Prediction on the Remaining Useful Life of Rolling Bearings Using Ensemble DLSTM

Table 3

Network parameter setting.

Name	Detail
Full connected layer (FC0)	in_features = 15, out_features = 128
LSTM layer	input_size = 128, hidden_size = 256, dropout = 0.2, num_layers = 2
Full connected layer (FC1)	in_features = 256, out_features = 128
Full connected layer (FC2)	in_features = 128, out_features = 64
Full connected layer (FC3)	in_features = 64, out_features = 32
Output layer (out)	in_features = 32, out_features = 1
Batch_size	8
Optimiser	ASGD
Loss function	MSE loss
Activation function	ReLU
Number of iterations	100