变负载轴承故障诊断卷积神经网络模型
DOI:
作者:
作者单位:

作者简介:

通讯作者:

中图分类号:

TH165 +.3

基金项目:

国家自然科学基金项目(面上项目,重点项目,重大项目)


Variable Load Bearing Fault Diagnosis Model Based on Convolutional Neural Network
Author:
Affiliation:

Fund Project:

The National Natural Science Foundation of China (General Program, Key Program, Major Research Plan)

  • 摘要
  • |
  • 图/表
  • |
  • 访问统计
  • |
  • 参考文献
  • |
  • 相似文献
  • |
  • 引证文献
  • |
  • 资源附件
  • |
  • 文章评论
    摘要:

    针对轴承故障在实际环境中存在负载变化导致故障难以诊断的问题,提出一种基于一维卷积神经网络的变负载适应轴承故障诊断模型:卷积结构使用小卷积核卷积层堆叠的形式,训练时对输入层进行均匀分布失活率的随机失活,以提高网络的变负载适应能力,且采用全局平均池化降低模型计算量和减轻过拟合程度;在实验验证阶段,提出以两种近邻负载条件的轴承数据构成变负载数据集,充分验证轴承故障诊断的变负载适应性。实验结果表明:本文模型在各测试集均能达到96%以上的准确率且平均准确率达到98.36%,能够在变负载环境下实现准确、稳定的轴承故障诊断,具有良好的泛化能力。

    Abstract:

    Aiming at the problem that bearing faults are difficult to diagnose due to load changes in the actual environment, a variable load adaptive bearing fault diagnosis model based on one-dimensional convolutional neural network is proposed. The convolution structure uses the form of convolutional layers stack of many small convolution kernels. During training, the input layer was subjected to random deactivation of the uniform distribution rate to improve the variable load adaptability of the model,and used global average pooling to reduce model calculations and reduce overfitting. In the experimental verification stage, it is proposed to use the bearing data of two neighboring load conditions to form the variable load data set, which fully verifies the variable load adaptability of bearing fault diagnosis. The experimental results show that the model can achieve more than 96% accuracy in each test set, and the average accuracy rate is 98.36%. The model can realize accurate and stable bearing fault diagnosis under variable load environment and has good generalization ability.

    参考文献
    相似文献
    引证文献
引用本文

祝道强,周新志,宁芊. 变负载轴承故障诊断卷积神经网络模型[J]. 科学技术与工程, 2020, 20(15): 6054-6059.
Zhu Daoqiang, Zhou Xinzhi, Ning Qian. Variable Load Bearing Fault Diagnosis Model Based on Convolutional Neural Network[J]. Science Technology and Engineering,2020,20(15):6054-6059.

复制
分享
文章指标
  • 点击次数:
  • 下载次数:
  • HTML阅读次数:
  • 引用次数:
历史
  • 收稿日期:2019-09-03
  • 最后修改日期:2020-06-14
  • 录用日期:2019-11-25
  • 在线发布日期: 2020-06-24
  • 出版日期:
×
律回春渐,新元肇启|《科学技术与工程》编辑部恭祝新岁!
亟待确认版面费归属稿件,敬请作者关注