摘要
针对滚动轴承故障诊断中振动信号易受强背景噪声干扰,出现非平稳、非线性的特性,导致故障诊断精度较低等问题,提出了一种基于奇异谱分解(SSD)和一维卷积神经网络(1DCNN)的滚动轴承故障诊断方法.首先,利用SSD将原始振动信号分解成若干个频率尺度的奇异谱(SSC)分量,并根据峭度准则选取有效SSC分量对信号进行重构;然后,构建一维卷积神经网络结构,先将重构后的信号输入模型进行训练,充分提取信号的特征,再由输出层输出诊断结果;最后,进行滚动轴承故障诊断实验,结果表明:提出的诊断方法诊断准确率达到98.9%,比传统方法具有更高的准确性和稳定性.
The rolling bearing signal under strong background noise has non-stationary and nonlinear characteristics,resulting in the low accuracy of fault diagnosis.To solve the problem,a method based on singular spectrum decomposition(SSD) and 1 D convolutional neural network(1 DCNN) was proposed.In this method,the original vibration signal was decomposed into several singular spectrum components(SSC) with different frequency scales by SSD,and the SSC was selected according to the kurtosis criterion to reconstruct the signal.In addition,the 1 D(one dimension) convolutional neural network was used to extract the fault feature from the reconstructed signal and obtain diagnosis results.Finally,the experimental results prove the effectiveness and superiority of the proposed method.The diagnosis accuracy is up to 98.9%,which is more accurate and stable than other traditional methods.
作者
宋霖
宿磊
李可
苏文胜
SONG Lin;SU Lei;LI Ke;SU Wensheng(School of Mechanical Engineering,Jiangnan University,Wuxi 214122,Jiangsu China;Jiangsu Key Laboratory of Advanced Food Manufacturing Equipment&Technology,Jiangnan University,Wuxi 214122,Jiangsu China;Jiangsu Province Special Equipment Safety Supervision Inspection Institute Branch of Wuxi,Wuxi214071,Jiangsu China)
出处
《华中科技大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2020年第12期38-43,共6页
Journal of Huazhong University of Science and Technology(Natural Science Edition)
基金
国家自然科学基金资助项目(51705203,51775243)
高等学校学科创新引智计划资助项目(B18027)。
关键词
滚动轴承
故障诊断
奇异谱分解
峭度
一维卷积神经网络
rolling bearing
fault diagnosis
singular spectrum decomposition(SSD)
kurtosis
1D convolutional neural network(1DCNN)
