Scholarlink

<p>The precise prediction of the bearing's remaining useful life (RUL) is of paramount importance in ensuring equipment reliability within the mechanical industry. Currently, the realization of intelligent operational management and maintenance for mechanical systems utilizing deep learning (DL) is regarded as the foremost appealing research direction. Different from traditional DL techniques, the graph convolutional neural network (GCN) has become a promising technique for predicting the bearing RUL, due to its capability of handling non-Euclidean space and aggregating neighbor data. In the paper, the traditional RUL prediction methods are reviewed firstly. Subsequently, the principle of GCN is elaborated in detail, and classifications of its variants are presented. The taxonomy distinguishes spatial-GCN architectures from spectral-GCN frameworks. Finally, current GCN applications in bearing RUL prediction are reviewed, with emerging research challenges systematically outlined.</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

Graph Convolutional Neural Network Algorithms for Bearing Remaining Useful Life Prediction: A Review

  • 期刊论文

  • 发表日期：

    2025年6月17日

查看全文
  • Zhenzhen Jin,
  • Zhangwei Wu,
  • Jiayang Zhao,
  • Deqiang He,
  • Yuan Zhuang

摘要

The precise prediction of the bearing's remaining useful life (RUL) is of paramount importance in ensuring equipment reliability within the mechanical industry. Currently, the realization of intelligent operational management and maintenance for mechanical systems utilizing deep learning (DL) is regarded as the foremost appealing research direction. Different from traditional DL techniques, the graph convolutional neural network (GCN) has become a promising technique for predicting the bearing RUL, due to its capability of handling non-Euclidean space and aggregating neighbor data. In the paper, the traditional RUL prediction methods are reviewed firstly. Subsequently, the principle of GCN is elaborated in detail, and classifications of its variants are presented. The taxonomy distinguishes spatial-GCN architectures from spectral-GCN frameworks. Finally, current GCN applications in bearing RUL prediction are reviewed, with emerging research challenges systematically outlined.