太原理工大学学报2026，Vol.57Issue(2)：243-252,10.DOI:10.16355/j.tyut.1007-9432.20250088
基于改进深度强化学习的电力物联网自动渗透测试技术研究
Research on Automatic Penetration Testing Technology for Power Internet of Things Based on Improved Deep Reinforcement Learning
摘要
Abstract
[Purposes]With the comprehensive application of the ubiquitous power internet of things,uncontrollable factors and changes in the physical contact environment have made power inter-net of things systems face severe information security issues.Penetration testing,as an important means of information security protection,can discover and fix security vulnerabilities in advance,ef-fectively reducing the risk of system intrusion.Current penetration testing mainly relies on manual testing,which requires high technical expertise and experience from personnel and has low testing effi-ciency.To address these,an automatic penetration testing method based on improved deep reinforce-ment learning is proposed.[Methods]First,the state space was established for the learning processes according to expert experience and an attention mechanism was introduced to solve the problem of dy-namic changes in the state space.Then,a deep reinforcement learning model was used for automatic exploration of attack penetration paths.Finally,an experimental simulation environment was set up for comparative testing.[Results]The results show that the proposed method exhibits improved con-vergence performance across different networks.关键词
泛在电力物联网/渗透测试/深度强化学习/注意力机制/先验知识Key words
ubiquitous power internet of things/penetration testing/deep reinforcement learning/attention mechanism/prior knowledge分类
信息技术与安全科学引用本文复制引用
孙守道,卢毅,吴迪..基于改进深度强化学习的电力物联网自动渗透测试技术研究[J].太原理工大学学报,2026,57(2):243-252,10.基金项目
国网辽宁省电力有限公司管理科技项目资助(2024YF-87) （2024YF-87）
