摘要
归一化LMS算法是用步长与输入信号功率的比值,对步长的归一化,但其使用的全局步长因子是固定的,会出现不能同时兼顾收敛速度和稳态误差的问题。为了解决这一问题,将该算法与基于对数函数的变步长LMS算法相结合,提出了一种基于对数函数的归一化变步长LMS算法。基于对数函数的归一化变步长LMS算法是利用步长μ和误差e满足的一种归一化的对数关系,通过误差e来调整步长μ,使得步长μ始终在一个合适的范围内。仿真结果表明,新算法在收敛速度和稳态误差方面都优于归一化LMS算法和基于对数函数的变步长LMS算法。
The normalized LMS algorithm is the normalization of step size by the ratio of step size to input signal power,but the used global step size factor is fixed,and the convergence rate and steadystate error can not be considered simultaneously.In order to solve this problem,the algorithm is combined with the variable step size LMS algorithm based on logarithmic function,and a new normalized variable step size LMS algorithm based on logarithmic function is proposed.The normalized step size LMS algorithm based on logarithmic function takes the step size滋and the error e to satisfy a normalized logarithmic relation,and adjusts the step size滋by the error e,so that the step size滋is always in a suitable range.The computer simulation results show that this new algorithm is better than normalized LMS algorithm and variable step size LMS algorithm based on logarithmic function in terms of convergence speed and steady-state error.
作者
明乐
周峰
MING Le;ZHOU Feng(School of Air and Missile Defense,Air Force Engineering University,Xi’an 710051,China)
出处
《火力与指挥控制》
CSCD
北大核心
2018年第8期173-175,180,共4页
Fire Control & Command Control
关键词
LMS算法
归一化
对数函数
收敛速度
稳态误差
LMS algorithm
normalization
logarithmic function
convergence speed
steady-state error
