Deep artificial neural network-based fatigue life modeling for Ti-6Al-4V alloy: Effects of heat treatment and surface defects


Filiz H. I., ATEŞ H., Bakir M., Aras C.

INTERNATIONAL JOURNAL OF FATIGUE, cilt.216, 2027 (SCI-Expanded, Scopus)

  • Yayın Türü: Makale / Tam Makale
  • Cilt numarası: 216
  • Basım Tarihi: 2027
  • Doi Numarası: 10.1016/j.ijfatigue.2026.110068
  • Dergi Adı: INTERNATIONAL JOURNAL OF FATIGUE
  • Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Compendex, INSPEC, zbMATH, Academic Search Ultimate (EBSCO), Engineering Source (EBSCO)
  • Gazi Üniversitesi Adresli: Evet

Özet

Understanding the influence of surface defects on fatigue behavior is essential for assessing the durability of aerospace alloys. This study investigated the fatigue behavior of annealed and (3-STOA Ti-6Al-4 V alloys containing EDM-induced surface defects under different stress ratios. A deep artificial neural network (deep ANN) framework was developed for fatigue life and S-N curve modeling. Following hyperparameter optimization, the deep ANN achieved testing values of R2 = 0.986, RMSE = 0.142, and MAE = 0.113, outperforming the shallow ANN. The modeled S-N curves reflected the combined effects of heat treatment, loading conditions, and defect depth. Fatigue knockdown factors for annealed specimens containing 0.25 mm and 0.10 mm scratches were approximately 70.9 % and 56.6 %, respectively. SEM observations revealed multiple crack initiation sites along the scratch root. These findings emphasize the importance of surface defects in fatigue assessment and demonstrate the applicability of deep ANN modeling to the investigated conditions.