报告题目：Toward Knowledge-Guided AI for Mechanics and Manufacturing Physics-Informed Learning and Multi-Agent LLM Systems

报告人：Seunghwa Ryu教授，韩国科学技术院

报告时间：2026年1月15日（周四）14:00开始（北京时间）

报告地点：线上Zoom会议（会议号: 8208950817）

主办单位：江苏省力学学会信息化工作委员会

 91麻豆-麻豆传媒-91视频

 91麻豆 人工智能与自动化学院

欢迎广大师生参加！

报告摘要：

Artificial intelligence is rapidly transforming the design of materials and manufacturing processes. However, in engineering applications, purely data-driven approaches often face fundamental limitations due to data scarcity, complex physics, and the need for interpretability. In this talk, I present a knowledge-guided AI framework that integrates physics-informed machine learning (PIML) or multi-agent large language model (LLM) systems to address these challenges. I begin with an overview of AI paradigms in engineering, spanning from purely data-driven models to knowledge- and physics-guided approaches. In the first part, I demonstrate how PIML enables quantitative prediction and inverse analysis for hyperelastic, plastic, and thermoelectric materials by embedding governing equations directly into the learning process, achieving robust material characterization even with minimal data. In the second part, I introduce multi-agent LLM systems that support human engineers through natural-language interaction, assisting in injection molding process design and numerical simulation workflows. Together, these examples illustrate how domain knowledge, physical laws, and language-driven human-AI collaboration can form a reliable, interpretable, and scalable foundation for next-generation AI-driven mechanics and manufacturing.