| 基于FIR神经网络的非线性盲信号分离 |
| |
| 引用本文： | 虞晓,胡光锐. 基于FIR神经网络的非线性盲信号分离[J]. 上海交通大学学报, 1999, 33(9): 1093-1096 |
| |
| 作者姓名： | 虞晓 胡光锐 |
| |
| 作者单位： | 上海交通大学电子工程系上海 200030 |
| |
| 摘 要： | 针对实验环境中可能出现的非线性卷积混合盲信号分离问题，在反馈结构最大似然盲信号分离算法的基础上，利用混合高斯模式概率密度函数估计方法，提出了一种采用有限冲激响应神经网络的非线性盲分离算法，并推导了新算法的权向量迭代公式．通过与其他盲信号分离算法的计算机模拟实验结果比较，新算法能更有效地进行非线性函数逼近，得到更小的输出均方误差，达到较好的非线性盲信号分离效果．
|
| 关 键 词： | 非线性盲信号分离 有限冲激响应神经网络 混合高斯模型 |
| 文章编号： | 1006-2467(1999)09-1093-04 |
| 修稿时间： | 1998-12-09 |
FIR Neural Network Based Nonlinear Blind Signal Separation |
| |
| YUXiao,HUGuang rui Dept. ofElectronic Eng.,ShanghaiJiaotong Univ.,Shanghai,China. FIR Neural Network Based Nonlinear Blind Signal Separation[J]. Journal of Shanghai Jiaotong University, 1999, 33(9): 1093-1096 |
| |
| Authors: | YUXiao HUGuang rui Dept. ofElectronic Eng. ShanghaiJiaotong Univ. Shanghai China |
| |
| Affiliation: | YUXiao,HUGuang rui Dept. ofElectronic Eng.,ShanghaiJiaotong Univ.,Shanghai200030,China |
| |
| Abstract: | Based on the feedback architecture and the m axim um likelihood blind signalseparation(BSS) algorithm principle, a new nonlinear BSSseparation algorithm w as proposed by using the FIR neural network and the Gaussian m ixture m odel(GMM) pdf estim ation. The corresponding iterative form ulas w ere deduced. Based on the com parison ofthe com puter sim ulation results, it can be concluded that the proposed algorithm has lower output MSE and better nonlinear blind signal separation effect than the traditionalBSSalgorithm . |
| |
| Keywords: | nonlinear blind signalseparation finite im pulse response(FIR) neuralnetwork Gaussian m ixture m odel(GMM) |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
