一种新的变步长ELMS算法在噪声抵消中的应用

A new Mixed Variable Step Size ELMS algorithm and its application in ANC

摘要:

为了寻求高效快速的自适应算法，在ELMS算法基础上，提出了一种用均方误差和误差的相关性来调节步长的混合变步长ELMS(MVSS-ELMS)算法。该算法符合步长调整原则，并在抗噪性、有效性方面有了很大改善，同时具有比传统的LMS,ELMS算法收敛速度快，稳态失调小等优点。计算机仿真结果表明，新算法在自适应噪声抵消中的综合性能优于LMS及ELMS算法。

Abstract:

In order to find an efficient adaptive algorithm，a new Mixed Variable Step Size ELMS algorithm(MVSS-ELMS) based on Extended Least Mean Square(ELMS) algorithm is proposed in this paper. This new algorithm combines the mean square error and the correlation of the error to modify the step size. The new algorithm has been improved in anti-noise capacity and effectiveness for it fitting the principal of adjusting step-size；meanwhile，it has the superiority in convergence rate and steady-state compared to Least Mean Square(LMS) and ELMS. Computer simulations indicate that the performance of the new algorithm is better than that of LMS and ELMS on Adaptive Noise Cancellation(ANC).