基于稀疏分量分析的欠定盲源分离用于表面肌电信号分解

Underdetermined Blind Source Separation Based on Sparse Component Analysis Applied in Decomposition of Surface Electromyography

目的解决表面肌电分解中电极数小于肌电源信号而产生的欠定问题,针对盲源分离求解欠定混合方程进行研究。方法首先采用匹配追踪(MP)算法将肌电信号稀疏化,再利用空间退化与Hough变换法估计聚类轴并求解混合矩阵,最后通过模板匹配法完成对运动单位动作电位(MUAP)波形的分类。结果从较少的观测信号中获得源信号的估计值,并得到MUAP的波形和发放间隔(IPI)信息。结论本文采用的方法对表面肌电信号的分解是有效的,且具有较好的分离效果。

Objective To achieve accurate information about motor unit（MU） basing on research of sparse component analysis（SCA）,by investigating a way for solving the problem of underdetermined mixed equations in the decomposition of surface electromyography（sEMG）.Methods First,matching pursuit（MP） algorithm was used to sparse the sEMG signal.Second,space degradation method and Hough transform were combined to estimate clustering axes and solve mixing matrix.At last,template matching technique was adopted to accomplish the MUAP waveform classification.Results The estimation of the source signal was got from less observed signals,and MUAP waveform template and iiter-pulse interval（IPI） were obtained too.Conclusion Experimental simulated results and real sEMG signals demonstrate that the method in this paper is effective in the decomposition of sEMG signal,and its separation results are satisfactory.