Laser-generated guided wave imaging using the PSO-SSWT-FFT method

In this study, we propose an adaptive signal denoising method to overcome the challenges of multi-mode wave mixing and low signal-to-noise ratio (SNR) in laser-generated guided wave (LGGW) signals used for weld defect detection. First, the LGGW signals are converted via synchrosqueezed wavelet transform (SSWT), after which a two-dimensional Hanning window is used to extract the A0 guided wave. Subsequently, the particle swarm optimization (PSO) algorithm is applied to adaptively select the window parameters. The study addresses two key aims: efficient separation of A0 mode guided wave and significant noise reduction. Finally, we propose a scanning imaging method, PSO-SSWT-FFT, which uses the differences in guided wave energy between defective and non-defective areas to generate images. We conduct the PSO-SSWT-FFT scan, B-scan, and C-scan experiments. Additionally, we compare the performance of three signal processing techniques: PSO-SSWT, wavelet threshold (WT), and empirical mode decomposition (EMD). Results demonstrate that PSO-SSWT significantly outperforms traditional methods in extracting the A0 guided waves under high-noise environments. Compared to WT and EMD, the SNR is improved by 6.3 dB and 6.5 dB, respectively. In addition, PSO-SSWT-FFT successfully detects a through-hole with a diameter of Φ0.5 mm, exhibiting higher sensitivity than both B-scan and C-scan.