应用CEEMD降噪与自适应MOMEDA的轴承故障特征提取方法
Rolling bearing fault feature extraction method based on CEEMD denoising and adaptive MOMEDA
宋宇博 1张宇飞1
作者信息
- 1. 兰州交通大学机电技术研究所,甘肃兰州 730070
- 折叠
摘要
针对滚动轴承早期故障信号中冲击成分能量低且易被强烈的背景噪声所淹没的问题,该文提出一种基于互补集合经验模态分解(complete ensemble empirical mode decomposition,CEEMD)-小波阈值降噪和参数自适应多点最优最小熵解卷积(multipoint optimal minimum entropy deconvolution adjusted,MOMEDA)的滚动轴承故障特征提取方法.将CEEMD与小波阈值降噪结合对原始信号进行降噪;提出一种新的复合指标:峭度-包络波形因子,并以其为适应度函数设计变步长搜索法,对MOMEDA算法的滤波器长度进行寻优;基于寻优的滤波器长度对降噪的信号进行MOMEDA解卷积,并通过包络谱分析识别滚动轴承的故障特征频率.对比实验结果表明:以该文寻找的最优滤波器长度作为MOMEDA的参数,解卷积后包络谱故障频率更加清晰;且相较于传统的MOMEDA算法和小波阈值降噪-MOMEDA方法,该文提出的方法能够更有效地提取强噪声背景下微弱的故障特征信息.
Abstract
Aiming at the problem that the impact component in the early fault signal of rolling bearing has low energy and is easy to be submerged by strong background noise,In this paper,a rolling bearing fault feature extraction method based on complete ensemble empirical mode decomposition(CEEMD)-wavelet threshold denoising and parameter adaptive multipoint optimal minimum entropy deconvolution adjusted(MOMEDA)is proposed.Combining CEEMD with wavelet threshold denoising to denoise the original signal;A new composite index:kurtosis envelope waveform factor is proposed,and taking it as the fitness function,a variable step search method is designed to optimize the filter length of MOMEDA algorithm;Based on the optimized filter length,the denoised signal is deconvoluted by MOMEDA,and the fault characteristic frequency of rolling bearing is identified by envelope spectrum analysis.The comparative experimental results show that taking the optimal filter length found in this paper as the parameter of MOMEDA,the envelope spectrum fault frequency is clearer after deconvolution;Compared with the traditional MOMEDA algorithm and wavelet threshold denoising MOMEDA method,the method proposed in this paper can extract weak fault feature information under strong noise background more effectively.
关键词
滚动轴承/故障诊断/多点最优最小熵解卷积/互补集合经验模态分解/小波阈值降噪Key words
rolling bearing/fault diagnosis/MOMEDA/CEEMD/wavelet threshold denoising引用本文复制引用
出版年
2024
NETL
NSTL
万方数据