Physics-AI Foundation Models for Advanced Manufacturing

Advanced manufacturing involves complex physical phenomena spanning multiple spatial and temporal scales, from melt-pool dynamics and microstructure evolution to residual stress, distortion, and component-level performance. Conventional physics-based simulation provides high-fidelity insights but remains computationally expensive, while purely data-driven artificial intelligence often suffers from limited generalization, poor physical consistency, and a strong dependence on large datasets. This talk presents a vision for Physics-AI Foundation Models for Advanced Manufacturing, aiming to integrate physical laws, multiscale simulation, heterogeneous engineering data, and modern foundation-model architectures within a unified computational framework. The proposed paradigm combines physics-grounded representation learning, neural operators, multimodal and multi-fidelity learning, and cross-scale knowledge transfer to construct reusable models capable of predicting manufacturing processes across materials, geometries, process parameters, and physical scales. Particular emphasis is placed on metal additive manufacturing and welding, where Physics-AI models can learn mappings among process conditions, thermal-fluid behavior, microstructure evolution, residual stress, and final performance. By coupling AI models with CAE solvers, experimental observations, and uncertainty-aware verification, such models can evolve from task-specific surrogate models toward general-purpose engineering intelligence. Ultimately, Physics-AI foundation models may enable faster simulation, data-efficient process optimization, and trustworthy decision-making, providing a new computational foundation for intelligent and autonomous manufacturing.

Xuyang Chen is a Research Fellow at the Institute of Engineering Thermophysics, Chinese Academy of Sciences, and a Marie Skłodowska-Curie Fellow.

She obtained her Ph.D. from the Institute of Mechanics, Chinese Academy of Sciences, in 2018. She conducted postdoctoral research at Tsinghua University from March 2018 to December 2020. From March 2021 to March 2025, she carried out research at institutions including the Karlsruhe Institute of Technology (KIT), the Technical University of Munich (TUM), and ETH Zurich. She maintains ongoing collaborative relationships with various research institutions in Germany, Singapore, and Spain. Research Interests: AI-integrated process simulation for multi-scale metal additive manufacturing and welding;Development and integration of process simulation software;Development of artificial intelligence algorithms.