一种用于自适应噪声消除器的变步长改进最小均方（LMS）算法

[A modified least mean square (LMS) algorithm with variable step-size for an adaptive noise canceller].

作者信息

Gao Hui, Niu Cong-min, Wu Wei

机构信息

Institute of Space Medico-Engineering, Beijing, China.

出版信息

Space Med Med Eng (Beijing). 2002 Oct;15(5):366-8.

PMID:12449145

Abstract

Objective. To study a modified Least Mean [correction of Mear] Square (LMS) algorithm that can be applied in aerospace and aviation fields. Method. A modified LMS algorithm was proposed and step-size was calculated by estimated signal noise ratio (SNR) of input signal. Parameters of filter were adjusted automatically until optimal estimation of disturbed speech was obtained. Result. Output SNR of estimated speech was increased respectively to 18.2 dB, 22.1 dB and 25.2 dB, when input SNR of disturbed speech was -6 dB, 0 dB, 6 dB. Conclusion. The performance of the modified algorithm has obvious advantages over that of NLMS algorithm. The results may be advantageous to related application of adaptive noise canceller (ANC) in aerospace and aviation fields.

摘要

目的。研究一种可应用于航空航天领域的改进最小均方（LMS）算法。方法。提出一种改进的LMS算法，并根据输入信号的估计信噪比（SNR）计算步长。自动调整滤波器参数，直到获得受干扰语音的最佳估计。结果。当受干扰语音的输入SNR为-6 dB、0 dB、6 dB时，估计语音的输出SNR分别提高到18.2 dB、22.1 dB和25.2 dB。结论。改进算法的性能明显优于NLMS算法。研究结果可能有利于自适应噪声消除器（ANC）在航空航天领域的相关应用。