基于振动灰度图的DCGAN结合ResNet50的发电机故障诊断OA

The fault diagnosis method based on deep learning can effectively realize the intelligent diagnosis of generator faults,but overfitting may occur easily when the training sample is too small or the sample distribution is not uniform,which affects the diagnosis effect and accuracy of the model.In order to solve these problems,this paper proposed a generator fault diagnosis method based on the deep convolutional generating adversal network(DCGAN)of vibrational grayscale image and residual network(ResNet50),which realizes the generator fault intelligent diagnosis under the condition of small fault samples and unbalanced fault samples.It was proved by experiment that the fault samples generated by DCGAN can be used to train the model in the case of small fault samples or unbalanced fault samples with higher diagnosis accuracy guaranteed.