Research on Rolling Bearing Fault Diagnosis Based on Classification Generative Adversarial Network

In practical engineering, it is usually difficult to label monitoring data, and the fault diagnosis accuracy is not high in strong noise environment. To solve the above problems, this paper uses Cwt-CatGAN method to process signals and diagnose faults. In this method, CWT is first used to convert one-dimensional original vibration signals into two-dimensional time–frequency graphs as the input of classification generation admixture network. Then, the CatGAN model is obtained through the adversarial training process, and fake samples similar to the extraction distribution of CWT are generated. Finally, the input samples are clustered into certain categories. In addition, the performance of the Cwt-CatGAN method is verified by using the classical rotating machine data set, the experiment is compared with the unsupervised fault diagnosis model CatAAE.