Evaluation and Prediction Method of Rolling Bearing Performance Degradation Based on Attention-LSTM

Table 5

Attention-LSTM prediction model parameters.

Parameter name	Parameter value
Number of attention-layer nodes	4
Number of input layer nodes	4
Number of output layer nodes	4
Proportion of training sequence	67%
Proportion of test sequence	33%
Number of middle layers	4
Number of middle layer nodes	6
Activation function	ELU
Loss function	MSE
Optimizer	Adam
Dropout layers	4
Learning rate	0.001