摘要
针对齿轮常见故障及信号在传统EMD算法分解中产生的端点效应,提出一种基于改进经验模态分解(EMD)与快速独立分量分析(FastICA)-样本熵的齿轮故障特征提取方法。首先对信号进行EMD分解,得到一系列IMF分量和残余量,在此过程中通过匹配差别最小的极值包络线段确定端点处极值,然后从每个信号中分别选取周期性明显的分量与原始信号组成混合信号作为FastICA的输入,获得ICA计算后的分量,最后分别计算EMD分量与各独立分量的样本熵。实验结果表明,改进后的EMD算法可以有效改善端点效应问题,并通过与EMD-样本熵的对比,表明FastICA-样本熵能更明显、稳定地反映齿轮故障,因此可作为一种有效的故障特征。
Aiming at common gear faults and the endpoint effect of signals in the traditional EMD Decomposition algorithm,a method based on the improved empirical mode decomposition (EMD) of fast independent component analysis (FastICA)-sample entropy (SampEn) is proposed to extract the characteristics of gear faults.Firstly,the EMD decomposition of the signal is carried out to obtain a series of IMF components and residual quantities.In this process,the extreme values of the endpoints are determined by matching the minimum envelope segments with the smallest difference,and then the periodic significant components and the original signals are selected from each signal.The mixed signal is composed as the input of FastICA to obtain the calculated component of ICA,and finally the sample entropy of the EMD component and each independent component is calculated separately.Through experimental study,it is verified that the improved EMD algorithm can effectively improve the endpoint effect problem,and the comparison with EMD-sample entropy shows that FastICA-sample entropy can reflect gear faults more obviously and stably,so it can be used as an effective fault feature.
作者
吕同昕
LV Tong-xin(School of Computer Science and Engineering,Shandong University of Science and Technology,Qingdao 266590,China)
出处
《软件导刊》
2019年第8期154-158,共5页
Software Guide
基金
国家重点研发计划项目(2017YFC0804406)
山东省重点研发计划项目(2016ZDJS02A05)
