摘要
提出一种基于二维图像和轻量级深度卷积神经网络相结合的集合型电机轴承故障诊断方法,用于提高电机轴承故障诊断性能。通过小波包变换(WPT)电机轴承振动信号重构为二维小波包图像。在ResNet基础上,将网络结构中的卷积操作改进为更轻量级的深度可分离卷积(DSConv)结构,在不损失诊断精度的前提下缩短了训练时间。仿真结果验证了所提基于图像的轻量级故障诊断方法用于故障诊断的有效性。
In this paper,a set motor bearing fault diagnosis method based on the combination of two-dimensional image and lightweight deep convolutional neural network is proposed to improve the motor bearing fault diagnosis performance.The vibration signal of motor bearing is reconstructed by wavelet packet transform(WPT).On the basis of ResNet,the convolution operation in the network structure was improved into a more lightweight Depthwise Sparable Convolution Residual Network(DS-ResNet)structure,which shortened the training time without loss of diagnostic accuracy.Simulation results show that the proposed WPT-DS-ResNet method is effective for fault diagnosis.
作者
付丽君
赵晨兵
杨青
张齐鹏
FU Lijun;ZHAO Chenbing;YANG Qing;ZHANG Qipeng(Shenyang Ligong University,Shenyang 110159,China)
出处
《沈阳理工大学学报》
CAS
2020年第5期8-12,共5页
Journal of Shenyang Ligong University
基金
辽宁省教育厅科学研究项目计划(LG201917)
辽宁省自然科学基金指导计划(20180550801)。