Physics-based prediction, uncertainty quantification and decision-making for IN718 crystallographic texture intensity across LPBF defocus regimes

This study introduces a two‑stage physics‑based model to predict the build‑direction crystallographic texture intensity of Inconel 718 produced by laser powder bed fusion. Stage 1 maps process parameters to melting mode and melt‑pool geometry, while Stage 2 combines an empirical physics model with a random‑forest residual correction, attenuated by k‑nearest‑neighbor weighting and a beam‑power‑density gate. The physics‑anchored approach achieves substantially higher predictive accuracy (R² ≈ 0.78) than black‑box models and provides calibrated uncertainty estimates that allow selective withholding of predictions outside the model’s valid domain.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv Machine Learning.