改进的变步长LMS自适应滤波算法  

Improved variable step size LMS adaptive filtering algorithm

作　　者：刘小东[1] 黄洪琼[1] 

机构地区：[1]上海海事大学信息工程学院,上海201306

出　　处：《舰船科学技术》2015年第10期115-118,共4页Ship Science and Technology

基　　金：国家自然科学基金资助项目(61271446)

摘　　要：针对解决LMS(Least Mean Square)自适应滤波算法收敛速度及未知系统时变跟踪速度与稳态误差的矛盾,改进步长因子μ(n)与误差信号e(n)的非线性映射关系,提出一种新的变步长LMS算法。执行该算法时系统初始阶段或未知系统时变阶段步长自动增大,而稳态时步长缓慢变小,提高了收敛速度和时变跟踪能力,克服了稳态误差偏大的缺点。理论分析及实验结果表明,新算法的收敛、跟踪速度及稳态误差性能均优于现有常见的几种LMS算法。For solving the LMS （Least Mean Square） adaptive filtering algorithm convergence speed and the unknown system change tracking speed and steady - state error of contradictions. Improved step factor/z（n） and the error signal e（n） of the nonlinear relationship, a new variable step size LMS algorithm is given. Executing the algorithm , the step size increases automatically at the beginning or unknown system is changing with time, while it becomes smaller during the steady state, which improved convergence speed and time-varying tracking capability, and it also avoid the shortage of step sizeμ（n） is too large during the steady state. The theoretical analysis and experimental results show that the convergence, tracking speed and steady state error performance of the new algorithm is superior to existing common types of LMS algorithm.

关 键 词：自适应滤波 变步长 最小均方 收敛速度 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]