Table 1.

The network structure of fault diagnosis classification model based on MobileNet.

Layer	Input Shape
Convolution/2	(32, 32, 1)
Depth-wise convolution/1	(16, 16, 8)
Point-wise convolution/1	(16, 16, 16)
Depth-wise convolution/2	(8, 8, 16)
Point-wise convolution/2	(8, 8, 32)
Depth-wise convolution/1	(8, 8, 32)
Point-wise convolution/1	(8, 8, 64)
Depth-wise convolution/2	(4, 4, 64)
Point-wise convolution/2	(4, 4, 128)
Average pooling/1	(1, 1, 128)
Fully connected	(1, 1, 128)
Softmax	(1, 1, 5)