基于VMD-SWT的降噪方法在转子振动信号中的应用  

A Denoising Method Based on Variational Mode Decomposition and Soft Wavelet-thresholding and its Application to Vibration Signals

作　　者：孟湘 曾洪涛[1,2] 刘冬[3] 肖志怀[1,2] 黄宗磊 MENG Xiang;ZENG Hong-tao;LIU Dong;XIAO Zhi-huai;HUANG Zong-lei(Key laboratory of Hydraulic Machinery Transients,Wuhan University,Wuhan 430072,China;School of Power and Mechanical Engineering,Wuhan University,Wuhan 430072,China;School of Civil and Hydraulic Engineering,Huazhong University of Science and Technology,Wuhan 430074,China;China Yantze Power Co.Ltd.,Yichang 443000,Hubei Province,China)

机构地区：[1]武汉大学水力机械过渡过程教育部重点实验室,武汉430072 [2]武汉大学动力与机械学院,武汉430072 [3]华中科技大学土木与水利工程学院,武汉430074 [4]中国长江电力股份有限公司,湖北宜昌443000

出　　处：《中国农村水利水电》2021年第6期164-168,共5页China Rural Water and Hydropower

基　　金：国家自然科学基金项目(51979204)。

摘　　要：为提高水电机组转子故障振动信号降噪后的信噪比,获得更好的降噪效果,提出一种基于样本熵的变分模态分解和小波软阈值相结合的降噪方法,通过对转子试验台所产生的正常、转子不对中、不平衡和碰磨4种工况下的转子垂直振动信号进行变分模态分解、计算内禀模态函数的样本熵、用小波软阈值对样本熵较高的分量信号进行降噪处理、信噪比分析,发现与经验模态分解的结果相比较,基于变分模态分解和小波软阈值降噪后信号普遍具有较高的信噪比,由此证实该方法确实具有更好的降噪效果。Noise is a great disturbance in the analysis of vibration signals,which is an important part in fault diagnosis.In this paper,a denoising method based on Variational Mode Decomposition and soft Wavelet-thresholding is proposed.The new method works better than the EMD one in that it solves the EMD’s inherent defects like modal aliasing,endpoint effect and weak robotness to noises.A rotor test bed is used to generate signals in four different working conditions:misaligned rotor,unbalanced rotor,healthy rotor and rotor contact-rubbing.These signals are first decomposed by VMD and then,the high frequencies are denoised by soft wavelet-thresholding and then reconstructed with the remains.Lastly,signal to noise analysis is conducted to the reconstructed signals.The results show that the signal to noise ratio of VMD method basically doubles the EMD one.Therefore,it is confirmed that this method has a better effect on denoising than the EMD one.

关 键 词：信号降噪 变分模态分解 小波阈值 信噪比 振动信号 

分 类 号：TV734.1[水利工程—水利水电工程]