基于SVD降噪的机械振动信号经验模式分解及其应用

SVD noise control based EMD of mechanical vibration signal and its application

提出了1种基于奇异值分解降噪的振动信号经验模式分解方法,利用该法对原始振动信号进行相空间重构和奇异值分解,并根据奇异值降噪,再利用经验模式分解法提取降噪后振动信号的基本模式分量,对滤波前和滤波后振动信号进行经验模式分解分析。分析结果表明:奇异值分解能有效地提高信噪比,突出原始振动信号的特征,使降噪后振动信号分解出的基本模式分量具有更明确的物理意义,有利于对柴油机进行精确诊断。

EMD method is used to conduct phase space rebuilding and SVD of origin vibration signal to control noise according to the strange value. The IMFs of the signal obtained by SVD are extracted to perform END of the signal before and after wave filtering. Analysis shows that SVD method can make signal - to - noise ratio improved, original signal character revealed well and the IMFs of the signal obtained by SVD has cXcarer pnysical meaning, and this contributes significantly to accurate diesel engine＇s diagnosis.