EUVIMEDEUVIMED entdecken
Open Access · CC BY-NC-ND 4.0

Data-driven monitoring of laser metal deposition for enhanced process stability

Bohdan Vykhtar, Vishnuu Jothi Prakash, Markus Kogel-Hollacher, Heinz-Ingo Schneider, Ingomar Kelbassa

Journal of Laser Applications · 01.11.2026 · 38 (4)

DOI: 10.2351/7.0001872 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Journal of Laser Applications, online erschienen am 18.11.2026. Verfasst von Bohdan Vykhtar und 4 weiteren Autor:innen. Die Arbeit nennt 21 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Laser metal deposition (LMD) is an additive manufacturing (AM) process for producing near-net-shape parts. Maintaining consistent part quality over several production cycles remains challenging due to accumulating process instabilities. This work focuses on the integration of sensor systems and the development of a data-driven digital shadow to monitor and analyze long-term process behavior under constant process conditions. A comprehensive sensing architecture, including process and environmental sensors, as well as an industrial edge device for synchronized data acquisition up to 500 Hz were integrated. Twenty print jobs were conducted over several months, resulting in 320 Ti-6Al-4V specimens manufactured under fixed process parameters and characterized using destructive mechanical testing. The multisensor dataset was analyzed using exploratory statistical methods and multivariate machine learning models. Pairwise dependencies between sensor signals and mechanical quality metrics were investigated using Spearman rank correlation to capture monotonic relationships (ρ) and distance correlation (dCor) to identify nonlinear dependencies. In addition, machine learning models were trained to describe the combined influence of all sensor signals on the respective quality metrics. Model performance was evaluated using the coefficient of determination (R2) and mean squared error (MSE). Exemplary, across all sensors, one of the best-performing models achieved R2 value of 0.47 with an MSE of 234 for tensile strength. Melt pool width, optical coherence tomography based standoff distance, and machine vibration were identified via correlation analysis and in-sample attribution as the most influential signals for predicting density, hardness, and elongation at break.

Zitieren

APA

Vykhtar, B., Prakash, V. J., Kogel-Hollacher, M., Schneider, H., & Kelbassa, I. (2026). Data-driven monitoring of laser metal deposition for enhanced process stability. Journal of Laser Applications, 38(4). https://doi.org/10.2351/7.0001872

Vancouver

Vykhtar B, Prakash VJ, Kogel-Hollacher M, Schneider H, Kelbassa I. Data-driven monitoring of laser metal deposition for enhanced process stability. Journal of Laser Applications. 2026;38(4). doi:10.2351/7.0001872

BibTeX
@article{Vykhtar2026,
  author = {Vykhtar, Bohdan and Prakash, Vishnuu Jothi and Kogel-Hollacher, Markus and Schneider, Heinz-Ingo and Kelbassa, Ingomar},
  title = {Data-driven monitoring of laser metal deposition for enhanced process stability},
  journal = {Journal of Laser Applications},
  year = {2026},
  volume = {38},
  number = {4},
  doi = {10.2351/7.0001872}
}
Kostenlos starten →