摘要
针对频域盲分离算法中存在的排序不确定性问题,提出一种基于K-means算法的排序方法。该方法在排序中使用K-means算法对分离子信号进行整体排序,与传统相邻频点排序方法相比,大大提高了盲源分离的鲁棒性。仿真结果表明了该方法的有效性。
For the ambiguous permutation in the frequency-domain blind source separation （BSS）, this paper presents a method for improving permutation. Compared with the traditional method with which to deal with permutation problem, the method based on K-means algorithm greatly improves the ro- bustness and accuracy of BSS. The simulation results show that this method is effective.
出处
《海军工程大学学报》
CAS
北大核心
2013年第3期1-6,共6页
Journal of Naval University of Engineering
基金
国家自然科学基金资助项目(50979109)
关键词
盲源分离
幅度相关
排序
频域
blind source separation~ amplitude correlation~ permutation~ frequency-domain
