基于EEMD和SOM的轴承故障诊断

Fault Diagnosis Method of Bearing Based on EEMD and SOM Neural Networks

EMD算法所存在的端点效应及模态混叠问题会导致故障轴承特征值提取时精度不高等问题。为提高故障诊断系统的准确性,文中选用一种基于EMD的改进算法EEMD,通过提取时域特征参数,衡量故障诊断程度,判断故障类型。运用SOM神经网络进行诊断识别。文中在进行实际实验分析的基础上,总结了实验平台的框架、典型的故障轴承类型,分析了选用的特征参数对故障诊断的不同判别方向,并提出基于集合经验模态与SOM神经网络的轴承故障诊断方法。

The end-point effect and the mode overlapping existing in the EMD algorithm can lead to problems such as low accuracy when extracting the eigenvalues of faulty bearings.In order to improve the accuracy of the fault diagnosis system,this paper selects an improved algorithm EEMD based on EMD,which measures the degree of fault diagnosis and determines the fault type by extracting time-domain feature parameters.The SOM neural network is used for diagnosis and identification.Based on the actual experimental analysis,this paper summarizes the framework of the experimental platform,the typical types of faulty bearings,analyzes the different discriminative directions of the selected feature parameters for fault diagnosis,and proposes a bearing fault diagnosis method based on ensemble empirical modalities and SOM neural networks.