摘要
在故障诊断过程中,为了更好地提取特征以及提高故障识别率,提出了一种基于离散小波变换和深度可分离神经网络算法以及SVM分类器的滚动轴承故障诊断方法。首先,模型利用离散小波变换对原始振动信号提取特征,形成多通道样本;然后对样本进行深度可分离卷积神经网络训练,最后在全连接层后接SVM分类器实现对故障信号的分类。实验所用数据来自CTU-2实验平台,故障标签共有10类。实验结果表明,相比较单一使用小波变换提取特征或者CNN卷积神经网络分类的方法,该模型的诊断效果更加优秀。
In order to better extract features and improve fault recognition rate in fault diagnosis process,a rolling bearing fault diagnosis method based on discrete wavelet transform and depth separable neural network algorithm and SVM classifier was proposed.Firstly,the model uses discrete wavelet transform to extract features from the original vibration signal to form multi-channel samples.Then,the samples were trained by the deep separable convolutional neural network,and finally the fault signal was classified by the SVM classifier after the full connection layer.The data used in the experiment came from CTU-2 experimental platform,and there were 10 categories of fault tags.The experimental results show that the diagnosis effect of the model is better than that of the method of extracting features by wavelet transform or CNN convolutional neural network.
作者
杨瑞恒
唐向红
陆见光
陈功胜
YANG Rui-heng;TANG Xiang-hong;LU Jian-guang;CHEN Gong-sheng(Key Laboratory of Modern 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
《组合机床与自动化加工技术》
北大核心
2021年第6期76-80,共5页
Modular Machine Tool & Automatic Manufacturing Technique
基金
边缘计算环境下的旋转机械复合故障诊断研究(黔科合基础-ZK[2021]一般271)
贵州省2018年本科教学内容和课程体系改革项目(2018520081)阶段性成果
贵州省公共大数据重点实验室开放基金资助项目(2017BDKFJJ019)
贵州大学引进人才基金(贵大人基合字(2016)13号)。
