Scholarlink

<p>Mechanical equipment generates a significant amount of operational data, leading to prolonged training times for fault diagnosis models. Therefore, it is necessary to reduce the scale of the data. This paper addresses this issue by reducing the size of the data from two perspectives. A PageRank-guided cluster sampling (PGCS) method is proposed to decrease the number of data items. Then, a Pearson-Shannon feature selection (PSFS) method is designed to reduce the dimensionality of the data. The experimental results show that while maintaining the average performance loss of the model within 10 %, the average training time of the model is reduced by more than 90 %, indicating the effectiveness of the proposed method. Furthermore, comparisons with five sampling method and five feature selection methods show that PGCS and PSFS outperforms these methods. Additionally, experiments using different combinations of three rotating machine datasets and six commonly used models indicate that the proposed method has universal applicability to some extent.</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

An efficient fault diagnosis method for rotating machinery based on PageRank-guided cluster sampling and Pearson-Shannon feature selection

  • 期刊论文

  • 发表日期：

    2025年10月8日

查看全文
  • Cheng Peng,
  • Haizhen Huang,
  • Xianming Huang,
  • Zhaohui Tang

摘要

Mechanical equipment generates a significant amount of operational data, leading to prolonged training times for fault diagnosis models. Therefore, it is necessary to reduce the scale of the data. This paper addresses this issue by reducing the size of the data from two perspectives. A PageRank-guided cluster sampling (PGCS) method is proposed to decrease the number of data items. Then, a Pearson-Shannon feature selection (PSFS) method is designed to reduce the dimensionality of the data. The experimental results show that while maintaining the average performance loss of the model within 10 %, the average training time of the model is reduced by more than 90 %, indicating the effectiveness of the proposed method. Furthermore, comparisons with five sampling method and five feature selection methods show that PGCS and PSFS outperforms these methods. Additionally, experiments using different combinations of three rotating machine datasets and six commonly used models indicate that the proposed method has universal applicability to some extent.