黑龙江大学自然科学学报
黑龍江大學自然科學學報
흑룡강대학자연과학학보
JOURNAL OF NATURAL SCIENCE OF HEILONGJIANG UNIVERSITY
2010年
1期
135-140
,共6页
盲分离%α稳定分布%非线性%几何线性化
盲分離%α穩定分佈%非線性%幾何線性化
맹분리%α은정분포%비선성%궤하선성화
blind separation%alpha-stable%nonlinear%Geometric linearization
在α稳定分布模型基础上提出一种非线性的盲分离算法,采用后非线性模型(PNL)几何线性化算法将输入信号线性化,然后用基于最小分散系数与旋转变换(MDC-RT)准则的神经网络算法实现信号盲分离.仿真实验证明该算法具有良好的分离特性以及较高的实际意义.
在α穩定分佈模型基礎上提齣一種非線性的盲分離算法,採用後非線性模型(PNL)幾何線性化算法將輸入信號線性化,然後用基于最小分散繫數與鏇轉變換(MDC-RT)準則的神經網絡算法實現信號盲分離.倣真實驗證明該算法具有良好的分離特性以及較高的實際意義.
재α은정분포모형기출상제출일충비선성적맹분리산법,채용후비선성모형(PNL)궤하선성화산법장수입신호선성화,연후용기우최소분산계수여선전변환(MDC-RT)준칙적신경망락산법실현신호맹분리.방진실험증명해산법구유량호적분리특성이급교고적실제의의.
A nonlinear BSS algorithm based on alpha-stable model is put forward in the composition. The geometric linearization algorithm based on PNL was used firstly, then the neural network algortihm based on MDC-RT was used to separate the signals. Experiment proves that this algorithm has a good separate character and has a high practical significance.