A Multi-Index Generative Adversarial Network for Tool Wear Detection with Imbalanced Data

Table 2

Multi-index decision-making algorithm.

---

Algorithm 2

Require: mean of -norm, CORT, fake-level score of normal signals, and fake-level score of

the generated signal, denoted as , , , and . Standard deviation of -norm,

CORT, fake-level score of normal signals, and fake-level score of the generated signal, denoted as

, , , and . Input signals , pretrained Generator , and pretrained

Generator .

(1) Invert to the latent space to search noises using Algorithm 1.

(2) Generate a batch of signals: .

(3) Calculate the -norm and CORT between and based on (4) and (5).

(4) Compute the fake-level scores of and , i.e., and , based on the discriminator

(5) Subtract thresholds:

(6)

(7)

(8)

(9)

(10) Scale the indexes:

(11)

(12)

(13)

(14)

(15) Combine the indexes:

(16)

(17) If:

(18) Input signals are normal.

(19) Else:

(20) Input signals are abnormal.

---