基于DRSN融合Transformer编码器的轴承故障诊断方法研究
Research on Bearing Fault Diagnosis Method Based on DRSN Fusion Trans-former Encoder
陈松 1陈文华 1张文广1
作者信息
- 1. 华北电力大学 控制与计算机工程学院,北京 102206
- 折叠
摘要
针对轴承故障在复杂工况环境中诊断准确率低和泛化性能弱的问题,提出了一种基于深度残差收缩网络(deep residual shrinkage network,DRSN)融合Transformer编码器的轴承故障诊断方法.首先,采用DRSN通过软阈值模块自动去掉振动信号中的噪声信息,并使用注意力机制增强提取到的特征;然后,采用Transformer编码器来进一步解决振动信号中的长期依赖性问题;最后,利用Softmax函数实现多故障模式识别.在凯斯西储大学轴承数据集上通过不同噪声等级对提出的模型进行测试,实验结果表明,该方法实现了对轴承故障分类,强噪声环境下准确率更高,训练时间更快.
Abstract
Aiming at the problems of low diagnosis accuracy and weak generalization performance of bearing faults in complex working conditions,a bearing fault diagnosis method based on deep residual shrinkage network(DRSN)fusion Transformer encoder is proposed.First,DRSN is used to automatically remove the noise information in the vibration signal through the soft threshold module,and the attention mechanism is used to enhance the extracted features.Then,the Transformer encoder is used to further solve the long-term dependence problem in the vibration signal,and finally the softmax function is used to realize multi-fault mode recognition.The proposed model is tested on the Case Western Reserve University(CWRU)dataset through different noise levels.The experimental results show that the method achieves classification of bearing faults,with higher accuracy in strong noise environments and fast training time.
关键词
故障诊断/轴承/深度残差收缩网络/Transformer编码器Key words
fault diagnosis/bearing/deep residual shrinkage network(DRSN)/Transformer encoder引用本文复制引用
出版年
2024
NETL
NSTL
维普
万方数据