基于VMD消噪处理的滚动轴承早期故障识别
发布时间：2018-11-12 07:36
【摘要】：提出了一种基于变分模态分解(VMD)消噪和核模糊C均值(KFCM)聚类相结合的滚动轴承早期故障识别方法。首先提出一种通过综合运用泄漏能量和互相关系数函数确定VMD预设尺度数K的新方法,弥补了VMD方法通常按经验选取预设尺度数方法的不足;然后对振动信号进行VMD分解得到K个限带的内禀模态函数(BIMF)分量,利用归一化的自相关系数函数能量集中比大于0.9的原则确定含有噪声的BIMF分量,并剔除这些含噪BIMF分量,再将剩余的BIMF分量叠加进行信号重构,实现了信号的消噪;最后计算各样本重构信号的均方根值和归一化能量值得到二维特征向量样本集,并输入到KFCM聚类器进行故障诊断。利用实测轴承故障数据进行验证,结果表明与经验模态分解(EMD)方法相比,可以有效地实现滚动轴承早期故障诊断。
[Abstract]:This paper presents an early fault identification method for rolling bearings based on variational mode decomposition (VMD) (VMD) de-noising and kernel fuzzy C-means (KFCM) clustering. Firstly, a new method is proposed to determine the preset scale number K of VMD by synthetically using leakage energy and correlation number function, which makes up for the deficiency of VMD method which usually selects preset scale number according to experience. Then the intrinsic mode function (BIMF) component of K band is obtained by VMD decomposition of the vibration signal, and the BIMF component with noise is determined by the principle of normalized autocorrelation coefficient function energy concentration ratio greater than 0.9. These noisy BIMF components are eliminated, and then the remaining BIMF components are superimposed to reconstruct the signal to realize signal de-noising. Finally, the mean square root value and normalized energy value of each sample reconstruction signal are calculated to obtain the two-dimensional eigenvector sample set, and input into the KFCM cluster for fault diagnosis. Compared with the empirical mode decomposition (EMD) method, the early fault diagnosis of rolling bearing can be realized effectively.
【作者单位】： 燕山大学河北省重型机械流体动力传输与控制重点实验室;燕山大学先进锻压成形技术与科学教育部重点实验室;郑州中车四方轨道车辆有限公司;
【基金】：国家自然科学基金(51475405) 国家重点基础研究发展计划(973计划)(2014CB046405) 秦皇岛市科学技术研究与发展计划(201502A041)
【分类号】：TH133.33
本文编号：2326488
[Abstract]:This paper presents an early fault identification method for rolling bearings based on variational mode decomposition (VMD) (VMD) de-noising and kernel fuzzy C-means (KFCM) clustering. Firstly, a new method is proposed to determine the preset scale number K of VMD by synthetically using leakage energy and correlation number function, which makes up for the deficiency of VMD method which usually selects preset scale number according to experience. Then the intrinsic mode function (BIMF) component of K band is obtained by VMD decomposition of the vibration signal, and the BIMF component with noise is determined by the principle of normalized autocorrelation coefficient function energy concentration ratio greater than 0.9. These noisy BIMF components are eliminated, and then the remaining BIMF components are superimposed to reconstruct the signal to realize signal de-noising. Finally, the mean square root value and normalized energy value of each sample reconstruction signal are calculated to obtain the two-dimensional eigenvector sample set, and input into the KFCM cluster for fault diagnosis. Compared with the empirical mode decomposition (EMD) method, the early fault diagnosis of rolling bearing can be realized effectively.
【作者单位】： 燕山大学河北省重型机械流体动力传输与控制重点实验室;燕山大学先进锻压成形技术与科学教育部重点实验室;郑州中车四方轨道车辆有限公司;
【基金】：国家自然科学基金(51475405) 国家重点基础研究发展计划(973计划)(2014CB046405) 秦皇岛市科学技术研究与发展计划(201502A041)
【分类号】：TH133.33
【相似文献】
相关重要报纸文章 前2条
1 李瀛寰;VMD联盟加速移动多媒体产业链[N];中国计算机报;2004年
2 沈蓓;VMD技术拓宽企业自主创新之路[N];人民邮电;2005年
，本文编号：2326488
本文链接：https://www.wllwen.com/jixiegongchenglunwen/2326488.html
