摘要
针对谐波减速器在不同工况下源域与目标域故障数据分布存在显著差异,以及难以有效刻画故障信号的非线性和非平稳性,所导致诊断精度低的问题,提出了一种集中式信息融合结合深度迁移学习网络的谐波减速器故障诊断方法。首先,通过基于蜻蜓优化算法的变分模态分解提取复杂信号的模态分量,结合希尔伯特时频映射得到单轴向时频图;其次,通过小波变换将三轴向时频图进行集中信息融合,构造融合图像样本;然后,在残差网络的基础上融入卷积块注意力机制,并引入联合最大均值差异方法来衡量不同域之间的联合分布差异,构建域迁移深度网络,实现对变工况下谐波减速器迁移故障诊断;最后,通过搭建的谐波减速器试验平台进行试验验证。试验结果显示,所提方法在变工况诊断任务中,最高识别率可达98.75%,平均诊断结果为95.00%,可实现变工况谐波减速器的故障诊断。
The issue of low diagnostic accuracy is caused by the significant differences in fault data distributions between the source and target domains under different operating conditions,as well as the difficulty in effectively characterizing the nonlinearity and non-stationarity of the fault signals.A fault diagnosis method for harmonic gear reducers was proposed,combining centralized information fusion with deep transfer learning networks.Firstly,modal components of the complex signals were extracted using variational mode decomposition based on the dragonfly algorithm,and a uniaxial time-frequency image was obtained through the combination with Hilbert time-frequency mapping.Secondly,the three-axial time-frequency images were fused by wavelet transform to construct the fused image samples.Thirdly,on the basis of the residual network,the convolutional block attention module was integrated,and the joint maximum mean discrepancy method was introduced to measure the joint distribution difference between different domains,and the domain migration deep network was constructed to realize the migration fault diagnosis of harmonic reducer under variable working conditions.Finally,the experimental verification was carried out by the experimental platform of the harmonic reducer.In the variable working condition diagnosis task,the highest recognition rate of the proposed method can reach 98.75%,and the average diagnosis result is 95.00%,which can realize the fault diagnosis of the variable working condition harmonic reducer.
作者
石超
郭世杰
吕贺
SHI Chao;GUO Shijie;L He(School of Mechanical Engineering,Inner Mongolia University of Technology,Hohhot 010051,China;Inner Mongolia Key Laboratory of Robotics and Intelligent Equipment Technology,Hohhot 010051,China)
出处
《振动与冲击》
北大核心
2025年第16期137-150,共14页
Journal of Vibration and Shock
基金
国家自然科学基金资助项目(52365064)
内蒙古自然科学基金资助项目(2023LHMS05018)
内蒙古自治区高等学校青年科技英才支持计划资助项目(NJYT23043)。
关键词
谐波减速器
变工况
集中信息融合
域迁移
故障诊断
harmonic reducer
various working
centralized information fusion
domain migration
fault diagnosis
作者简介
第一作者:石超,男,硕士生,2000年生;通信作者:郭世杰,男,博士,副教授,1985年生,E-mail:sjguo@imut.edu.cn。
