We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
基于DDAE-GRA的滚动轴承早期故障诊断.
- Authors
廢冰姻; 回波; 邓振明; 史珂
- Abstract
A method for early fault diagnosis of rolling bearings is proposed by combining deep denoising auto-encoder (DDAE) with grey relation analysis (GRA). The DDAE is used to extract features from vibration signals of the bearings. The features of normal samples are utilized as a benchmark sequence for GRA. The relation degree between sample features of the bearings in full life cycle and normal sample features is calculated as an indicator for performance degradation of the bearings. The performance degradation curve is drawn, and the early failure time of the bearings is ascertained using a 3o threshold. The analysis results of the bearings exhibiting outer ring faults in bearing dataset from University of Cincinnati indicate that; the bearing faults identified by DDAE-GRA model occurs in 533rd sample, the envelope spectrum of the sample exhibits obvious feature frequencies associated with outer ring faults, as well as their frequency doubling; the DDAE - GRA model demonstrates better robustness than wavelet packet decomposition - GRA, DDAE-FCM, and DDAE-SVDD models, and is more suitable for early fault detection of the bearings.
- Publication
Bearing, 2023, Issue 11, p76
- ISSN
1000-3762
- Publication type
Article
- DOI
10.19533/j.issn1000-3762.2023.11.015