Material-Agnostic Zero-Shot Thermal Inference for Metal Additive Manufacturing via a Parametric PINN Framework

2026-04-16 · Source: Machine Learning · Field: Manufacturing & Industrial — Smart Manufacturing & Industry 4.0, Materials & Production Technology · Depth: Expert, quick

Summary

A new parametric physics-informed neural network (PINN) framework has been developed for zero-shot thermal inference in metal additive manufacturing (AM), specifically for bare plate laser powder bed fusion (LPBF). This framework enables generalization across arbitrary materials without requiring labeled data, retraining, or pre-training, addressing a significant challenge in AM thermal modeling. It features a decoupled architecture that encodes material properties and spatiotemporal coordinates separately, integrating them via conditional modulation to reflect the multiplicative influence of material parameters in governing equations. The framework also incorporates physics-guided output scaling, derived from Rosenthal's analytical solution, and a hybrid optimization strategy to improve physical consistency, training stability, and convergence. Experimental results show up to a 64.2% reduction in relative L2 error compared to non-parametric baselines, achieving superior performance with only 4.4% of the baseline's training epochs across diverse metal alloys, including out-of-distribution cases.

Key takeaway

For AI Scientists and Machine Learning Engineers developing thermal models for additive manufacturing, this parametric PINN framework offers a robust solution for material generalization. You can achieve significant reductions in training time and error rates, enabling more flexible and practical deployment of AM process simulations without extensive material-specific data or retraining.

Key insights

A parametric PINN framework enables zero-shot thermal modeling across diverse materials in metal AM with high efficiency.

Principles

• Decouple material properties from spatiotemporal coordinates.
• Align network architecture with governing equation structure.
• Incorporate physics-guided scaling for consistency.

Method

The framework uses a decoupled parametric PINN architecture with conditional modulation, physics-guided output scaling based on Rosenthal's solution, and a hybrid optimization strategy for enhanced training and generalization.

In practice

• Apply decoupled PINN for material-agnostic modeling.
• Use Rosenthal's solution for output scaling.
• Implement hybrid optimization for PINN stability.

Topics

• Parametric PINN Framework
• Metal Additive Manufacturing
• Zero-Shot Generalization
• Thermal Modeling
• Laser Powder Bed Fusion

Best for: AI Scientist, Research Scientist, Machine Learning Engineer

Related on AIssential

• Material-Agnostic Zero-Shot Thermal Inference for Metal Additive Manufacturing via a Parametric PINN Framework — A parametric PINN enables zero-shot thermal modeling across diverse materials in metal AM, significantly reducing error and training time.
• Finetuning-Free Diffusion Model with Adaptive Constraint Guidance for Inorganic Crystal Structure Generation — Adaptive constraint guidance in diffusion models enables finetuning-free generation of inorganic crystal structures with targeted properties.
• Physics-Informed Neural Network Modeling of Biodegradable Contaminant Transport through GCL/SL Composite Liners — Hard-constrained PINNs significantly improve accuracy and stability for modeling contaminant transport in composite liners.
• Finite Element-Based Material Learning via Automatic Differentiation: Learning constitutive neural network models from full-field deformation data — FE-MAD uses automatic differentiation within a JAX-FEM solver to learn constitutive neural network models from full-field deformation data.
• Force-free molecular dynamics through autoregressive equivariant networks — TrajCast uses equivariant MPNNs to directly predict molecular states, enabling faster, data-efficient MD simulations with strong generalization.
• They Said It Was Impossible… This Simulation Solved It — Numerical homogenization enables simulating complex granular materials by characterizing bulk properties rather than individual particles.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Machine Learning.