EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Automated damage detection and localization in acoustic images using autoencoder-based dynamic mode decomposition

Rishant Pal, Anowarul Habib

Smart Materials and Structures · 01.06.2025 · 34 (6) · 065029

DOI: 10.1088/1361-665x/ade20d ↗

Kurz gefasst

Fachartikel in der Zeitschrift Smart Materials and Structures, online erschienen am 19.06.2025. Verfasst von Rishant Pal und 1 weiteren Autor:innen. Die Arbeit wurde bisher 5-mal zitiert (Crossref) und nennt 51 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Quantifying reliability is essential in evaluating and assessing inspection technologies, whether they fall under nondestructive evaluation or structural health monitoring methods. A key challenge in computer vision is uncovering the underlying dynamics of objects within image sequences, especially in scenarios involving damage. Lead zirconate titanate (PZT) is a piezoelectric material that is highly effective in generating ultrasonic waves for various engineering and scientific purposes. An experimental method utilizing Coulomb coupling has been developed to visualize ultrasonic wave interactions with microscale defects in PZT. A modified convolutional autoencoder-based dynamic mode decomposition (CAE-DMD) algorithm was developed and applied to spatiotemporal data from healthy and damaged conditions to capture image dynamics. The modified CAE-DMD algorithm is applied in a reference-free manner, enabling the identification of both the background mode and the specific damage location without the need for healthy state data. The framework was validated under two scenarios: one involving only damaged specimens and another incorporating both damaged and healthy specimens. The algorithm demonstrated exceptional performance in both cases, achieving a damage detection and localization accuracy of 96%, verified through visual evaluation. Including an image registration framework and damage localization techniques based on K -means clustering and contour detection further improved the system’s precision and reliability.

Zitieren

APA

Pal, R., & Habib, A. (2025). Automated damage detection and localization in acoustic images using autoencoder-based dynamic mode decomposition. Smart Materials and Structures, 34(6), 065029. https://doi.org/10.1088/1361-665x/ade20d

Vancouver

Pal R, Habib A. Automated damage detection and localization in acoustic images using autoencoder-based dynamic mode decomposition. Smart Materials and Structures. 2025;34(6):065029. doi:10.1088/1361-665x/ade20d

BibTeX
@article{Pal2025,
  author = {Pal, Rishant and Habib, Anowarul},
  title = {Automated damage detection and localization in acoustic images using autoencoder-based dynamic mode decomposition},
  journal = {Smart Materials and Structures},
  year = {2025},
  volume = {34},
  number = {6},
  pages = {065029},
  doi = {10.1088/1361-665x/ade20d}
}
Kostenlos starten →