首页|基于边界辅助判别的滚动轴承故障特征增强及诊断方法

基于边界辅助判别的滚动轴承故障特征增强及诊断方法

扫码查看
原文链接
  • NETL
  • NSTL
  • 万方数据
滚动轴承作为机械设备重要部件,对保障设备安全稳定运行具有重要意义.针对实际诊断中的滚动轴承故障数据不平衡问题,提出了一种基于边界辅助判别的辅助分类生成对抗网络模型(BD-ACGAN).首先,设计了一种可用于提取故障样本边界细节特征的边界辅助判别器,以引导生成器生成更真实的样本,并采用该生成样本解决了数据不平衡的问题;其次,采用了自适应权重损失模块,动态调整了损失权重,使该模型更加关注重要的特征信息,从而提高了该模型的生成质量和特征表达能力;利用生成样本和真实样本数据对BD-ACGAN模型进行了增强训练,提高了该模型的泛化能力和诊断能力;最后,进行了消融实验及对照实验,对BD-ACGAN模型的特征增强能力和诊断效果进行了验证,分别采用美国凯斯西储大学和西安交通大学滚动轴承数据集对模型进行了实验验证.研究结果表明:该BD-ACGAN模型能够有效利用故障样本的边界特征解决数据不平衡问题,并且故障诊断精确度为98.79%,优于其他对照模型,为滚动轴承故障诊断提供了一种新的方法.
Fault feature enhancement and diagnosis method of rolling bearing based on boundary-assisted discrimination
As the vital component of mechanical equipment,rolling bearings play a critical role in ensuring the safety and stability of equipment.To solve the problem of data imbalance in rolling bearing fault diagnosis,a boundary-assisted discriminative auxiliary classifier generative adversarial network(BD-ACGAN)model was proposed.Firstly,a boundary-assisted discriminator was designed to extract the boundary feature details of faulty samples,to guide the generator to generate more realistic samples for addressing the issue of data imbalance.Secondly,an adaptive weight loss module was employed to dynamically adjust the loss weights,enabling the model to focus more on important feature information and improve the quality of sample generation and feature representation.Further,generated samples and real samples were employed for enhancement training of BD-ACGAN model,to improve the generalization ability and diagnostic capability.Finally,the feature enhancement capability and diagnostic effectiveness of BD-ACGAN model were verified through ablation experiments and comparison experiments.The rolling bearing data sets from Case Western Reserve University and Xi'an Jiaotong University were used to verify the model.The experimental results show that BD-ACGAN model can effectively use the boundary characteristics of fault samples to solve the problem of data imbalance,and the fault diagnosis accuracy is 98.79%,which is better than other control models,and provides a new method for the fault diagnosis of rolling bearings.

bearing fault diagnosisdata imbalanceboundary-assisted discriminative auxiliary classifier generative adversarial network(BD-ACGAN)fault feature enhancementadaptive weight lossdataset augmentation

李佰霖、鲁大臣、付文龙、陈禹朋

展开 >

三峡大学 电气与新能源学院,湖北 宜昌 443002

轴承故障诊断 数据不平衡 边界辅助判别的辅助分类生成对抗网络 故障特征增强 自适应权重损失 数据集增广

国家自然科学基金资助项目梯级水电站运行与控制湖北省重点实验室开放基金资助项目

517419072022KJX10

2024

10.3969/j.issn.1001-4551.2024.04.009

机电工程
浙江大学　浙江省机电集团有限公司

机电工程

CSTPCD北大核心
影响因子：0.785
ISSN：1001-4551
年,卷(期)：2024.41(4)
  • 19