摘要
基于最注二乘准则Ｊ（ｎ）＝ ，利用最徒梯度下降法，得到一种新的梯度型自适应滤波算法．该算法避免了递推最小二乘ＲＬＳ（Ｒｅｃｕｒｓｉｖｅ Ｌｅａｓｔ Ｓｑｕａｒｅｓ）算法需递推估计更新自相关矩阵Ｒｘｘ（ｎ）的逆的不足，计算机模拟仿真结果表明该算法有良好的收敛性能，收敛速度快于ＬＭＳ（Ｌｅａｓｔ Ｍｅａｎ Ｓｑｕａｒｅｓ）算法、ＮＬＭＳ（ＮｏｒｍａｌｉｚｅｄＬｅａｓｔ Ｓｑｕａｒｅｓ）算法和ＲＬＳ算法．
This paper puts forward a new gradient-based adaptive filtering algorithm based on least squares criterion J(n) = by using the steepest decent gradient method. The algorithm avoids estimating inverse matrix of input signal self-correlation matrix Rxx(n). The results of computer simulation show this algorithm performance is better than those of other algorithms such as LMS algorithm、NLMS algorithm and RLS algorithm.
出处
《广州大学学报（综合版）》
2001年第2期32-34,共3页
Journal of Guangzhou University
关键词
自适应滤波算法
最小二乘准则
RLS算法
adaptive filtering algorithm
least square criterion
RLS algorithm
