open access

https://hdl.handle.net/2115/100468

A Model Context Protocol Framework for Bridge Data Generation with Large Language Model Integration

英語

Finite Element Analysis

2026-07-01

Proceedings of the 10th International Symposium on Reliability Engineering and Risk Management (ISRERM 2026)

Large Language Models (LLMs) have demonstrated strong capabilities in natural language understanding and reasoning, however, integrating them with domain-specific engineering tools remains a persistent challenge. This paper introduces BrDSL-MCP, a Model Context Protocol (MCP) server designed to connect conversational AI interfaces with structural engineering workflows for bridge specification data generation. The proposed system supports real-time interaction between LLMs and bridge engineering tools, enabling parametric design generation, finite element analysis (FEA) via Abaqus, within a unified, natural-language-driven loop. We detail the architectural choices that facilitate reliable tool composition and demonstrate how the system supports bridge data generation through conversational interaction. Case studies indicate that engineers can complete design iterations faster than with conventional manual workflows, while retaining full authority over final engineering decisions.

会議発表論文

Version of Record