新型变阶LMS自适应算法

Novel variable tap-length LMS adaptive algorithm

详细介绍了现有成熟的基于LMS(最小均方)变阶长算法,提出了更适合实际应用高效的最优变阶算法。在变阶长的研究中FT-LMS(分数阶长LMS算法)是目前最高效的算法,本算法继续保持分数阶长变化量累积到一定数值再对滤波器的阶数进行更新的优势,通过对误差进行统计算法分析来得到稳健的更新阶长,使算法中参数更少且选择阈度更高。此算法经MATLAB仿真软件验证,其收敛速度快,最优阶长的跟踪性能好。

This paper introduced the existing mature length of variable order based on the LMS（ least mean square） algorithm,then put forward more suitable for practical application of efficient optimal variable step algorithm. In the study of variable order length,FT-LMS（ fractional LMS algorithm） is the most efficient algorithm. The algorithm maintained the advantage of updating the order of the tap when the number of fractional accumulated to a certain value,through statistical algorithm analysis for error to get an steady update tap,it made fewer parameters to select and had an higher threshold to choose. Verified the algorithm by MATLAB simulation software,its convergence speed is fast and show the good tracking performance for optimal tap-length.