Network Construction for Bearing Fault Diagnosis Based on Double Attention Mechanism

Table 2

Structural parameters of this model.

Layers	Network name	Key parameter	Output size
0	Input layer	—	1024  1
1	Convolutional block1	Number of neurons: 512, convolution kernel	512  1
2	Convolutional block12	Number of neurons: 128, convolution kernel	128  1
3	Spatial feature attention layer	—	128  1
4	LSTM1	Number of neurons: 32, dropout: 0.3	32 1
5	LSTM2	Number of neurons: 32, dropout: 0.3	32  1
6	LSTM3	Number of neurons: 32, dropout: 0.3	32  1
7	LSTM4	Number of neurons: 32, dropout: 0.3	32  1
8	Temporal feature attention layer	—	32  1
9	Softmax layer	—	10  1