Application of Generative Adversarial Nets in Bearing Fault Diagnosis
Aiming at the problems that conditional generative adversarial nets(CGAN)can only judge truth or false and not judge multiple classification,and semi-supervised generative adversarial nets(SGAN)needs to discriminate multiple classification and judge truth and false simultaneously,an improved generative adversarial nets called conditional semi-supervised generative adversarial nets(CSGAN)is proposed in this paper,and its specific design is also given.The generator of the CSGAN is based on CGAN and composed of multi-layer perceptron(MLP),and the discriminator of the CSGAN is based on SGAN and consists of convolutional neural networks(CNN).Based on CSGAN,a 2-D GAN method for bearing fault diagnosis is proposed.Firstly,the original fault signals are normalized to the interval[-1,1],and then a sliding window is used to intercept 1024 length data from the normalized data,which is converted into a 2-D matrix with a size of 32×32 as the input of CSGAN.The validation of experiments on several public data sets shows that this method can effectively improve the diagnostic accuracy of the discriminator under different sample proportions and has good applicability.
国家科技期刊平台
NSTL
万方数据
维普
