文章摘要
田召阳,王风涛,熊元昊,王子豪,钱居楠.噪声干扰下基于多头注意力机制的圆锥滚子轴承故障诊断[J].井冈山大学自然版,2025,46(6):95-102
噪声干扰下基于多头注意力机制的圆锥滚子轴承故障诊断
Multi-attention mechanism based fault diagnosis of tapered roller bearings under noise interference
投稿时间:2025-03-11  修订日期:2025-04-21
DOI:10.3969/j.issn.1674-8085.2025.06.010
中文关键词: 轴承  故障诊断  多头注意力  噪声  神经网络
英文关键词: bearings  fault diagnosis  multi-attention  noise  neural network
基金项目:国家自然科学基金项目(51905001); 安徽未来技术研究院企业合作项目(2023qyhz22); 安徽工程大学校级项目(Xjky2022012)
作者单位E-mail
田召阳 安徽工程大学机械与汽车工程学院, 安徽, 芜湖 241000  
王风涛 安徽工程大学机械与汽车工程学院, 安徽, 芜湖 241000 wangfengt1985@163.com 
熊元昊 安徽工程大学机械与汽车工程学院, 安徽, 芜湖 241000  
王子豪 安徽工程大学机械与汽车工程学院, 安徽, 芜湖 241000  
钱居楠 安徽工程大学机械与汽车工程学院, 安徽, 芜湖 241000  
摘要点击次数: 39
全文下载次数: 40
中文摘要:
      针对噪声干扰下的圆锥滚子轴承故障诊断问题,本研究提出了一种基于多头注意力机制的长短时记忆网络(MHA-LSTM)方法,该方法通过从复杂运行环境中提取有效特征,实现对轴承故障状态的精准识别。具体流程为:首先通过采集圆锥滚子轴承运行过程中的振动信号,并构建噪声干扰模型模拟真实工况;随后将含噪信号直接输入MHA-LSTM模型进行特征提取和分类。实验结果表明,该方法能够有效识别噪声中的故障信号,显著提升故障诊断的准确性与可靠性。本研究为圆锥滚子轴承故障监测提供了新思路,对于保障机械设备的安全稳定运行具有重要工程价值。
英文摘要:
      In order to study the problem of fault diagnosis of tapered roller bearings under noise interference, a method based on multihead attention mechanism long short-term memory network(MHA-LSTM) is proposed.This method aims to extract the useful features from complex operating environments for the accurate identification of bearing fault conditions. By collecting vibration signals during the operation of tapered roller bearings, a noise interference model is constructed to simulate the most realistic working conditions. The noisy signals are directly used as inputs to the MHA-LSTM model for the feature extraction and classification. The experimental results show that this method can effectively identify the signals buried in the noise and improve the accuracy and reliability of fault diagnosis. This research provides new insights and methods for fault monitoring of tapered roller bearings, which is of great significance for ensuring the safe and stable operation of mechanical equipment.
查看全文   查看/发表评论  下载PDF阅读器
关闭