A16-a4 - Enhancing air-coupled ultrasonic inspection: resolving signal overlapping in long-duration chirp signals
- Event
- 2025 ICU PADERBORN - 9th International Congress on Ultrasonics
2025-09-21 - 2025-09-25
Paderborn - Band
- Lectures
- Chapter
- A16-a - Ultrasonic Signal Processing and Instrumentation
- Author(s)
- M. Tayyib, P. Tervydis, L. Svilainis, M. Yousufi - Kaunas University of Technology, Kaunas (Lithuania)
- Pages
- 118 - 121
- DOI
- 10.5162/Ultrasonic2025/A16-a4
- ISBN
- 978-3-910600-08-9
- Price
- free
Abstract
Air-coupled ultrasound investigations are crucial for non-destructive testing of thin and delicate materials. Conventional pulse excitation yields suboptimal signal-to-noise and narrow bandwidth. Chirp signals improve these parameters, however, requires long duration to do so, which results in overlapping of reflections. To solve the overlapping this paper proposed the use of residual deep learning approach using convolutional UNet1D architecture and composite loss function. Results show successful separation of overlapping reflections.