Table 3 LSTM Architecture.

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

Layers

Name

Number of parameters

 

Input

0

1

LSTM Activation (tanh)

376,800

2

Dropout (0.5)

0

3

LSTM Activation (tanh)

721,200

4

Fully connected (128 units)

38,528

5

Dropout (0.5)

0

6

Fully connected

774

 

Output

Activation (SoftMax)

0

  1. The GWO parameters are set as follows:
  2. Maximum number of iterations 500.
  3. Number of wolves: 5, 10, 15, 20, 25, 30.
  4. c1 = c2 = 0.5, c3 = 0.5.
  5. w = 0.5 + rand ()/2, and l ∈ [1,0].
Back to article page