Application of Deep Learning for Defect Detection by X-Ray Computed Tomography in Large Aerospace Structures


Metiner A., Nikishkov Y., Makeev A., KOÇYİĞİT M. T.

AIAA Science and Technology Forum and Exposition, AIAA SciTech Forum 2025, Florida, Amerika Birleşik Devletleri, 6 - 10 Ocak 2025, (Tam Metin Bildiri)

  • Yayın Türü: Bildiri / Tam Metin Bildiri
  • Doi Numarası: 10.2514/6.2025-0633
  • Basıldığı Şehir: Florida
  • Basıldığı Ülke: Amerika Birleşik Devletleri
  • Boğaziçi Üniversitesi Adresli: Evet

Özet

Structural health monitoring has an essential role in the maintenance of aerospace structures. An important part of structural health monitoring is nondestructive inspection (NDI) that may include various techniques such as visual inspection, ultrasonic and acoustic testing, and X-ray Computed Tomography (CT). Among them, X-ray CT stands out as a technique that has unmatched fidelity leading to a wide range of applications. However, large aerospace structures, such as helicopter blades, aircraft wings or fuselage, are difficult to inspect using a standard X-ray CT due to size limitations of the technique. Inclined CT is a novel method recently developed to overcome the size limitations, making use of limited angle views and 3D reconstruction. Due to incomplete viewing range, the result of the 3D reconstruction may have artifacts limiting recognition of target defects. Deep neural network-based techniques are capable of improved recognition of defects thanks to recent achievements in image segmentation. This study shows how the use of machine learning (ML) can help detect defects with considerably less artifacts than the state-of-the-art CT reconstruction techniques. Segment Anything Model of META AI was used as the basis of ML image segmentation. Current study and further research into ML modeling aims to make CT for large structures an accepted NDI method for the aerospace industry.