基于改进深度残差收缩网络的轴承故障诊断  

Bearing Fault Diagnosis Based on Improved Depth Residual Shrinkage Network

作　　者：李雪松 李劲华 吕智涵 LI Xue-song;LI Jin-hua;LÜ Zhi-han(School of Data Science and Software Engineering, Qingdao University, Qingdao 266071, China)

机构地区：[1]青岛大学数据科学与软件工程学院,青岛266071

出　　处：《青岛大学学报（自然科学版）》2022年第2期38-43,50,共7页Journal of Qingdao University(Natural Science Edition)

基　　金：国家自然科学基金(批准号:61902203)资助;山东省重点研发计划重大科技创新工程(批准号:2019JZZY020101)资助。

摘　　要：为解决噪声背景中轴承故障诊断精度不高的问题,提出了一种新的轴承故障诊断方法。利用连续小波变换将采集到的振动信号转换成小波时频图,采用多尺度膨胀卷积对深度残差收缩网络进行改进,扩大卷积核的感受野,并将交叉熵损失函数改进成加权交叉熵损失函数。实验结果表明,与其他深度学习算法相比,本算法故障诊断的准确率较高。The accuracy of bearing fault diagnosis is not high under the background of noise.In order to solve this problem,a new bearing fault diagnosis method was proposed.Firstly,it used continuous wavelet transform to convert the collected vibration signals into wavelet time-frequency graph.Then the deep residual shrinkage network is improved by multi-scale dilation convolution,and the receptive field of the convolution kernel was expanded.Finally,The cross entropy loss function is improved to a weighted cross entropy loss function.The experimental results show that compared with other deep learning algorithms,the fault diagnosis accuracy of this algonthm is higher.

关 键 词：轴承 故障诊断 深度残差收缩网络 小波时频图 多尺度膨胀卷积 

分 类 号：TH-3[机械工程]