Scholarlink

<p>Aiming at the problem of low prediction accuracy in the life prediction of rolling bearings, a method for predicting the remaining useful life of rolling bearings based on optimized Gated Recurrent Unit (GRU) and Transformer is proposed. The method first optimizes the GRU network using multi-head attention (MHA) to enhance the temporal information capturing ability of the GRU network, thereby achieving deep extraction of temporal degradation features. Then, it trims the Transformer framework to construct a positional feature learning unit, which learns and encodes the critical positional information of the data, thereby achieving deep extraction of positional degradation features. Finally, the deep degradation characteristics of timing and position are fused, and the remaining life of the rolling bearing can be predicted through a fully connected layer. Experimental verification using the PHM2012 and XJTU - SY rolling bearing datasets shows that the proposed method can effectively predict the remaining useful life of rolling bearings with high prediction accuracy.</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

Remaining Useful life Prediction of Rolling Bearings Based on Optimized GRU and Transformer

  • 期刊论文

  • 发表日期：

    2025年8月23日

查看全文
  • Jinbao Yao,
  • Jun Wang,
  • Xiang Li,
  • Keqin Zhou

摘要

Aiming at the problem of low prediction accuracy in the life prediction of rolling bearings, a method for predicting the remaining useful life of rolling bearings based on optimized Gated Recurrent Unit (GRU) and Transformer is proposed. The method first optimizes the GRU network using multi-head attention (MHA) to enhance the temporal information capturing ability of the GRU network, thereby achieving deep extraction of temporal degradation features. Then, it trims the Transformer framework to construct a positional feature learning unit, which learns and encodes the critical positional information of the data, thereby achieving deep extraction of positional degradation features. Finally, the deep degradation characteristics of timing and position are fused, and the remaining life of the rolling bearing can be predicted through a fully connected layer. Experimental verification using the PHM2012 and XJTU - SY rolling bearing datasets shows that the proposed method can effectively predict the remaining useful life of rolling bearings with high prediction accuracy.