一种基于改进双曲正切函数的变步长自适应滤波算法  

Variable step size adaptive filtering algorithm based on modified hyperbolic tangent function

作　　者：郭瑞[1] 李宝华[1] 马奥运 

机构地区：[1]辽宁工程技术大学电气与控制工程学院,辽宁葫芦岛125105

出　　处：《计算机应用研究》2015年第8期2335-2338,共4页Application Research of Computers

摘　　要：为了改进现有变步长LMS(least mean square)算法性能方面存在的缺陷,提出一种改进的变步长谐波检测算法。该算法在原有双曲正切函数的基础上引入包含输入信号的因子T(n),跟踪输入信号变化,以便分析算法性能,提高其抗干扰能力,并采用增加补偿项来确保算法的收敛速度;同时将步长迭代公式中固定约束范围转变为动态范围,使步长变化相对平滑,稳态失调相对较小;最后利用归一化的处理方法改进权值公式,增大输入信号的动态范围。仿真结果表明,新算法在收敛速度、跟踪能力、抗干扰能力、稳态误差等方面较现有的变步长谐波检测算法有较大提高,是一种可行、有效、具有一定工程应用价值的算法。In order to improve the disadvantages of the existing variable step size LMS （ least mean square） algorithm in terms of performance, this paper proposed an improved variable step size adaptive harmonic detection algorithm. The algorithm was based on the original hyperbolic tangent function and introduced factor T（n） that contained the input signal, tracked the changes of input signal, in order to analyze the performance of the algorithm to improve its anti-jamming capability. And it added compensation term to ensure the algorithm convergence speed. At the same time it replaced the fixed step size range restriction of step iterative formula by dynamic change restriction, which made the step change tend to be smoother and steady-state misadjustment was relatively small. Finally, it used normalization method to update weight formula to improve the dynamic range of the input signal. The results of simulation show that the new algorithm compares to the existing variable step harmonic detection algorithm improves greatly in terms of convergence speed, tracking capability, anti-interference ability, stable error . It is a feasible, effective, and has certain engineering application value algorithm.

关 键 词：变步长LMS算法 双曲正切函数 抗扰能力 归一化 稳态失调 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]