FastICA算法对非高斯混合信号分离的性能分析  

FastICA algorithm for non-Gaussian mixture signal separation performance analysis

作　　者：房宏威[1] 李良敏[1] 安占飞[1] 贾燕红 

机构地区：[1]长安大学,陕西西安710064

出　　处：《汽车实用技术》2013年第7期48-51,共4页Automobile Applied Technology

摘　　要：基于独立分量分析的盲源分离方法,采用快速独立分量分析(Fast Independent ComponentAnalysis,FastICA)的算法对混合的非高斯信号进行分离.通过对仿真信号的分离比较,使用峭度值和信噪比对分离函数tanh和gauss函数的性能进行了分析,研究分离函数的选择对于分离效果的影响。The blind source separation based on independent component analysis is performed by means of FastlCA algorithms, with which the non-gussian signals will be separated. Through the separation and comparison of simulation signal,analysising the performance of the separation function ＇tanh＇ and ＇gauss＇ by using the Kurtosis information and SNR, and also discuss the selection of nonlinear functions on separation effect.

关 键 词：盲源分离 独立分量分析 超高斯 亚高斯 快速独立分量分析 

分 类 号：TN911.23[电子电信—通信与信息系统]