摘要
提出了一种自适应可变步长最小方差算法(λNVSS),以解决基本最小方差算法LMS中收敛速度和稳态误差之间的矛盾.该算法利用当前和过去共m(滤波器阶数)个误差信息,并通过引入修正系数ρ和遗忘因子λ(i),来确定下一步的迭代步长.文中对这种算法进行了分析和仿真验算.对比一般的变步长算法,λNVSS算法对于平稳过程中的滤波器能获得更快的收敛速度和更小的稳态误差,同时还具有较好的跟踪跃变系统的能力.
In this paper, a novel LMS （Least Mean Square） algorithm with adaptive variable step size, namely, A NVSS, is proposed to solve the trade-off between the misadjustment and the convergence speed. In the proposed algorithm, the iteration step size is determined according to the present and past m error signals and by introducing a modification coefficient ρ and a forgetting factor λ （i）. The algorithm is then analyzed and verified by simulation, with the results indicating that the λNVSS algorithm constitutes a significant improvement in the convergence speed with very small misadjustment in stationary environments and is of better tracking capability, as compared with the traditional algorithms with variable step size.
出处
《华南理工大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2006年第4期61-64,共4页
Journal of South China University of Technology(Natural Science Edition)
基金
广东省自然科学基金资助项目(020906)
广东省科技计划项目(A10502004)
关键词
最小方差算法
自适应滤波
变步长最小方差算法
遗忘因子
least mean square algorithm
adaptive filtration
variable step size least mean square algorithm
forgetting factor
