东北大学学报:自然科学版 ›› 2018, Vol. 39 ›› Issue (8): 1108-1113.DOI: 10.12068/j.issn.1005-3026.2018.08.009

• 信息与控制 • 上一篇    下一篇

基于EMD改进算法的欠定混合盲分离

季策, 孙梦雪, 张君   

  1. (东北大学 计算机科学与工程学院, 辽宁 沈阳110169)
  • 收稿日期:2017-04-07 修回日期:2017-04-07 出版日期:2018-08-15 发布日期:2018-09-12
  • 通讯作者: 季策
  • 作者简介:季策(1969-),女,辽宁沈阳人,东北大学副教授.冯明杰(1971-), 男, 河南禹州人, 东北大学副教授; 王恩刚(1962-), 男, 辽宁沈阳人, 东北大学教授,博士生导师.
  • 基金资助:
    国家自然科学基金资助项目(61370152,61671141,61501038); 沈阳市科技计划项目(F16-205-1-01).国家自然科学基金资助项目(51171041).

Underdetermined Blind Separation Based on Improved EMD Algorithm

JI Ce, SUN Meng-xue, ZHANG Jun   

  1. School of Computer Science & Engineering, Northeastern University, Shenyang 110169, China.
  • Received:2017-04-07 Revised:2017-04-07 Online:2018-08-15 Published:2018-09-12
  • Contact: SUN Meng-xue
  • About author:-
  • Supported by:
    -

摘要: 为改善拟合效果,针对经验模态分解(empirical mode decomposition,EMD)算法存在的端点效应,提出一种改进的EMD算法——端点极值延拓方法.利用改进的EMD算法对观测信号进行分解,将分解分量连同之前的观测信号构成新的观测信号,从而将欠定情况转化为超定情况,最后利用独立成分分析(independent component analysis,ICA)算法得到源信号的估计.通过仿真实验对比,证明了本文算法的有效性.

关键词: 欠定盲源分离, 经验模态分解, 端点效应, 极值延拓, 独立成分分析

Abstract: In order to improve the effect of data fitting, a method of extremum extension on endpoints is proposed, which is aimed at the endpoint effect of empirical mode decomposition (EMD) algorithm. The improved EMD algorithm is used to decompose the observed signals, and then the decomposed components together with the prior observed signals are regarded as new observed signals. Thus the underdetermined situation is changed into an overdetermined case. Finally, we use independent component analysis (ICA) algorithm to obtain the estimation of source signals. Simulation result shows that the proposed algorithm is effective.

Key words: underdetermined blind source separation, EMD(empirical mode decomposition), endpoint effect, extremum extension, ICA(independent component analysis)

中图分类号: