Scholarlink

<p>Under the strategic context of constructing new industrial systems, the rapid development of Industrial Internet services has significantly intensified interactions between public networks and industrial private communication networks. This trend has introduced emerging cybersecurity risks, highlighting an urgent need for research on effective risk assessment frameworks. First,we conduct cybersecurity risk analysis based on Industrial Internet characteristics to establish multidimensional static assessment metrics. Subsequently, leveraging large language models (LLMs) and knowledge graph technologies, we model cross-domain data flows (informational, physical, and cyber-physical) through attack path evolution mechanisms, thereby constructing an intelligent dynamic risk assessment framework. Finally, we develop quantitative calculation methods for these indicators and validate the system via experimental simulations. Results demonstrate that the proposed framework effectively evaluates Industrial Internet security risks, identifies potential vulnerabilities, and proposes targeted mitigation strategies. Validation in real-world industrial scenarios further confirms its operational efficiency, providing a scientific foundation for enhancing proactive risk prevention capabilities in cyber-physical systems.</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

Research on the construction method of industrial internet risk assessment indicator system based On LLMs

  • 期刊论文

  • 发表日期：

    2026年3月21日

查看全文
  • Ziyi Wang,
  • Zhenyu Guan,
  • Xu Liu,
  • Ben Qian,
  • Xuan Sun,
  • Jun Li,
  • Tingwen Liu

摘要

Under the strategic context of constructing new industrial systems, the rapid development of Industrial Internet services has significantly intensified interactions between public networks and industrial private communication networks. This trend has introduced emerging cybersecurity risks, highlighting an urgent need for research on effective risk assessment frameworks. First,we conduct cybersecurity risk analysis based on Industrial Internet characteristics to establish multidimensional static assessment metrics. Subsequently, leveraging large language models (LLMs) and knowledge graph technologies, we model cross-domain data flows (informational, physical, and cyber-physical) through attack path evolution mechanisms, thereby constructing an intelligent dynamic risk assessment framework. Finally, we develop quantitative calculation methods for these indicators and validate the system via experimental simulations. Results demonstrate that the proposed framework effectively evaluates Industrial Internet security risks, identifies potential vulnerabilities, and proposes targeted mitigation strategies. Validation in real-world industrial scenarios further confirms its operational efficiency, providing a scientific foundation for enhancing proactive risk prevention capabilities in cyber-physical systems.