基于VMD能量熵与SVM的行星齿轮箱故障诊断  

Fault Diagnosis of Planetary Gearbox Based on VMD Energy Entropy and SVM

作　　者：彭智慧 崔宝珍[1] 任川 王浩楠 PENG Zhihui;CUI Baozhen;REN Chuan;WANG Haonan(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：《机械设计与研究》2021年第6期78-82,共5页Machine Design And Research

基　　金：国家自然科学基金项目(51175480);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：由于行星齿轮箱振动信号的故障特征难被提取,故采用变分模态分解(VMD)能量熵与支持向量机(SVM)相结合的方式实现行星齿轮箱故障诊断。首先利用VMD方法将振动信号分解为不同尺度的内稟模态函数(IMF)并提取各IMF的能量熵值构成特征矩阵,其次利用粒子群算法(PS0)对支持向量机的惩罚因子和核函数优化,最后将特征矩阵输入支持向量机进行故障模式识别。通过行星齿轮箱的实验研究,验证了该方法的有效性并且识别准确率高达99.625%。It is difficult to extract the fault characteristics of the vibration signal of the planetary gearbox.Therefore,the method of combining the variational modal decomposition(VMD)energy entropy and the support veetor machine(SVM)is used to realize the fault diagnosis of the planetary gearbox.Firstly,the VMID method is used to deconpose the vibration signal into intrinsie modal funetions(IMF)of diferent scales,and the energy entropy values of each IMF are extracted to form a characteristic matrix.Secondly,particle swarm optimization(PSO)is used to optimize the penalty factor and kernel function of the support vector machine.Finally,the feature matrix is input into the support vector machine for failure mode recognition.Through experimental research on planetary gearboxes,the ffectiveness of the method is verifed,and the recognition accuracy is as high as 99.625%.

关 键 词：变分模态分解 能量熵 粒子群算法 支持向量机 故障诊断 

分 类 号：TH132.41[机械工程—机械制造及自动化]