Double-level adversarial domain adaptation network for intelligent fault diagnosis

作者：

Highlights：

• A double-level adversarial domain adaptation network is proposed to bridge the domain distribution differences for intelligent fault diagnosis.

• Domain-level and class-level alignments are jointly conducted by two minimax games.

• Wasserstein metric is adopted to construct a reliable discrepancy measure in class-level alignment.

• Extensive experiments on two mechanical equipment are constructed to verify the efficacy and superiority of the proposed method.

摘要

•A double-level adversarial domain adaptation network is proposed to bridge the domain distribution differences for intelligent fault diagnosis.•Domain-level and class-level alignments are jointly conducted by two minimax games.•Wasserstein metric is adopted to construct a reliable discrepancy measure in class-level alignment.•Extensive experiments on two mechanical equipment are constructed to verify the efficacy and superiority of the proposed method.

论文关键词：Domain adaptation,Intelligent diagnosis,Domain-level alignment,Class-level alignment,Machine

论文评审过程：Received 2 April 2020, Revised 22 May 2020, Accepted 8 July 2020, Available online 12 July 2020, Version of Record 17 July 2020.

论文官网地址：https://doi.org/10.1016/j.knosys.2020.106236