DC-NNMN: Across Components Fault Diagnosis Based on Deep Few-Shot Learning

Table 1

Architecture of CNN.

Layer	Symbol	Operate	Parameter size
1	Input	Input samples	864
		Convolution	3 × 1 × 64
2	Conv1	Batch normalization	—
		Leaky ReLU	Leaky
3	Pool1	Max pooling	2 × 1
		Convolution	3 × 1 × 64
4	Conv2	Batch normalization	—
		Leaky ReLU	Leaky
5	Pool2	Max pooling	2 × 1
		Convolution	3 × 1 × 64
6	Conv3	Batch normalization	—
		Leaky ReLU	Leaky
		Convolution	3 × 1 × 64
7	Conv4	Batch normalization	—
		Leaky ReLU	Leaky
8	Output	Output features	216 × 1 × 64