Leveraging Generative AI to Modify and Query MBSE Models

Abstract

Generative AI tools, such as large language models (LLMs), offer a variety of ways to gain efficiencies and improve systems engineering processes from requirements generation and management through design analysis and formal testing. Large acquisition programs may be particularly well poised to take advantage of LLMs to help manage the complexities of system and system of systems acquisitions. However, generative AI tools are prone to a variety of errors. Our research explores the ability of current LLMs to generate, modify, and query Systems Modeling Language (SysML) v2 models. Techniques such as Retrieval-Augmented Generation (RAG) are utilized to add domain-specific knowledge to an LLM and improve model accuracy. A preliminary case study is presented where the number of prompts to generate the models is minimized. We also discuss the limitations of LLMs and future systems engineering research related to LLMs.