基于深度残差网络的滚动轴承故障诊断  

Rolling Bearing Fault Diagnosis Based on Deep Residual Network

作　　者：龙腾宇 Long Tengyu(School of Mechanical Engineering and Automation,Wuhan Textile University,Wuhan,China)

机构地区：[1]武汉纺织大学机械工程与自动化学院,湖北武汉

出　　处：《科学技术创新》2024年第19期21-24,共4页Scientific and Technological Innovation

摘　　要：针对基于传统深度学习的滚动轴承故障诊断方法在变工况条件下,存在特征值提取困难、泛化性差的问题,本文提出了一种基于连续小波变换和深度残差网络的滚动轴承故障诊断模型。将采集到的原始振动信号进行连续小波变换(CWT),转换为二维时频图;利用分组卷积(Group Convolution)具有降低模型复杂程度的优点,用其代替传统残差网络的标准卷积层;通过加入Dropout层和使用全局平均池化层以避免过拟合,从而提高模型的抗干扰能力。实验结果证明,该模型识别准确率达到了98.33%;变工况条件下的滚动轴承故障诊断准确率超过90%,相较于现主流故障诊断方式具有更高的准确率和更好的泛化性。In view of the difficulties in extracting eigenvalues and poor generalization of common deep learning-based rolling bearing fault diagnosis methods under varying working conditions,a rolling bearing fault diagnosis model based on continuous wavelet transform and deep residual network was proposed.The collected original vibration signal is transformed into a two-dimensional time-frequency graph by continuous wavelet transform(CWT).Group Convolution has the advantage of reducing the complexity of models,and it replaces the standard convolution layer of traditional residual networks.The anti-jamming capability of the model is improved by adding the Dropout layer and using the global average pooling layer to avoid overfitting.The experimental results show that the recognition accuracy of this model reaches 98.33%.The fault diagnosis accuracy of rolling bearings under variable working conditions is more than 90%.

关 键 词：故障诊断 滚动轴承 连续小波变换 深度残差网络 

分 类 号：TP391.5[自动化与计算机技术—计算机应用技术] TH133.3[自动化与计算机技术—计算机科学与技术]