Research on Fault Diagnosis Model of Generative Adss Based on Improved Semisupervised Diagnosis Algorithm

Table 1

Improved SGAN algorithm steps.

---

Algorithm: Improved SGAN model training based on small batch random gradient descent algorithm

Inputs: p (z): Random noise distribution, p (x): True data distribution, λ: Specific gravity coefficient, m: Iteration times

Output: Trained discriminator D

For i = 1 to m do

 N Noise samples {} subject to P (z) distribution

 N labeled samples obeying p (x) distribution {( ), ( )....(, )}

 N unlabeled samples following p (x) distribution {}

 Keeping the parameters of generator G unchanged, update the parameters of discriminator D according to the following formula

 Keeping the parameters of discriminator D unchanged, update the parameters of generator G according to the following formula

 end

---