基于ICA的白噪声消除方法研究  

Method for eliminating white noise based on ICA

作　　者：宋佳[1] 邱天爽[1] 

机构地区：[1]大连理工大学电子信息与电气工程学部,辽宁大连116024

出　　处：《计算机工程与设计》2012年第2期782-786,共5页Computer Engineering and Design

基　　金：国家自然科学基金项目(60872122)

摘　　要：为了消除润滑油内金属磨粒检测系统(metal debris detection system,MDDS)输出信号中混合的高斯白噪声,提出了一个基于ICA的算法对两路输出信号进行消噪处理。对两路信号添加前缀信号,并按照所述步骤进行两次ICA后得到三路独立源信号,根据ICA前后前缀信号幅值和相位的变化校正ICA分离结果的幅值和相位,完全恢复源信号。对MDDS的输出信号进行仿真以验证算法的去噪效果,实验结果表明,该算法可以有效地消除输出信号中的白噪声。An algorithm based on ICA is presented,which is used to eliminate Gaussian white noise in the output signals of the metal debris detection system（MDDS）.First,add the prefixs for two signals and composite them by ICA twice according to the steps.With this method,three independent source signals are obtained.Then the changes of the prefix signals＇ amplitude and phase after ICA are used to correct the amplitude and phase of the separation results and source signals are completely recovered finally.The two output signals of MDDS is simulated to verify the algorithm.Experimental results show that the proposed algorithm can effectively eliminate the Gaussian white noise in the output signals.

关 键 词：独立分量分析 盲源分离 源识别或重建 白噪声 噪声消除 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]