一种改进的变步长自适应滤波算法  

An Improved Variable Step-Size Adaptive Filtering Algorithm

作　　者：赵光丛[1] 石玉[1] 

机构地区：[1]南京航空航天大学自动化学院,江苏南京210016

出　　处：《自动化技术与应用》2013年第6期32-36,共5页Techniques of Automation and Applications

摘　　要：为了提高LMS自适应滤波算法的性能,在对传统LMS算法及其改进算法研究的基础上,提出了一种改进的变步长算法。在改进算法中,步长因子与误差信号自相关函数之间建立了一种改进的非线性函数关系。将改进算法应用到系统辨识中,通过计算机仿真结果看出,自适应滤波性能在收敛速度和稳态失调误差等方面得到改善。In order to improve the performance of the variable size-step LMS （least meansquare） adaptive filtering algorithm, on the basis of the traditional LMS algorithm and its improved algorithms,a new variable step size algorithm is proposed. In the new algorithm, a new nonlinear relationship is established between the step size factor and the error signal＇s autocorrelation functions. This paper applies the new algorithm to the system iden-tification, through the computer simulations, the adaptive filter performance is improved in convergence speed and steady-state offset errors etc.

关 键 词：LMS自适应滤波算法 变步长 系统识别 滤波性能 

分 类 号：TN911.7[电子电信—通信与信息系统]