摘要
由于无法采集到齿轮和轴承的某样故障类型的充足故障样本,使其呈现长尾分布形式,导致无法有效构建神经网络诊断模型;当引入联邦学习方法解决上述长尾问题时,无法有效地提取尾部故障类型样本的特征信息。针对上述问题,本文提出一种改进联邦学习方法。首先,采用联邦特征对诊断模型再训练,提高对尾部样本的故障特征提取能力;其次,引入CBAM注意力机制,对联邦学习中的ResNet网络模型进行改进,增强对通道和空间的关键局部特征信息的提取能力和效率;再次,将传统卷积替换为非对称卷积,增强对样本的非对称特征信息的提取能力和效率;最后,采用间隔校准算法优化网络模型的分类边距,以获取更高的诊断准确率和效率。基于齿轮和轴承的实测故障样本的实验分析可知,所提改进联邦学习方法可有效地提升平均和最高准确率,二者分别为8.78%和3.40%。
Due to the inability to collect sufficient fault samples of a certain fault type of gears and bearings failures,the data exhibits a long tail distribution,making it impossible to effectively construct a neural network diagnosis model.When the federal learning method is introduced to solve the above long tail problem,the feature information of the tail fault type sample cannot be effectively extracted.In view of the problems,this paper proposes an improved federated learning method.Firstly,the diagnosis model is retrained by using federal features to improve the fault feature extraction ability of tail samples.Secondly,the CBAM(convolutional block attention module)attention mechanism is introduced to improve the ResNet(residual network)network model in federated learning,boosting its ability and efficiency of extracting key local feature information of channel and space.Thirdly,the traditional convolution is replaced by asymmetric convolution to enhance the ability and efficiency of extracting asymmetric feature information of samples.Finally,the interval calibration algorithm is used to optimize the classification margin of the network model to obtain higher diagnostic accuracy and efficiency.The experimental analysis based on the measured fault samples of gears and bearings shows that the proposed improved federated learning method can effectively improve the average and highest accuracy,by 8.78%and 3.40%,respectively.Due to the inability to collect sufficient fault samples of a certain fault type of gears and bearings,it presents a long tail distribution form,which makes it impossible to effectively construct a neural network diagnosis model;when the federal learning method is introduced to solve the above long tail problem,the feature information of the tail fault type sample cannot be effectively extracted.In view of the above problems,this paper proposes an improved federated learning method.Firstly,the diagnosis model is retrained by using federal features to improve the fault feature extraction ability of tail samples.Secondly,the CBAM(convolutional block attention module)attention mechanism is introduced to improve the ResNet(residual network)network model in federated learning,and enhance the ability and efficiency of extracting key local feature information of channel and space.Thirdly,the traditional convolution is replaced by asymmetric convolution to enhance the ability and efficiency of extracting asymmetric feature information of samples.Finally,the interval calibration algorithm is used to optimize the classification margin of the network model to obtain higher diagnostic accuracy and efficiency.The experimental analysis based on the measured fault samples of gears and bearings shows that the proposed improved federated learning method can effectively improve the average and highest accuracy,which are 8.78%and 3.40%respectively.
作者
刘伟民
展翼鹤
郑爱云
黄继德
郑直
Liu Weimin;Zhan Yihe;Zheng Aiyun;Huang Jide;Zheng Zhi(College of Mechanical Engineering,University of Science and Technology,Tangshan 063210,China;HBIS Group Co,Tangshan 063600,China)
出处
《仪器仪表学报》
EI
CAS
CSCD
北大核心
2024年第9期145-156,共12页
Chinese Journal of Scientific Instrument
基金
河北省自然科学基金项目(E2022209086)
河北省科技重大专项项目(22282203Z)资助。
关键词
联邦学习
长尾数据
注意力机制
非对称卷积
间隔校准算法
federated learning
long-tail data
attention mechanism
asymmetric convolution
margin calibration algorithm
作者简介
刘伟民,1996年于东北大学获得学士学位,2009年于北京科技大学获得博士学位,现为华北理工大学副教授,主要研究方向为制造系统优化、工业物联网技术等。E-mail:lzhjia@ncst.edu.cn;通信作者:郑爱云,1996年于东北大学获工学学士学位,1999年于东北大学获得硕士学位,现为华北理工大学副教授,主要研究方向为制造系统优化、工业物联网技术等。E-mail:zay@ncst.edu.cn;黄继德,于北京科技大学获得学士学位,现为唐山钢铁集团有限责任公司正高级工程师,主要研究方向为设备管理、智能运维等。E-mail:13933398653@163.com。
