Rolling Bearing Fault Diagnosis Based on STFT-Deep Learning and Sound Signals

Table 3

Settings of SAE.

Layer 1	The number of input layer nodes	6500
	The number of hidden layer nodes	1000
	Sparsity	0.1
	Sparsity penalty factor of the loss function	3
	Weight decay factor of the loss function	0.003
Layer 2	The number of input layer nodes	1000
	The number of hidden layer nodes	100
	Sparsity	0.1
	Sparsity penalty factor of the loss function	3
	Weight decay factor of the loss function	0.003
Layer 3	The number of input layer nodes	100
	The number of hidden layer nodes	10
	Sparsity	0.1
	Sparsity penalty factor of the loss function	3
	Weight decay factor of the loss function	0.003
Softmax classifier	The number of input layer nodes	10
	The number of output layer nodes	4