EUVIMEDEUVIMED entdecken
Open Access · CC BY-NC-ND 2.5

Comparative Analysis of Machine Learning Models for Delamination Prediction in CFRP Drilling Using Intelligent Manufacturing Approaches

Bhanu Murthy Soppari, Preety Singh, Votarikari Naveen Kumar

Defence Science Journal · 27.10.2026 · 76 (6) · 932-941

DOI: 10.14429/dsj.22649 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Defence Science Journal, online erschienen am 27.10.2026. Verfasst von Bhanu Murthy Soppari und 2 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 2.5).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Machine learning (ML) methods used to study whether delamination could be predicted in CFRP samples being drilled with methods other than using empirical methods. Five sample types were manufactured via vacuum bagging methods and compared against neat carbon fibre prior to drilling via a Design of Experiment (DOE) conducted using the Taguchi method to investigate push-out delamination as a function of cutting speed, feed rate, and drill point angle. The use of three different types of computational models for measuring delamination (Random Forest, XG Boosting and Artificial Neural Networks) was also considered, together with Gradient Boosting as a fourth benchmark model evaluated under identical cross-validation for comparability. A high-resolution numerical database was generated using a digital microscope to capture drill point angles, spindle speeds and feed rates. Subsequent analysis identified that the originally reported R² values (0.97–0.9998) resulted from an input feature (“Readings”, the measured hole diameter) that is an exact function of the delamination factor being predicted — a data-leakage artefact rather than a learned relationship. With this feature removed and the delamination factor recomputed correctly as Fd = Dmax/D per Eqn. 4, none of the three process parameters, individually or in combination, showed a statistically significant relationship with push-out delamination in this dataset (one-way ANOVA across the 9 Taguchi parameter combinations: F = 0.87, p = 0.54), and all four cross-validated models performed at or below a naive mean-prediction baseline. We report this corrected, rigorously validated analysis and identify the likely missing predictors (nanofiller/material system, thrust force, tool wear) needed to achieve genuine predictive accuracy in future work.

Zitieren

APA

Soppari, B. M., Singh, P., & Kumar, V. N. (2026). Comparative Analysis of Machine Learning Models for Delamination Prediction in CFRP Drilling Using Intelligent Manufacturing Approaches. Defence Science Journal, 76(6), 932-941. https://doi.org/10.14429/dsj.22649

Vancouver

Soppari BM, Singh P, Kumar VN. Comparative Analysis of Machine Learning Models for Delamination Prediction in CFRP Drilling Using Intelligent Manufacturing Approaches. Defence Science Journal. 2026;76(6):932-941. doi:10.14429/dsj.22649

BibTeX
@article{Soppari2026,
  author = {Soppari, Bhanu Murthy and Singh, Preety and Kumar, Votarikari Naveen},
  title = {Comparative Analysis of Machine Learning Models for Delamination Prediction in CFRP Drilling Using Intelligent Manufacturing Approaches},
  journal = {Defence Science Journal},
  year = {2026},
  volume = {76},
  number = {6},
  pages = {932--941},
  doi = {10.14429/dsj.22649}
}
Kostenlos starten →