张韦 张永 骈晓琴 苏赫 蔺相东

摘要：风机齿轮箱振动信号成分复杂，而经验模态分解（ EMD ）在故障诊断中存在模态混叠和端点效应问题。针对此问题，研究了一种 EEMD 样本熵和高斯径向基核函数的 SVM分类器的滚动轴承故障诊断方法。以风机齿轮箱滚动轴承为研究对象，提取了内圈故障、外圈故障、滚动體故障和正常轴承4种状态振动信号，利用 EEMD和小波分别对振动信号分解降噪并筛选主要 IMF 分量;计算前4阶 IMF 分量的样本熵作为特征向量;最后将特征向量输入高斯径向基核函数的 SVM模型进行故障识别。结果表明：EEMD 算法对端点效应和模态混叠都有一定抑制作用，EEMD 样本熵和 SVM 相结合可有效识别滚动轴承故障类型，故障识别率为97.5%，为工程应用中风机齿轮箱滚动轴承故障诊断提供参考。

关键词：滚动轴承;EEMD 分解;样本熵;SVM;故障诊断

中图分类号：TH17             文献标志码：A

文章编号：1009-9492（2021）12-0038-04

开放科学（资源服务）标识码（OSID）：

Fault Diagnosis of Rolling Bearing Based on Improved EMD SampleEntropy and SVM

Zhang Wei1，Zhang Yong1，Pian Xiaoqin2，Su He1※，  Lin Xiangdong1

（1. School of Mechanical and Electrical Engineering， Inner Mongolia Agricultural University， Hohhot 010018， China;

2. Department of Hematology， Affiliated Hospital of Inner Mongolia Medical University， Hohhot 010050， China）

Abstract： The components of the vibration signal of the wind turbine （WT） gearbox are complex， and empirical mode decomposition （EMD） has problems of modal aliasing and end effect. Aiming at this problem， a rolling bearing fault diagnosis method based on EEMD sample entropy and Gaussian radial basis kernel function SVM was studied. Taking the rolling bearing of the wind turbine gearbox as the research object， the four state vibration signals of inner ring failure， outer ring failure， rolling element failure and normal bearing were extracted， and the vibration signals were decomposed and denoised respectively by EEMD and wavelet and the main IMF components were screened. The sample entropy of the first four-order IMF components was calculated as the feature vector. Finally， the feature vector was input into the SVM model of the Gaussian radial basis kernel function for fault identification. The results show that the EEMD algorithm has a certain inhibitory effect on the end effect and modal aliasing. The combination of EEMD sample entropy and SVM can effectively identify the type of rolling bearing fault， and the fault recognition rate is 97.5%， which provides a reference for the fault diagnosis of rolling bearing of WT gearbox in engineering application.

Key words： rolling bearing; EEMD decomposition; sample entropy; SVM; fault diagnosis

0 引言

风能作为一种无污染、可再生的绿色能源，近年来获得了井喷式发展，2018年风能累积量（GW）比2015年高出约43%[1]。由于风电机组运行环境极度恶劣，齿轮箱各部件故障频发，滚动轴承作为风电机组齿轮箱的关键旋转部件，也是最容易损坏的零件之一，而故障停机导致的维护成本非常高，因此轴承故障诊断技术越来越得到重视[2-4]，这对于保障机械装备的安全可靠运行、维护维修具有重要的工程实际意义。

经验模态分解 （Empirical  Mode  Decomposition， EMD ）是 Huang等[5]于1998年提出的一种自适应信号处理方法，其本身存在模态混叠和端点效应问题[1，6]。……