Intelligent Fault Diagnosis of Aeroengine Sensors Using Improved Pattern Gradient Spectrum Entropy

Table 3

Network parameters.

Name	Parameter	Output feature size
Input layer	Input data
Convolutional layer (C1)	6 filters, size , stride 1
Pooling layer (P1)	Filter size , stride 1
Convolutional layer (C2)	16 filters, size , stride 1
Pooling layer (P2)	Filter size , stride 1
Fully connected layer (F1)	120 nodes,
Fully connected layer (F2)	84 nodes,
Fully connected layer (F3)	7 nodes
Output layer	Output data
Learning rate	0.00004	None
Training dataset	2450	None
Test dataset	1050	None
Iteration times	35000	None
Scale factor	25	None