摘要
提出基于信息生成对抗网络和卷积神经网络的轴承小样本故障诊断方法。短时傅里叶变换从一维原始时域信号中提取时频图像特征;将提取的时频图像特征输入信息生成对抗网络进行对抗训练生成更多的图像样本;将生成的样本图像添加到训练的时频图像中,建立用于诊断的卷积神经网络模型。为了证明所提方法的有效性,在CWRU轴承数据集上进行了一系列对比试验。结果表明,所提的诊断方法优于其他算法和模型,能有效地实现轴承故障诊断;采用另一轴承故障数据集验证了该方法的泛化性。
Aiming at the problem of limited performance of bearing fault diagnosis due to single historical operating data and small amount of data,a small sample fault diagnosis method of bearing based on information maximizing generative adversarial nets(infoGAN)and convolutional neural network(CNN)was proposed.The short-time Fourier transform extracted the time-frequency image features from the one-dimensional original time-domain signal;The extracted time-frequency image features were input into infoGAN for confrontation training to generate more image samples;It will generate The sample images of are added to the trained time-frequency images,and the CNN model for diagnosis was established.In order to prove the effectiveness of the proposed method,a series of comparative experiments were carried out on the CWRU bearing data set.Experimental results show that the proposed diagnosis method is superior to other algorithms and models,and can effectively realize bearing fault diagnosis.Another bearing fault data set was used to verify the generalization of the method.
作者
杨青
陆见光
唐向红
顾鑫
盛晓静
杨瑞恒
YANG Qing;LU Jianguang;TANG Xianghong;GU Xin;SHENG Xiaojing;YANG Ruiheng(Key Laboratory of Advanced Manufacturing Technology, Ministry of Education, Guizhou University, Guiyang 550025, China;School of Mechanical Engineering, Guizhou University, Guiyang 550025, China;State Key Laboratory of Public Big Data, Guizhou University, Guiyang 550025, China)
出处
《兵器装备工程学报》
CSCD
北大核心
2021年第11期235-240,共6页
Journal of Ordnance Equipment Engineering
基金
贵州省公共大数据重点实验室开放基金项目(2017BDKFJJ019)
贵州省2018年本科教学内容和课程体系改革项目阶段性成果(2018520081)
贵州省科学技术基金项目(黔科合基础-ZK〔2021〕一般271)。
关键词
小样本
短时傅里叶变换
信息生成对抗网络
卷积神经网络
故障诊断
small sample
short-time Fourier transform
information maximizing generative adversarial nets
convolutional neural network
fault diagnosis
作者简介
杨青(1995—),女,硕士研究生,E-mail:1813507811@qq.com;通信作者:陆见光(1986—),男,博士,副教授,E-mail:jglu@gzu.edu.cn。
