摘要
最小均方(LMS)自适应滤波算法易于实现,并在很多领域得到了广泛的应用,但它存在着难于同时获取较快的收敛速度与较小的稳态误差这一矛盾,而固定步长LMS算法无法解决。基于这一矛盾,在此提出一种新的变步长LMS自适应滤波算法,用最佳Wiener解和反正切函数建立了步长因子与误差之间一种新的非线性函数关系,该算法具有良好的收敛性能。并通过DSP验证,结果表明该方法实现简单,不依赖模型,具有较强的稳健性。
LMS algorithm can be easily realized and applied in many fields, but there is an inconsistency between the fast algorithm convergence and the low steady state error, fixed step size I.MS algorithm can not resolve this problem. On the basis of the problem, an improved LMS adaptive algorithm is proposed, which established a new non-linear functional relationship be- tween μ（n） and e（n） use arc tangent function and the best Wiener result,which is easily controlled, And the flit method has been tested and adopted in DSP. Experiments show that the realization method is simple and doesn＇t rely on model and has strong stability.
出处
《现代电子技术》
2009年第24期103-105,共3页
Modern Electronics Technique
