摘要
为解决旋转机械故障类型多、等级不均衡的故障诊断难题,构建了一种基于ID3决策树与卷积神经网络(ID3-CNN)的故障诊断模型。首先对原始信号进行人工时域特征提取,使用t-SNE降维可视化提取出特征混叠的故障,而后利用卷积运算对特征混叠的故障进行二次特征提取,提高模型的特征表达能力,最后使用ID3决策树和卷积神经网络对不同等级的故障进行分类。在轴承数据集上对模型进行了验证,结果表明,严重故障的诊断准确率达到100%,轻微故障的诊断准确率达到95%。与传统的支持向量机及二维卷积神经网络比较,提高了模型的诊断准确率及特征提取能力。
To solve the diagnostic problem of multiple rotating mechanical fault types and unbalanced grades,this paper constructs a fault diagnosis model based on ID3 decision tree and convolutional neural network(ID3-CNN).The original signal artificial time domain feature extraction is carried out,t-SNE dimension reduction visualization is used to extract feature overlapping fault.And then the feature aliasing fault secondary feature extraction is performed using the convolution operation to improve the feature expression ability of the model.Finally the ID3 decision tree and convolutional neural network is used to classify different levels of fault.The model is validated on the bearing dataset,and the results show that the diagnostic accuracy of severe faults reaches 100%,and the diagnostic accuracy of minor faults reaches 95%.Compared with the traditional support vector machine and two-dimensional convolutional neural network,the diagnostic accuracy and feature extraction ability of the model are improved.
作者
王承超
王湘江
WANG Chengchao;WANG Xiangjiang(School of Mechanical Engineering,University of South China,Hengyang 421001,China)
出处
《机械工程师》
2024年第3期38-43,共6页
Mechanical Engineer
关键词
旋转机械
故障诊断
特征提取
卷积神经网络
ID3决策树
rotating machinery
fault classification diagnosis
feature extraction
convolutional neural network
ID3 decision tree
