应用ICEEMDAN和SVM的行星齿轮箱故障诊断  

Fault Diagnosis of Planetary Gearbox using ICEEMDAN and SVM

作　　者：王浩楠 崔宝珍[1] 彭智慧 任川 WANG Haonan;CUI Baozhen;PENG Zhihui;REN Chuan(College of Mechanical Engineering,North University of China,Taiyuan 030051,China)

机构地区：[1]中北大学机械工程学院,太原030051

出　　处：《机械科学与技术》2023年第1期24-30,共7页Mechanical Science and Technology for Aerospace Engineering

基　　金：国家自然科学基金项目(51175480);山西省重点研发计划(国际合作)项目(201903D421008);中北大学先进制造技术山西省重点实验室开放课题研究基金(XJZZ202007)。

摘　　要：针对行星齿轮箱复合故障准确分类问题,应用了改进自适应噪声完备集合经验模态分解(ICEEMDAN)和支持向量机(SVM)相结合的故障诊断方法。首先,将行星齿轮箱的不同故障信号分别进行ICEEMDAN分解,得到各阶内禀模态函数(IMF);其次,利用各阶IMF分量与原信号的相关性大小,剔除虚假的IMF分量;最后,以优选IMF分量的多尺度模糊熵均值作为特征向量,输入到多分类SVM中进行故障分类,分类准确率高达100%,实验结果证明了该方法的可行性。Aiming at the problem of accurate classification of compound faults of planetary gearboxes,a fault diagnosis method combining improved adaptive noise complete set empirical mode decomposition(ICEEMDAN)and support vector machine(SVM)is proposed in this study.First,the different fault signals of the planetary gearbox are decomposed by ICEEMDAN to obtain the intrinsic mode function(IMF)of each order.Second,the correlation between the IMF component of each order and the original signal is used to remove the false IMF component.Finally,the multi-scale fuzzy entropy average value of the preferred IMF component is used as the feature vector and input into the multi-class SVM to accomplish fault classification.The classification accuracy is as high as 100%.The experimental results prove the feasibility of this method.

关 键 词：改进自适应噪声完备集合经验模态分解 频域互相关 多尺度模糊熵 支持向量机 行星齿轮箱故障 

分 类 号：TH132.41[机械工程—机械制造及自动化]