文档介绍：2013年4月噪声与振动控制第2期
文章编号:1006-1355(2013)02-0152-05
一种EMD阈值寻优降噪方法
黄建招,谢建,李良,许建国
(第二炮兵工程大学兵器发射理论与技术国家重点学科实验室,西安 710025)
摘要:为提高 EMD 方法的降噪效果,针对基于能量直接提取法中跳变点定量指标的确定和奇异本征模态函数的
处理问题,提出一种基于经验模态分解的阈值寻优降噪方法。方法首先从能量角度给出确定跳变点位置的判定标准,
选取符合标准的本征模态函数。然后通过将所有被剔除本征模态函数的平均能量和已选取的各本征模态函数能量进
行比较,确定奇异本征模态函数并对其进行阈值降噪处理,最后通过对已选取本征模态函数求和得到降噪信号。方法
将跳变点的选取和软阈值消噪处理相结合,与其他三种降噪方法进行降噪对比试验,证明方法的有效性和优越性。试
验结果为将该方法应用于工程实际,进一步提高信号降噪水平提供支撑。
关键词:声学;EMD;阈值降噪;IMF选取;奇异模态
中图分类号:;TH137 文献标识码:A DOI编码:.1006-
EMD Based Threshold-optimizing De-noising Method
HUANG Jian-zhao , XIE Jian , LI Liang , XU Jian-guo
( State Key Subject Lab of Weapon Launching Theory and Technology,
The Second Artillery Engineering University, Xi’an 710025, China )
Abstract : The EMD-based de-noising methods can be classified into direct extraction method and threshold method.
Aiming at the problems of the sharp-change point determination in energy-based direct extraction method and the processing
of the singular IMFs, a threshold-optimizing de-noising method based on EMD was put forward. Firstly, the criterion of
sharp change point determination from energy viewpoint was offered, and the IMFs up to the criterion were selected.
Secondly, paring the average energy of all eliminated IMFs with the energy of each selected IMF, the singular
IMFs were confirmed and de-noised