基于改进MEEMD与DE-OSELM的滚动轴承故障诊断方法

Fault diagnosis method of rolling bearing based on improved MEEMD and DE-OSELM

针对现场采集的信号因混有噪声而使故障特征提取困难的问题,基于奇异值分解(SVD)、改进的集合经验模态分解(MEEMD)、差分进化算法(DE)、在线贯序极限学习机(OSELM),提出了一种基于改进MEEMD与DE-OSELM的滚动轴承故障诊断方法.首先进行MEEMD分解,对MEEMD算法中经排列熵筛选出的异常IMF分量进行SVD降噪,与剩余信号重构后直接进行EMD分解;其次提取各IMF分量的能量作为特征构造特征集;最后将获得的特征集作为DE-OSELM的输入进行训练并识别故障类型.对实际4种不同健康状态的滚动轴承样本进行分类识别,并与常用分类方法进行比较.结果表明:该方法具有更高的准确率,可以更有效地实现故障诊断.

In order to solve the difficult problem of fault feature extraction due to the mixed noise in the field collected signals,a rolling bearing fault diagnosis method based on improved MEEMD and DE-OSELM was proposed based on SVD,EEMD,DE and OSELM.Firstly,the signal was decomposed by improved MEEMD,the abnormal IMF components selected by permutation entropy were denoised by SVD,and the denoised signal and the remaining signal were reconstructed by EMD.Secondly,the energy of each IMF component was extracted as the feature set.Finally,the obtained feature set was used as the input of DE-OSELM to train and identify the fault type.The actual four types of rolling bearing samples with different health conditions were classified and identified,and compared with common classification methods,the results showed that the proposed method had higher accuracy and could effectively realize fault diagnosis.