摘要
通过分析谱图小波变换在平面图像、三维实体分析中的应用,提出一种用于一维数字信号分析的谱图小波阈值降噪方法。该方法将一维数字信号定义到路图上,利用谱图小波变换将其分解成尺度系数和谱图小波系数,对谱图小波系数进行阈值过滤处理,再进行谱图小波逆变换得到降噪信号。首先,利用四种典型仿真信号进行降噪试验,并分析不同分解层数对降噪性能的影响;接着,将其与经典小波阈值降噪方法进行仿真对比;最后,采用该方法进行滚刀主轴振动信号降噪,并与经典小波阈值降噪方法对比。仿真及试验结果表明,该方法实现了一维数字信号的快速非迭代降噪,且降噪信号平滑度高、畸变小,优于经典小波阈值降噪方法。
By analyzing the application of spectral graph wavelet transform in plane image and three-dimensional solid analysis, a spectral graph wavelet threshold denoising method is proposed for one-dimensional digital signal analysis. In this method, the one-dimensional digital signal is defined on the path graph, which is decomposed into scale coefficients and spectral graph wavelet coefficients by spectral graph wavelet transform, and the spectral graph wavelet coefficients are filtered by threshold. Thus the denoising signal is obtained by spectral graph wavelet inverse transform. Firstly, four typical simulation signals are used for denoising test, and the influence of different decomposition layers on denoising performance is analyzed. Then, the performance of the proposed method is compared with that of the classical wavelet threshold denoising method by simulation. Finally, the method is applied to the denoising experiment of hob spindle vibration signal, and compared with the classical wavelet threshold denoising method. Simulation and experimental results show that the proposed method can realize the fast non-iterative denoising of one-dimensional digital signal, and the denoising signal has high smoothness and small distortion. Also, it is obviously superior to the classical wavelet threshold denoising method.
作者
董鑫
李国龙
何坤
贾亚超
徐凯
李彪
DONG Xin;LI Guolong;HE Kun;JIA Yachao;XU Kai;LI Biao(State Key Laboratory of Mechanical Transmission,Chongqing University,Chongqing 400030;College of Mechanical Engineering,Chongqing Technology and Business University,Chongqing 400067)
出处
《机械工程学报》
EI
CAS
CSCD
北大核心
2020年第11期96-107,共12页
Journal of Mechanical Engineering
基金
国家自然科学基金资助项目(51875066)
重庆市基础研究与前沿探索(cstc2018jcyjAX0578)资助项目。
关键词
谱图小波变换
阈值
降噪
滚刀主轴
振动信号
spectral graph wavelet transform
threshold
denoising
hob spindle
vibration signal
作者简介
董鑫,男,1991年出生,博士研究生。主要研究方向为智能制造技术与系统,刀具故障预测与健康管理。E-mail:xindong0531@foxmail.com;通信作者:李国龙,男,1968年出生,博士,教授,博士研究生导师。主要研究方向为智能制造技术与装备,精密与超精密加工技术。E-mail:glli@cqu.edu.cn。
