基于深度卷积神经网络的滚动轴承迁移故障诊断  

Transfer Fault Diagnosis of Rolling Bearings Based on Deep Convolutional Neural Network

作　　者：李欢 吕勇 袁锐[1,2] 杨旭 LI Huan;LV Yong;YUAN Rui;YANG Xu(Key Laboratory of Metallurgical Equipment and Control Technology,Ministry of Education,School of Mechanical Automation,Wuhan University of Science and Technology,Wuhan 430081,China;Hubei Key Laboratory of Mechanical Transmission and Manufacturing Engineering,School of Mechanical Automation,Wuhan University of Science and Technology,Wuhan 430081,China)

机构地区：[1]武汉科技大学机械自动化学院冶金装备及其控制教育部重点实验室,武汉430081 [2]武汉科技大学机械自动化学院机械传动与制造工程湖北省重点实验室,武汉430081

出　　处：《组合机床与自动化加工技术》2023年第2期90-94,共5页Modular Machine Tool & Automatic Manufacturing Technique

基　　金：国家自然科学基金面上项目(51875416);湖北省自然科学基金创新群体项(2020CFA033);中国博士后科学基金面上项目(2020M682492)。

摘　　要：针对滚动轴承故障诊断中故障样本不足、诊断精度与诊断效率不高的问题,提出一种基于深度卷积神经网络的滚动轴承迁移故障诊断方法。首先,通过VMD对原始振动信号进行分解,利用中心频率法确定分解个数k;其次,按照最大峭度准则筛选出最佳固有模态函数(intrinsic mode function, IMF),并对其进行连续小波变换(continuous wavelet transform, CWT)生成时频图;最后,将预处理过的时频图输入到在ImageNet数据集预训练过的深度残差网络(residual network, ResNet)模型中微调,实现故障分类识别。在某大学公开轴承数据集和题课组数据集上验证,测试精度分别达到99.60%和100%,可有效实现滚动轴承故障诊断。Aiming at the problems of insufficient fault samples and low diagnosis accuracy and efficiency in rolling bearing fault diagnosis, a rolling bearing transfer fault diagnosis method based on deep convolution neural network was proposed.Firstly, the original vibration signal is decomposed by VMD,and the number of decomposition k is determined by the center frequency method.Secondly, the optimal intrinsic modal function(IMF) is screened out according to the maximum kurtosis criterion, and the continuous wavelet transform(CWT) is used to generate the time-frequency map.Finally, the preprocessed time-frequency map is input into a deep residual network(ResNet) model pre-trained on the ImageNet dataset for fine-tuning to achieve fault classification and identification.It is verified on the dataset of the research group and the public bearing dataset of an university, and the test accuracy has reached 99.60% and 100%,which can effectively realize the fault diagnosis of rolling bearings.

关 键 词：滚动轴承 深度卷积神经网络 变分模式分解 深度迁移学习 故障诊断 

分 类 号：TH133.3[机械工程—机械制造及自动化] TG659[金属学及工艺—金属切削加工及机床]