Loading...

Investigating and Detecting Cracks to Assess the Health of Titanium Samples Made by Additive Method, with Machine Vision Method and Identifying the Issue

Mohammadi, Zahra | 2025

0 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 58670 (45)
  4. University: Sharif University of Technology
  5. Department: Aerospace Engineering
  6. Advisor(s): Adibnazari, Saeed
  7. Abstract:
  8. In this study, an intelligent model based on the YOLOv11 object detection network was developed to enhance the monitoring and detection of surface damages caused by fretting fatigue in titanium alloys. The main innovation of this research lies in the application of a data-driven and deep learning-based approach for the automated inspection of fretting surfaces in additively manufactured titanium alloys, a subject that has not been independently investigated before and is explored here for the first time. A dataset of microscopic images of damaged surfaces obtained from fretting tests was used to train and evaluate the model. Two configurations were designed and compared: a single-class model for crack detection and a dual-class model for simultaneous detection of cracks and interlayer delamination, aiming to assess the performance of the model under various industrial monitoring conditions. The experimental results demonstrated that both configurations achieved high accuracy, stability, and processing speed, enabling reliable and effective surface damage detection. Their performance indicates an optimal balance between detection accuracy and computational efficiency, confirming the feasibility of implementing the model in real-time monitoring systems and industrial production lines. Overall, the findings show that deep learning methods can automate and improve the precision and efficiency of surface damage monitoring in metals, representing a significant step toward developing intelligent inspection systems and advancing technology in high-tech industries, particularly in additive manufacturing and aerospace engineering
  9. Keywords:
  10. Machine Vision ; Deep Learning ; Crack Detection ; Fretting Fatigue ; You Only Look Once, Version 11 (YOLOv11)Model ; Additively Titanium Alloy ; Surface Damage Monitoring

 Digital Object List

 Bookmark

No TOC