Fast and Robust Damage Imaging With a Cascaded Deep Learning Technique

Abstract

Quantitatively measuring the health status of mechanical structures is a long-standing challenge in industrial fields. For plate structure inspection in tank and pressure vessels, traditional techniques using point by point scan with probes are tedious and time-consuming. Although algorithms like full waveform inversion (FWI) and diffraction tomography (DT) provide solutions to the problems, those methods are slow and suffer from convergence problems. In this article, we provided an effective damage imaging technique using dispersive A0 mode Lamb guided wave and an inversion algorithm called DLIS. The proposed algorithm adopts convolutional neural network (CNN) as first iteration to provide a fast and low-resolution background, and further optimizes the inversion results with descent direction matrix. In this way, the nonlinearity of the problem is effectively decomposed and yields better results. To test the robustness of the proposed method, we generated 1000 samples consists of corrosion defects with various sizes and shapes using 2D acoustic wave modeling. The inversion results prove the feasibility of our approach in plate heath monitoring and inspections. Note that this method also has full potential to be applied in the fast inspection of plates made of composite materials, pipes, geophysical prospecting and medical imaging because all these inverse problems share similar physics.