改进注意力机制的滚动轴承故障诊断方法研究  

Research on Fault Diagnosis Method of Rolling Bearing with Improved Attention Mechanism

作　　者：肖安 李开宇[1] 范佳能 仲志强 贾银亮[1] XIAO An;LI Kaiyu;FAN Jianeng;ZHONG Zhiqiang;JIA Yinliang(College of Automation Engineering,Nanjing University of Aeronautics and Astronautics,Nanjing 211100,China;NARI Group Corporation,Nanjing 211106,China)

机构地区：[1]南京航空航天大学自动化学院,南京211100 [2]南瑞集团有限公司,南京211106

出　　处：《计算机测量与控制》2023年第11期22-30,共9页Computer Measurement &Control

摘　　要：针对滚动轴承在实际工作环境中噪声较大和负载变化的问题,提出一种基于双注意卷积机制的残差神经网络(DACM_ResNet,double attention convolution mechanism ResNet)轴承故障诊断方法;首先,对滚动轴承振动信号进行短时傅里叶变换(STFT,short-time fourier transform)并使用伪彩色处理得到三通道图像数据;然后,对残差神经网络在轴承故障诊断上进行研究,在残差单元的卷积层之后,使用DACM模块,将残差特征在通道和空间维度上进行进一步提取,最后,在凯斯西储大学(CWRU)数据集上进行试验验证,试验结果表明所提出的方法在噪声环境下及负载变化时,平均诊断准确率达到了98%以上,说明所提出的模型有较好的鲁棒性。Aiming at the problems of high noise and load changes in actual working environments of rolling bearings,a bearing fault diagnosis method based on double attention convolution mechanism(DACM)_ResNet is proposed.Firstly,a short-time Fourier transform(STFT)is performed on the vibration signal of rolling bearings,and a pseudo-color processing is used to obtain three-channel image data.Then,the residual neural network is studied on the bearing fault diagnosis.After the convolution layer of the residual block,the DACM module is used to further extract the residual features in the channel and spatial dimensions,and the connection between the residual and input is established.Finally,the experiments are verified on the dataset in Case Western Reserve University(CWRU),and the test results show that the proposed method has an average accuracy of over 98%under the noise environment and load changes,indicating that the proposed model has a good robustness of noise.

关 键 词：轴承故障诊断 短时傅里叶变换 伪彩色处理 双注意卷积机制模块 残差网络 

分 类 号：TH133.3[机械工程—机械制造及自动化] TG66[金属学及工艺—金属切削加工及机床]