基于HMFDE和t-SNE的旋转机械故障诊断方法  

Fault diagnosis method of rotating machinery based on HMFDE and t-SNE

作　　者：尹久[1] 张杰[2] YIN Jiu;ZHANG Jie(School of Mechanical and Electrical Engineering,Hubei Light Industry Technology Institute,Wuhan 430070,China;School of Mechanical and Electrical Engineering,Wuhan University of Technology,Wuhan 430070,China)

机构地区：[1]湖北轻工职业技术学院机电工程学院,湖北武汉430070 [2]武汉理工大学机电工程学院,湖北武汉430070

出　　处：《机电工程》2024年第6期1058-1067,共10页Journal of Mechanical & Electrical Engineering

基　　金：国家自然科学基金资助项目(51577065)。

摘　　要：针对旋转机械的故障特征提取较难,以及单一类型的特征无法全面反映故障特性的问题,提出了一种基于混合多尺度波动散布熵(HMFDE)、t分布-随机邻域嵌入(t-SNE)和郊狼优化算法(COA)优化极限学习机(ELM)的旋转机械故障诊断方法。首先,采用特征加权提出了混合多尺度波动散布熵方法,并将其用于提取旋转机械振动信号的故障特征;随后,采用t-SNE方法对混合故障特征进行了特征降维,挑选出了最能够反映故障特性的特征子集,构建了敏感特征样本;最后,采用郊狼优化算法对极限学习机的输入权重和隐含层阈值进行了优化,完成了旋转机械的故障识别和分类;以齿轮箱和滚动轴承故障数据集为对象,对基于HMFDE、t-SNE和COA-ELM的故障诊断方法进行了实验,验证了方法的有效性。研究结果表明:采用HMFDE-t-SNE-CAO-ELM故障诊断方法可以取得100%的故障识别准确率,该方法能够有效地诊断旋转机械的不同故障类型和损伤;相较于基于单一类型特征的故障诊断方法,其准确率分别可以提高0.68%、22.42%、29.18%(齿轮箱)和1.43%、8.23%、23.67%(滚动轴承),虽然牺牲了一定的计算效率,但准确率得到了明显的提高;相较于其他常规故障分类器,COA-ELM的故障识别准确率具有明显的优势。Aiming at the problem of fault feature extraction of rotating machinery and the fault features of a single type cannot fully reflect the fault characteristics,a rotating machinery fault diagnosis method based on hybrid multi-scale fluctuation dispersion entropy(HMFDE),t-distributed stochastic neighbor embedding(t-SNE),and coyote optimization algorithm(COA)-extreme learning machine(ELM) was proposed.Firstly,a hybrid multiscale fluctuation dispersion entropy method was proposed by using feature weighting and used to extract fault characteristics of vibration signals of rotating machinery.Then,the feature dimension reduction of mixed fault features was carried out by using t-SNE method,and the feature subset that could best reflect fault characteristics was selected to construct sensitive feature samples.Finally,coyote optimization algorithm was used to optimize the input weight value and hidden layer threshold of the extreme learning machine to realize fault identification and classification of rotating machinery.Taking gearbox and rolling bearing fault data set as objects,experiments were conducted on the fault diagnosis methods based on HMFDE,t-SNE and COA-ELM,and the effectiveness of the mothods was overified.The results show that the HMFDE-t-SNE-CAO-ELM fault diagnosis method achieves 100% fault identification accuracy and can effectively diagnose different fault types and damage of rotating machinery.Comparing to fault diagnosis methods based on single type features,the accuracy has been improved by 0.68 %,22.42 %,29.18%(gearbox) and 1.43%,8.23%,23.67%(rolling bearing),respectively.Although some efficiency has been sacrificed,the accuracy of this method has been significantly improved.Comparing with other conventional fault classifiers,COA-ELM has obvious advantages in fault recognition accuracy.

关 键 词：旋转机械 故障诊断 齿轮箱 滚动轴承 混合多尺度波动散布熵 t分布-随机邻域嵌入 郊狼优化算法 极限学习机 

分 类 号：TH133.3[机械工程—机械制造及自动化]