收稿日期:2009 -10 -08 修回日期 2009 -11 -25 第 27 卷 第 11 期 计 算 机 仿 真 2010 年 11 月 文章编号:1006 -9348( 2010) 11 -0359 -04 基于箕舌线变步长 LMS 算法的分析与改进 胡春娇,杨 顺 ( 辽宁工程技术大学,辽宁 葫芦岛 125105) 摘要: 研究系统跟踪性能和误差问题,为了改善基于箕舌线的变步长最小均方( LMS) 算法的收敛速度和对信号输入端不相关噪声的抗干扰性,首先对现有的典型变步长 LMS 算法进行了分析归纳。在箕舌线变步长 LMS 算法( TCLMS) 的基础上通过结合其他典型算法的优点,提出了 TCLMS 算法的改进算法。算法不仅继承了 TCLMS 算法计算复杂度低的优点,并且通过引入动量项与前后误差的自相关估计对其收敛速度慢,抗不相关噪声干扰性能差的特性进行了改善。在 MATLAB 进行仿真,结果表明,改进算法具有收敛速度快,计算复杂度低,稳态误差小,抗噪声能力强等特点。关键词: 关键词: 变步长; 最小均方算法; 箕舌线; 动量项中图分类号: TN911. 4 文献标识码: A Analysis and Improvement of a Variable Step LMS Algorithm Based on Correlation Tongue - Like Curve HU Chun - jiao,YANG Shun ( Liaoning Technical University,Huludao Liaoning 125105,China) ABSTRACT: In order to improve the convergence rate and anti - noise performance of variable step size LMS algo-rithm based on correlation tongue - like curve,this paper analyzes some typical variable step adaptive LMS algo-rithms. Through combining the advantages of some other typical algorithms,an improved algorithm which is a new variable step adaptive algorithm based on tongue - like curve( TCLMS) is proposed. The new algorithm not only in-herits the advantages of TCLMS algorithm,but also uses the estimation of the autocorrelation about adjacent error sig-nals and momentum to speed up the convergence rate and improve the anti - noise performance. MATLAB simulation results show that the new algorithm has fast convergence rate,low computational complexity and the steady - state er-ror or strong anti -