A Imaging Method for Delamination Damage Based on Energy Mapping and Data-Driven Reconstruction

To prevent catastrophic failures and extend the service life of laminated composite structures, it is essential to promptly detect delamination within the structure. In this paper, a delamination imaging method based on 2D frequency domain spectroscopy and Segnet is proposed. A field expansion learning (FEL) strategy is introduced, focusing on a small wavefield region surrounding the defect in the experiments, and a small dataset is constructed. Numerical simulations are conducted to simulate wavefields with four times the field of view to increase dataset diversity. The two-dimensional FFT is used to extract the frequency domain spectra for model training. Testing results show that the model achieves high imaging accuracy on the test set and generalizes well to experimental samples with four times the field of view.