Scholarlink

This paper introduces a novel approach to troubleshooting decision-making that leverages the Markov property of the problem. Specifically, the proposed method transforms the problem into a reinforcement learning task and utilizes the agent’s action selection process to facilitate troubleshooting. To further enhance the method’s performance, several optimization techniques are employed, such as setting prior fault probabilities based on failure rate, introducing a new troubleshooting action called “observe-act”, and implementing action masking. The efficacy of the proposed method is demonstrated through numerical experiments, which reveal its superiority over traditional methods.

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

Troubleshooting Decision-Making Method Using Reinforcement Learning Approach

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Qiang Zhang,
  • Limei Tian,
  • Baoding Liu,
  • Yingguang Wang,
  • Jiyang Zhang,
  • Ming Lu,
  • Yuewei Hu,
  • Weiheng Zhao,
  • Nuo Su

摘要

This paper introduces a novel approach to troubleshooting decision-making that leverages the Markov property of the problem. Specifically, the proposed method transforms the problem into a reinforcement learning task and utilizes the agent’s action selection process to facilitate troubleshooting. To further enhance the method’s performance, several optimization techniques are employed, such as setting prior fault probabilities based on failure rate, introducing a new troubleshooting action called “observe-act”, and implementing action masking. The efficacy of the proposed method is demonstrated through numerical experiments, which reveal its superiority over traditional methods.