Adaptive sparse reconstruction of damage localization via Lamb waves for structure health monitoring

The application of sparse reconstruction method for damage localization via ultrasonic Lamb waves in structure health monitoring is studied theoretically and experimentally in this paper. In this method, the oblique probes are used to detect damages in thin plate structures by recording all pairwise signals. By using the baseline signals of nondestructive structure and the sparsity of structural damages, a dictionary matrix is constructed through scattering signals from each grid of the detected area. The possible location of damage can be represented by atoms in the dictionary. In order to reduce the effect of noise and unknown sparsity, an adaptive BPDN algorithm is proposed for damage imaging. It combines greedy algorithm and convex optimization algorithm by firstly estimating a sparsity value as the initial iteration step to choose atoms that may contain damages and then renewing the dictionary. The ultrasonic Lamb wave image of the detected plate structure can be obtained by sparsely reconstructing the signals of the new dictionary, and the damages can be located in the image. The results of simulation and experiment manifested the effectiveness of the proposed method.