originated at the impact location. On thin structures, impacts generate guided stress waves (Lamb waves) which can be analyzed to infer the actual impact position. A novel strategy to determine the Lamb wave direction of arrival, generated by an impact occurring in thin plate structures, is proposed in this paper. As already known, the wave mode propagation of GWs (i.e. Lamb Waves) along thin structures is characterized by a dispersive behaviour which could hinder an accurate estimation of the wave direction of arrival when monitoring the DToA associated to different sensors. Evidently, this is to be avoid in SHM (Structural Healt Monitoring) techniques. Overcoming this limitation is the aim of the paper, therefor a new estimation technique in the wavelet domain has been carried out for passive sensors. The proposed procedure, based on the CWT decomposition of non-linear signals acquired by three closely-located piezoelectrics transducers, causes a resampling of the time-frequency plane allowing to separate the wave components travelling at different velocities as well as to highlight the frequency dependence of the DToA. The multiresolution analysis has the advantage that the DToA among two different transducers can be computed multiple times, at different frequencies range, by applying the cross-correlation method in the CWT domain for each computed scale. In this way, the GW direction of arrival can be estimated making an averaging procedure across scales to the arc-tangent of the ratio between the DToA among specific sensors. The optimization of the transducers disposition on the plate has been also carried out and, in the end, experimental results are shown.