基于一维卷积神经网络的轴承故障诊断方法研究

Research on bearing fault diagnosis based on one-dimensional convolutional neural network

轴承故障诊断为机械故障诊断领域的研究热点,传统轴承故障诊断方法需要在特征选择的基础上实现信号特征提取,这类方法诊断效率低,且诊断效果依赖于特征参数的选择。因此,为了提高滚动轴承故障诊断效率和准确度,考虑到一维卷积神经网络具有端到端的特点,文章提出一种基于一维卷积神经网络(1DCNN)的滚动轴承故障诊断方法。利用CWRU轴承数据集对方法的性能进行验证,实验结果表明在相同负载条件下的故障诊断准确度为99.59%以上,且该方法在变负载条件下具有强的诊断鲁棒性。

Bearing fault diagnosis is a research hot spot in the field of mechanical fault diagnosis.The traditional bearing fault diagnosis method need to extract bearing fault signals on the basis of feature selection.So the method is inefficient,and the diagnostic results depend on the selection of characteristic parameters.In order to improve the efficiency and accuracy of rolling bearing fault diagnosis,according to the end-to-end characteristics of one-dimensional convolutional neural network,a fault diagnosis rolling bearing method based on one-dimensional convolutional neural network is proposed.The performance of the fault diagnosis method is verified by the CWRU bearing fault data set.The experimental results show that the recognition rate of fault diagnosis is more than 99.59%under the same load conditions,and it has strong adaptability under variable load conditions.