摘要
针对滚动轴承振动信号的非平稳非线性特性,提出了一种采用集合经验模态分解(EEMD)熵特征提取、t-分布邻域嵌入(t-SNE)和粒子群优化-概率神经网络(PSO-PNN)的滚动轴承故障诊断方法。首先,对振动信号应用EEMD算法实现分解,生成多个固有模态函数(IMFs),对生成的含有主要故障信息的模态分量进行选择,以进一步实现熵特征提取,然后对高维特征数据应用t-SNE算法进行降维,最后利用PSO-PNN分类器进行故障识别。通过案例1和案例2的分析结果表明:该方法对滚动轴承故障识别率均达到100%,具有较高的故障识别率,能对滚动轴承的故障类型有效的识别。
Aiming at the nonstationary nonlinearity characteristics of rolling bearing vibration signal,a fault diagnosis method for rolling bearing that uses ensemble empirical mode decomposition(EEMD)entropy feature extraction,t-distributed stochastic neighbor embedding(t-SNE)and particle swarm optimization-probabilistic neural network(PSO-PNN)is proposed.Firstly,EEMD algorithm is applied to decompose the vibration signal to generate several intrinsic modal functions(IMFs),and the generated modal components containing the major fault information are selected to further realize entropy feature extraction.Then,t-SNE algorithm is applied to the high-dimensional feature data for dimensionality reduction,and finally,PSO-PNN classifier is used for fault identification.The analysis results of case 1 and case 2 show that the method has a high fault recognition rate of 100%for rolling bearing fault recognition,and can effectively identify the type of rolling bearing fault.
作者
高淑芝
王拳
张义民
GAO Shu-zhi;WANG Quan;ZHANG Yi-min(Institute of Equipment Reliability,Shenyang University of Chemical Technology,Liaoning Shenyang 110000,China;School of Information Engineering,Shenyang University of Chemical Technology,Liaoning Shenyang 110000,China)
出处
《机械设计与制造》
北大核心
2023年第6期229-233,共5页
Machinery Design & Manufacture
基金
NSFC-国家自然科学重点基金—辽宁联合基金(U1708254)
辽宁省特聘教授(No[.2018]3533)。
