摘要： 经验模态分解被广泛用于复杂的非线性及非平稳信号处理,但其分解过程中存在端点效应问题,并且端点效应会随着筛选的进行而更加明显,从而导致数据处理发生畸变。镜像延拓法和窗函数法均是抑制端点效应的有效方法,但也各自存在不足。为此,提出将镜像延拓和余弦窗函数结合的改进方法,以弥补镜像延拓法中原始信号端点不是极值点和窗函数法中需要改变原信号的不足,实现对端点效应抑制方法的改进。仿真实验结果表明,与传统方法相比,改进方法能更好地抑制端点效应的发生。

关键词: Hilbert-Huang 变换, 经验模态分解, 端点效应, 镜像延拓, 窗函数, 均方根

Abstract: Empirical Mode Decomposition(EMD) is widespread used in complex signal processing,but it exists the end effect in decomposition processing,and the end effect will further deepen with the process of filtration. Thus,it will cause data error. The methods of mirror extension and window function are effective methods for dealing with the end effect,but they are flawed. This paper presents a lifting method combining mirror extension and cosine window function. It makes up the deficiencies of mirror extension in original signal endpoint being not extreme value point and window function method in the need to change the original signal. Experimental result manifests that this lifting method can better restrain the end effect compared with some traditional methods.

Key words: Hilbert-Huang Transform(HHT), Empirical Mode Decomposition(EMD), end effect, mirror extension, window function, Root-Mean-Square(RMS)

中图分类号: 

• TN911. 7