基于WGAN和CNN的轴承故障诊断研究

Bearing Fault Diagnosis Based on WGAN and CNN

轴承作为机械系统的关键部件,可以解析出整个机械系统的故障信息和健康状态,提出一种基于改进生成对抗网络和卷积神经网络的轴承故障诊断方法.首先通过小波变换将一维轴承振动数据转换为二维时频图像数据;然后经设计的改进生成对抗网络训练轴承的二维图像数据,将达到纳什平衡后生成的数据补充到原始数据中增加轴承样本数据;最后将扩充完成的数据集输入卷积神经网络进行训练.测试结果显示轴承故障诊断平均准确率达94%,验证了故障数据不充足时该方法用于轴承故障诊断的可行性.

As the key component of mechanical system,rolling bearings can be used to analyze the fault information and health status of the whole mechanical system.It is of practical significance to study bearing fault diagnosis in industrial process.In this paper,a method of bearing fault diagnosis based on the improved generative adversarial network,Wasserstein GAN(WGAN)and convolution neural network(CNN)is proposed.Firstly,one-dimensional bearing vibration data is converted into two-dimensional time-frequency image data through wavelet transform.Then,the 2D image data of bearings trained by the designed WGAN network are supplemented with the data generated after reaching Nash equilibrium to add bearing sample data to the original data.Finally,the expanded data set is input into the CNN network for training.And the test results show that the average accuracy of bearing fault diagnosis is 94%,which confirms the feasibility of the proposed method for bearing diagnosis when the fault data is insufficient.