High-temperature augmented neighborhood metric learning for cross-domain fault diagnosis with imbalanced data

作者：

Highlights：

• A novel high-temperature neighborhood metric-learning network for imbalanced cross-domain fault diagnosis is proposed.

• Several fault diagnosis experiments were conducted on three vibration datasets to analyze the effectiveness of the proposed method.

• The effect of high-temperature mechanisms and pseudo labels on the network is discussed in depth.

摘要

•A novel high-temperature neighborhood metric-learning network for imbalanced cross-domain fault diagnosis is proposed.•Several fault diagnosis experiments were conducted on three vibration datasets to analyze the effectiveness of the proposed method.•The effect of high-temperature mechanisms and pseudo labels on the network is discussed in depth.

论文关键词：Fault diagnosis,Metric learning,Variable working conditions,Class imbalance

论文评审过程：Received 6 August 2022, Revised 9 September 2022, Accepted 19 September 2022, Available online 23 September 2022, Version of Record 6 October 2022.

论文官网地址：https://doi.org/10.1016/j.knosys.2022.109930