Engineering Proceedings

Journal Browser

► Journal Browser

Eng. Proc., 2026, IOCAE 2026

Issues are regarded as officially published after their release is announced to the table of contents alert mailing list.
You may sign up for e-mail alerts to receive table of contents of newly released issues.
PDF is the official format for papers published in both, html and pdf forms. To view the papers in pdf format, click on the "PDF Full-text" link, and use the free Adobe Reader to open them.

Number of Papers: 1

22 pages, 2261 KB

Open AccessProceeding Paper

Physics-Regularized Transfer Learning for Cross-Alloy Generalization in Aerospace Additive Manufacturing

by Aswin Karkadakattil

Eng. Proc. 2026, 142(1), 1; https://doi.org/10.3390/engproc2026142001 - 3 Jun 2026

Abstract

Additive manufacturing (AM) is increasingly used in aerospace applications; however, most machine-learning (ML) models remain alloy-specific and require retraining when applied to new materials. To address this limitation, this study proposes a Physics-Regularized Transfer Learning (Physics-TL) framework for cross-alloy prediction in additive manufacturing. A neural network was first trained using Ti-6Al-4V as a source alloy and subsequently adapted to AlSi10Mg and 316L stainless steel using approximately 40 samples per alloy. Physics-based descriptors, including volumetric energy density and thermal diffusivity, were incorporated through a regularized loss function to improve physical consistency and data efficiency. The proposed framework was compared with a baseline neural network trained without transfer learning or physics-based constraints. Across repeated randomized train–test evaluations, the Physics-TL model achieved lower prediction errors, improved training stability, and better agreement with physically meaningful process–property relationships. Additional dataset sensitivity, ablation, and training–validation analyses provided supporting evidence of stable learning behaviour under limited-data conditions. Although the study is limited by dataset size and should be regarded as a proof-of-concept investigation, the results demonstrate the feasibility of combining transfer learning with physics-guided learning to support cross-alloy knowledge transfer in additive manufacturing. The proposed framework offers a promising pathway toward more data-efficient and physically informed predictive modelling for aerospace material qualification, process optimization, and future multi-material manufacturing systems. Full article

► Show Figures

Figure 1

Journal Menu

Journal Browser

Eng. Proc., 2026, IOCAE 2026

Other

Further Information

Guidelines

MDPI Initiatives

Follow MDPI