Table 1 Parameter Settings for PSO-BP Neural Network.

From: Prediction of frequency response of sub-frame bushing and study of high-order fractional derivative viscoelastic model

Algorithm

Parameters

Values

BP neural network

Input neuron

1

Hidden layer

5

Output neuron

1

Training samples

1–40 Hz

Test samples

41–50 Hz

Prediction

51–70HZ

Learning rate

0.1

PSO algorithm

Particle dimension

16

Population size

30

Number of iterations

50

\([{v}_{min} , {v}_{max}]\)

[− 1,1]

\([{x}_{min} , {x}_{max}]\)

[− 5,5]

\({c}_{1}\)

1.5

\({c}_{2}\)

1.5

  1. The particle dimension refers to the sum of the threshold and weight count of the entire neural network. The weight count is calculated as follows: 1 × 5 + 5 × 1 = 10, and the threshold count is 5 + 1 = 6. Therefore, the particle dimension is 10 + 6 = 16.
Back to article page