Rotating machinery fault detection and diagnosis based on deep domain adaptation: A survey

• DOI: 10.1016/j.cja.2021.10.006
• Publication Date: 2021-11-09T16:09:43Z
• Authors: Siyu ZHANG; Lei SU; Jiefei GU; Ke LI; Lang ZHOU; Michael PECHT
• Keywords: Transfer of learning; Domain Adaptation

In practical mechanical fault detection and diagnosis, it is difficult expensive to collect enough large-scale supervised data train deep networks. Transfer learning can reuse the knowledge obtained from source task improve performance of target task, which performs well on small reduces demand for high computation power. However, significantly reduced by direct transfer due domain difference. Domain adaptation (DA) distribution information solve a series problems caused difference data. this survey, we review various current DA strategies combined with (DL) analyze principles, advantages, disadvantages each method. We also summarize application DL in field diagnosis. This paper provides summary research results proposes future work based analysis key technologies.