摘要
轴承是旋转机械中的重要部件,在现实应用中,轴承的运行环境复杂多变。复杂工况条件下对轴承的故障进行准确判断是一个研究热点,因此针对复杂工况下轴承故障诊断中的样本不平衡和变工况下的迁移学习问题展开论述。在样本不平衡问题中,从重采样技术和基于模型的生成方法两种角度,分析所涉及方法的优缺点及适用场景。在迁移学习部分,详细解释了基于样本、特征和参数的迁移方法,并探讨了其在变工况轴承故障诊断中的应用前景。此外,还展望了未来可能涌现的新技术和方法,如结合深度学习和领域自适应的算法,以应对更复杂的工况和数据情境。旨在为轴承故障诊断领域的研究者提供参考,以进一步提升模型诊断的准确性和可靠性。
Bearing is an important part of rotating machinery.In practical applications,the operating environment of bearing is complex and changeable.It is a research hotspot to accurately judge the bearing fault under complex working conditions.Therefore,the sample imbalance in bearing fault diagnosis under complex working conditions and the transfer learning under variable working conditions are discussed.In the problem of sample imbalance,the advantages,disadvantages and applicable scenarios of the involved methods are analyzed from the perspectives of resampling technology and model-based generation method.In the part of transfer learning,the transfer method based on samples,features and parameters is explained in detail,and its application prospect in variable condition bearing fault diagnosis is discussed.In addition,it also looks forward to new technologies and methods that may emerge in the future,such as algorithms that combine deep learning and domain adaptation to deal with more complex working conditions and data scenarios.The purpose is to provide reference for researchers in the field of bearing fault diagnosis,so as to further improve the accuracy and reliability of model diagnosis.
作者
马新娜
张策
李豪
何畔
MA Xinna;ZHANG Ce;LI Hao;HE Pan(School of Information Science and Technology,Shijiazhuang Railway University;Hebei Key Laboratory of Electromagnetic Environmental Effects and Information Processing,Shijiazhuang Railway University;Shijiazhuang Key Laboratory of Artificial Intelligence,Shijiazhuang Railway University,Shijiazhuang 050043,China)
出处
《软件导刊》
2025年第9期9-18,共10页
Software Guide
基金
国家自然科学基金项目(12172234)
河北省自然科学基金项目(A2021210022)
河北省“三三三人才”资助项目(A202101018)。
关键词
复杂工况
轴承故障诊断
样本不平衡
迁移学习
complex working conditions
bearing fault diagnosis
sample imbalance
transfer learning
