VFEAgent: A Multimodal Agent Framework for End-to-End Automated Finite Element Analysis

Jiachen Zhang, Junyi Lao, Chenghao Liu, Siyuan Liu, Shixin Wu, Linsen Zhang, Boyu Wang, Songfang Huang

arXiv:2605.28978 · 2026-05-29 공개 · arXiv · PDF

vlm finite-element-analysis code-synthesis self-debugging engineering-design llm-integration simulation-verification automated-modeling

Abstract

Finite Element Analysis (FEA) serves as the cornerstone of modern engineering design. However, its workflow is inherently complex and relies heavily on domain expertise. Although recent efforts have integrated Large Language Models (LLMs) into FEA, existing approaches face limitations in handling multimodal inputs and executing complex tasks. To address these limitations, we propose VFEAgent, an end-to-end multi-agent system designed to automate FEA modeling and simulation directly from input images and problem descriptions. Our methodology integrates two core components: (1) a multimodal vision-language multi-agent pipeline that employs ReAct-driven reasoning to extract structured FEA specifications from heterogeneous inputs and (2) a verification-first code synthesis framework, incorporating robust self-debugging and fallback mechanisms to ensure executability and physical validity. We systematically evaluated the system across various engineering mechanics scenarios. The results demonstrate that VFEAgent achieves a high success rate in generating complete and physically valid simulations, outperforming LLM-based baseline methods in reliability and correctness. These findings validate the feasibility of automating the complete FEA workflow, highlighting the framework's potential to liberate engineers from tedious manual analysis.

한국어 요약

📋 한 줄 요약

**[엔지니어링 AI / FEA 자동화]** VFEAgent가 이미지·문제 설명만으로 FEA modeling·simulation 자동화 — ReAct multi-agent vision-language 파이프라인이 구조 specification 추출·verification-first code synthesis가 self-debug·fallback으로 실행성·물리 타당성 보장.

🎯 핵심 기여도

💡 핵심 아이디어

FEA workflow 자동화는 multimodal 입력(이미지·텍스트)을 ReAct 기반 multi-agent reasoning으로 structured specification으로 변환하고, verification-first code synthesis로 self-debug·fallback을 결합해 실행성·물리 타당성을 design-level로 보장하면 LLM baseline 대비 reliability에서 outperform할 수 있다.

🔬 기술적 접근법

📊 주요 결과

💭 의의 및 한계

**의의**: 엔지니어가 복잡 FEA 수작업에서 해방되는 비전 제시, ReAct multi-agent + verification-first의 일반 응용 가능 패턴, multimodal 입력 처리로 도면·이미지 활용 가능. **한계**: Engineering mechanics 중심으로 다양 FEA 도메인 일반화 추가 검증, multi-agent overhead·latency, vision pipeline의 도면 품질 의존, 물리 validity 검증의 완전성.

🚀 실용적 활용