Pointwise Quantitative Depth Profile of Defects With Adjustable Resolution via Partial-Field Laser Ultrasonic-Guided Waves

Laser ultrasonic guided waves imaging technology is increasingly attractive to defect depth detection in industrial field, leveraging its non-contact measurement and capability for rapid full wavefield scanning. The traditional wavenumber-based methods can effectively process full wavefield measurement data acquired through laser ultrasonic technique for quantitative characterization of defect depth, but it has the drawback of insufficient accuracy in quantifying the depth of defects. Additionally, it necessitates complete full wavefield datasets, and its imaging efficiency is significantly constrained by computational complexity. To address this challenge, this study introduces a novel deep learning framework that incorporates spatial distribution information by employing localized spatial matrices of A0mode direct wave signals to train the model. This framework enables end-to-end pixel-level defect depth mapping and refined imaging of spatial depth distributions. Experimental validation on specially designed aluminum alloy specimens with varying thinning defect types demonstrates that the proposed method outperforms the frequency domain instantaneous wavenumber (FDIW) approach in defect depth visualization. Furthermore, by utilizing scanning point signals as the dataset, this methodology significantly reduces the complexity of experimental data acquisition while allowing adjustable imaging resolution of defect contours through scanning step size modulation. The research results hold substantial potential for practical engineering applications in large-scale inspection scenarios.