Scholarlink

<p>This paper discusses innovative approaches to model-based systems engineering (MBSE) in the context of the space industry by presenting the results of a digital assistant that implements artificial intelligence to support the automation of systems engineering tasks. The design of complex engineering systems requires an exhaustive and continuous verification and validation process at each stage of the mission lifecycle. While at the early phases of a project the system design might not yet include a high level of detail, it is still necessary to ensure that it meets the mission requirements. Even if the design data are recorded using MBSE tools, these do not usually offer integrated verification capabilities to cross check the design against the requirements. This paper explores Natural Language Processing techniques to address advancements, limitations, and customization of MBSE approaches in space. The application scenarios are mainly focused on learning from past mission data to identify potential links between requirements and model artefacts, and to detect inconsistencies in the models also making use of Large Language Models.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Implementation of an AI-powered digital assistant to support space system engineering activities

  • 期刊论文

  • 发表日期：

    2025年11月3日

查看全文
  • Paloma Maestro Redondo,
  • Gérald Garcia,
  • Annalisa Riccardi,
  • Paul Darm,
  • Jaime Bernar Jiménez Eguizabal,
  • Antoine Théate,
  • Sam Gerené,
  • Alberto González Fernández,
  • Gwendolyn Kolfschoten

摘要

This paper discusses innovative approaches to model-based systems engineering (MBSE) in the context of the space industry by presenting the results of a digital assistant that implements artificial intelligence to support the automation of systems engineering tasks. The design of complex engineering systems requires an exhaustive and continuous verification and validation process at each stage of the mission lifecycle. While at the early phases of a project the system design might not yet include a high level of detail, it is still necessary to ensure that it meets the mission requirements. Even if the design data are recorded using MBSE tools, these do not usually offer integrated verification capabilities to cross check the design against the requirements. This paper explores Natural Language Processing techniques to address advancements, limitations, and customization of MBSE approaches in space. The application scenarios are mainly focused on learning from past mission data to identify potential links between requirements and model artefacts, and to detect inconsistencies in the models also making use of Large Language Models.