Research Article
A Deep Transfer Learning Method for Bearing Fault Diagnosis Based on Domain Separation and Adversarial Learning
Table 5
Distribution of data sets in the transfer fault diagnosis experiments.
| | Transfer fault diagnosis experiments | Training data set | Test data set (30%) |
| | CWRU⟶Paderborn | CWRU (labeled 100%) | Paderborn | | Paderborn (unlabeled 70%) |
| | CWRU⟶XJTU-SY | CWRU (labeled 100%) | XJTU-SY | | XJTU-SY (unlabeled 70%) |
| | Paderborn⟶CWRU | Paderborn (labeled 100%) | CWRU | | CWRU (unlabeled 70%) |
| | Paderborn⟶XJTU-SY | Paderborn (labeled 100%) | XJTU-SY | | XJTU-SY (unlabeled 70%) |
| | XJTU-SY⟶CWRU | XJTU-SY (labeled 100%) | CWRU | | CWRU (unlabeled 70%) |
| | XJTU-SY⟶Paderborn | XJTU-SY (labeled 100%), | Paderborn | | Paderborn (unlabeled 70%) |
|
|
