【作者】 高鹰； 谢胜利；

【Author】 Gao Ying Xie Shengli (Dept. of Electron. and Commun. Eng., South China University of Technology, Guangzhou, 510641,China) (Dept. of Computer Science and Technology., Guangzhou University, Guangzhou.510405. China)

【机构】 华南理工大学电子与通信工程系；

【摘要】 本文先对LMS牛顿自适应滤波算法中的梯度矢量用当前时刻的梯度估计来代替前一时刻的梯度估计,然后利用矩阵求逆定理给出了LMS牛顿自适应滤波算法的一种修正形式。对该修正形式算法的收敛性的理论分析表明算法对步长因子μ>0是绝对收敛的,计算机数值仿真显示该算法的收敛性能好于LMS牛顿自适应滤波算法的收敛性能。更多还原

【Abstract】 In this paper, a modified LMS Newton algorithm is proposed by replacing the previous gradient vector with the current gradient vector in the LMS Newton algorithm and using the matrix inversion lemma. The results of theoretical analysis indicate that the modified LMS Newton algorithm is stabile for step size factor μ > 0 .Computer simulation results show that the proposed algorithm’s convergence performance is better than that of the LMS Newton algorithm.更多还原