本文關(guān)鍵詞： 盲源分離 改進(jìn)EFICA算法 最速下降法　出處：《空軍預(yù)警學(xué)院學(xué)報(bào)》2016年06期 　論文類型：期刊論文

【摘要】：針對(duì)EFICA算法穩(wěn)定性會(huì)因隨機(jī)選擇初始迭代矩陣而受到影響的問(wèn)題,提出使用最速下降法選出合適的迭代初始矩陣對(duì)EFICA算法進(jìn)行了改進(jìn),給出了EFICA算法以及EFICA改進(jìn)算法與Fast ICA算法的對(duì)比實(shí)驗(yàn).仿真實(shí)驗(yàn)結(jié)果表明,對(duì)于通信信號(hào),EFICA改進(jìn)算法的平均串音誤差在數(shù)值上小于原算法,其方差也由原來(lái)的0.740 8降低到0.008 6,有效提高了算法穩(wěn)定性.
[Abstract]:Aiming at the problem that the stability of EFICA algorithm will be affected by random selection of initial iteration matrix, this paper proposes to use the steepest descent method to select the appropriate iterative initial matrix to improve the EFICA algorithm. The EFICA algorithm and the EFICA improved algorithm are compared with the Fast ICA algorithm. The simulation results show that the communication signal is obtained. The average crosstalk error of the improved EFICA algorithm is smaller than that of the original algorithm numerically, and its variance is reduced from 0.7408 to 0.008.6, which effectively improves the stability of the algorithm.
【作者單位】： 空軍預(yù)警學(xué)院;
【分類號(hào)】：TN911.7
【正文快照】： 盲源分離是在源信號(hào)與傳輸通道特性都未知的情況下,對(duì)所得到的觀測(cè)信號(hào)進(jìn)行分離得出源信號(hào)的估計(jì)方法[1].盲源分離技術(shù)目前已被廣泛運(yùn)用在各個(gè)領(lǐng)域[2-6].獨(dú)立分量分析通過(guò)從多維數(shù)據(jù)中尋找具有統(tǒng)計(jì)獨(dú)立和非高斯特征的分量對(duì)混合信號(hào)進(jìn)行分離,其中快速獨(dú)立分量分析,即Fast ICA

【相似文獻(xiàn)】

相關(guān)期刊論文 前1條

1 陶玉福;劉慶華;黃斌;;基于EFICA的混合語(yǔ)音盲分離時(shí)域算法[J];聲學(xué)與電子工程;2009年02期

，

本文編號(hào)：1486839