EUVIMEDEUVIMED entdecken
Crossref

Modeling and Reconstruction of Air-Coupled Ultrasonic Signals via CNNs

Héctor Calas

e-Journal of Nondestructive Testing · 2026-07 · 31 (7)

DOI: 10.58286/32981 ↗

Kurz gefasst

Fachartikel in der Zeitschrift e-Journal of Nondestructive Testing, 2026-07. Verfasst von Héctor Calas.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

In air-coupled ultrasonic testing (ACUT), signal interpretation is often limited by strong noise and propagation distortions. This work introduces a data-driven reconstruction framework based on Convolutional Neural Networks (CNNs) guided by the concept of a Latent Ideal Signal (LIS) — the underlying, noise-free representation that expert operators intuitively infer from degraded measurements. Synthetic datasets simulating ACUT conditions between 250 kHz and 1 MHz were generated to train and evaluate several 1D CNN architectures. The deepest model, CNN_9C, achieved the best performance, improving the output Signal-to-Noise Ratio (SNR) from ≈20 dB to ≈30 dB and the Peak Signal-to-Noise Ratio (PSNR) to ≈42 dB, while maintaining real-time inference below 1 ms per A-scan. When applied to experimental data, the model nearly doubled the Contrast-to-Noise Ratio (CNR) of a CFRP C-scan, enhancing defect visibility and suppressing background noise. These results confirm that deep CNNs can effectively reconstruct the physical structure of ultrasonic echoes, bridging the gap between raw data and expert interpretation, and enabling AI-driven, high-fidelity inspection in non-contact ultrasonic testing.

Zitieren

APA

Calas, H. (2026). Modeling and Reconstruction of Air-Coupled Ultrasonic Signals via CNNs. e-Journal of Nondestructive Testing, 31(7). https://doi.org/10.58286/32981

Vancouver

Calas H. Modeling and Reconstruction of Air-Coupled Ultrasonic Signals via CNNs. e-Journal of Nondestructive Testing. 2026;31(7). doi:10.58286/32981

BibTeX
@article{Calas2026,
  author = {Calas, Héctor},
  title = {Modeling and Reconstruction of Air-Coupled Ultrasonic Signals via CNNs},
  journal = {e-Journal of Nondestructive Testing},
  year = {2026},
  volume = {31},
  number = {7},
  doi = {10.58286/32981}
}
Kostenlos starten →