Adaptive Fisher-Based Deep Convolutional Neural Network and Its Application to Recognition of Rolling Element Bearing Fault Patterns and Sizes

Table 2

Parameters of AFDCNN model during training.

Layer part parameter	The first layer	The second layer
	AFDCNN1	AFDCNN2-1	AFDCNN2-2	AFDCNN2-3
No. of layers	10	10	10	10
Learning rate	Adaptive	Adaptive	Adaptive	Adaptive
No. of epochs	600	200	200	200
Batch size	10	10	10	10
No. of kernels	[6, 12, 12]	[6, 12, 12]	[6, 12, 12]	[6, 12, 12]