Table 2 Hyperparameters setting used in LSTM network.

From: A hybrid LSTM random forest model with grey wolf optimization for enhanced detection of multiple bearing faults

Variables

Value

Number of features

13

LSTM layer hidden units

200

Training dataset

2600

Validation test

2600

Training epochs

1000

Gradient Threshold

2

Initial learn rate

0.01

Learn rate drop period

350

Learn rate drop factor

0.1

State and Gate activation function

tanh and sigmoid

Optimizer

sgdm

Back to article page