LS-SVM与小波指数阈值结合的振动信号降噪  

Vibration Signal Denoising Method Combining LS-SVM and Wavelet Index Threshold

作　　者：张亚靓 林郁 纪俊卿 孟祥川 许同乐[1] ZHANG Ya-jing;LIN Yu;JI Jun-qing;MENG Xiang-chuan;XU Tong-le(Shandong University of Technology,College of Mechanical Engineering,Zibo,Shandong 255000;Shandong Energy Longkou Mining Group Co.,Ltd.,Yantai,Shangdong 265700)

机构地区：[1]山东理工大学机械工程学院,山东淄博255000 [2]山东能源龙口矿业集团有限公司,山东烟台265700

出　　处：《液压与气动》2020年第3期71-75,共5页Chinese Hydraulics & Pneumatics

基　　金：国家自然科学基金(51465009);山东省自然科学基金(ZR2016EEM20);山东省高等学校科技计划(J10LG22)

摘　　要：为了能够更好的保持液压泵轴承振动信号的峰形,解决软阈值函数存在的恒定偏差和硬阈值函数存在间断点的问题,引进了一种最小二乘支持向量机(LS-SVM)与小波指数阈值结合的振动信号降噪方法。首先对液压泵的机械轴承振动信号进行小波阈值分解,进而采用LS-SVM方法将小波系数分为与噪声相关的及与噪声无关的小波系数,采用小波指数阈值函数将与噪声相关的小波系数滤除,将处理后的信号重组,提取微弱的振动信号特征。不仅实现了振动信号的降噪,也利于轴承的故障诊断。In order to better maintain the peak shape of the piston pump bearing vibration signal,to solve the problem of the constant deviation of the soft threshold function and the discontinuous point of the hard threshold function,this paper introduces a vibration signal combining LS-SVM and wavelet index threshold.Noise reduction method.Firstly,wavelet threshold decomposition is performed on the mechanical bearing vibration signal of the plunger pump.Then the wavelet coefficient is divided into noise-related and noise-independent wavelet coefficients by LS-SVM method.The wavelet-like threshold function is used to correlate the noise-related wavelet coefficients.Filtering,recombining the processed signal,extracting weak vibration signal characteristics.This method not only realizes the noise reduction of the vibration signal,but also beneficial to the fault diagnosis of the bearing.

关 键 词：软硬阈值函数 小波指数阈值函数 信号降噪 

分 类 号：TH137[机械工程—机械制造及自动化]