Skip to content

Leolando/FYP

Repository files navigation

FYP-Unsupervised Domain Adaptation for bearing fault diagnosis

XJTU Dataset Selection

  • Only choose stable data. Data with significant shakes will be dropped.
Dataset Element Good (Normal) Starts at Fault Starts at
Bearing 2_1 37.5 Inner 1-452 454-484
Bearing 2_2 Outer 1-50 51-159
Bearing 2_4 Outer 1-30 31-40
Bearing 2_5 Outer 1-120 121-337
Bearing 3_1 40Hz Outer 1-2463 2464-2536
Bearing 3_3 Inner 1-340 341-369
Bearing 3_4 Inner 1-1416 1417-1514
Bearing 3_5 Outer 1-10 11-110

Methodology

Purposed two different types of Unsupervised Domain Adaptation in adversarial manners, and applied pseudo label semi-supervised learning strategy. The two different types of models are compared and analyzed.

Proposed structure of feature extractor

image-20220427173456789

First type of Unsupervised Domain Adaptation model

image-20220427173803408

Experiment results

Confusion matrix of the first proposed model

image-20220427174004039

Loss and accuracy of the second proposed model

image-20220427174035953

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages