基于领域自适应的故障诊断研究与进展

Research and Progress of Fault Diagnosis Based on Domain Adaptive

实际工业生产过程中,不同工况的机械设备,所采集到的数据普遍存在时空分布不一致或数据不平衡的问题,从而影响故障诊断模型的精度及泛化能力。迁移学习作为一种利用现有知识对不同但相关领域问题进行求解的学习方法,可有效解决传统机器学习数据分布不一致和小数据集下模型训练的问题。因此,如何将迁移学习应用于故障诊断领域成为学术界和工业界研究的热点。本文首先综述迁移学习领域自适应的研究现状,之后对领域自适应在故障诊断领域的应用进行分析,并对迁移学习在故障诊断领域未来的研究方向进行了探讨。

In the actual industrial production process,the data collected by mechanical equipment under different working conditions are generally inconsistent in spatial and temporal distribution or unbalanced in data,which affects the accuracy and generalization ability of fault diagnosis model.As a learning method that uses existing knowledge to solve problems in different but related fields,transfer learning can effectively solve the problem of inconsistent distribution of traditional machine learning data and model training under small data sets.Therefore,how to apply migration learning to fault diagnosis has become a hot topic in academia and industry.This paper first reviews the current research status of adaptive learning in the field of migration learning,then summarizes the application of adaptive learning in the field of fault diagnosis,and discusses the future research direction of adaptive learning in the field of fault diagnosis.