摘要
源信号数目较多的混合信号盲源分离中,快速独立分量分析FastICA算法存在迭代次数过高、收敛速度慢、分离性能恶化等问题,为此,提出了一种基于有理多项式非线性函数的FastICA算法N-FastICA,并通过实验分析了有理多项式函数的阶数和系数选取。仿真结果表明,NFastICA算法的PI值小于FastICA算法的PI值,且都接近于0,分离性能优于FastICA算法。随着信号数目的增多,由相似系数矩阵判断N-FastICA算法能成功分离源信号,而FastICA算法已不能成功分离出源信号,前者迭代次数远少于后者。
Fast independent component analysis(FastICA)algorithm has many problems such as high number of iterations,slow convergence rate and poor separation performance in mixed-signal blind source separation with a large number of source signals.To improve the performance of FastICA,one FastICA algorithm based on rational polynomial nonlinear function(N-FastICA)is proposed.The order and coefficients selection of the rational polynomial function is analyzed experimentally.The results of computer simulation show that the PI values of N-FastICA algorithm is smaller than FastICA algorithm and closing to 0,which means the separation performance of N-FastICA algorithm superior to FastICA algorithm.With the increase of the number of signals,the similarity coefficient matrix shows the N-FastICA algorithm can successfully separate the source signal,and the FastICA algorithm can not successfully separate the source signal.The former iterations number is far less than the latter.
作者
赵知劲
黄艳波
强芳芳
ZHAO Zhijin;HUANG Yanbo;QIANG Fangfang(School of Communication Engineering,Hangzhou Dianzi University,Hangzhou Zhejiang 310018,China;School of Electronic Information,Hangzhou Dianzi University,Hangzhou Zhejiang 310018,China)
关键词
盲源分离
快速独立分量分析
非线性函数
有理多项式
blind source separation
fast independent component analysis
nonlinear function
rational polynomial
