ELMD与排列熵在滚动轴承故障诊断中的应用

摘要

Aiming at the problem that the bearing fault signal was submerged by strong background noise so that it was difficult for the traditional envelope demodulation method to extract fault features, a method of bearing fault diagnosis ,which is based on ensemble local mean decomposition( ELMD) and permutation en-tropy(PE) was proposed. First, bearing vibration signal was decomposed into a series of narrow band prod-uct functions( PFs) using ELMD method. Then, the PE value of PFs was calculated to construct high-di-mensional feature vectors. Finally, the high-dimensional feature vectors were transformed as input of multi-fault classifier to identify bearing fault types. The experimental results indicated that ELMD method can re-strain the mode mixing effectively, permutation entropy distribution of PF components can response signal features under different working status and the intelligent diagnosis approach based on ELMD and permuta-tion entropy can identify the operating conditions and fault types of bearing accurately.%针对轴承故障信号往往被强背景噪声淹没,采用传统包络解调方法难以提取故障特征的问题,提出总体局部均值分解(ensemble local mean decomposition,ELMD)与排列熵(permutation entropy,PE)相结合的轴承故障诊断方法.首先,对轴承振动信号进行ELMD分解并得到一系列窄带乘积函数(product function,PF),然后,计算各PF分量排列熵以构造高维特征向量,最后将高维特征向量作为多故障分类器的输入来识别轴承故障类型.实验结果表明ELMD方法可以有效地抑制模态混叠;PF分量的排列熵分布可以反应轴承不同工作状态下的信号特征;基于ELMD与排列熵的智能诊断方法可以准确地识别轴承的工作状态和故障类型.