Algorithm 1. The training process of DamperGAN.

Input: Real damper image dataset $I_o=\left\{I_{o_1},\dots ,I_{o_N}\right\}$; Generator $G$; Discriminator     ${\left\{D_i\right\}}_{i=1}^{i=k}$; g-steps, the training step of the generator; d-steps, the training step of     the discriminators.

Output: $G$, generator after training.

1: Initialize generator $G$ and discriminator ${\left\{D_i\right\}}_{i=1}^{i=k}$ with random weights;

2: repeat

3:  for g-steps, perform

4:     $G$ generate fake images;

5:     Calculate the penalty value $V_D^G$ via Equation (6);

6:     Minimize Equation (9) to update the parameters of the generator $G$;

7:  end for

8:  for d-steps, perform

9:      Use $G$ to generate fake images $I_g=\left\{I_{g_1},\dots ,I_{g_N}\right\}$;

10:    Use real images $I_o=\left\{I_{o_1},\dots ,I_{o_N}\right\}$ and fake images $I_g=\left\{I_{g_1},\dots ,I_{g_N}\right\}$ to update the discriminator parameters by minimizing Equation (10);

11:  end for

12: until DamperGAN completes convergence

13: return