Abstract:

A modified CFxLMS(Correlation FxLMS) algorithm is proposed in this paper, which makes use of the correlation function of the filtered reference-error signal to control the step size update. The step size decreases along with the decrease of the correlation function, thereby improves algorithm real-time capability, and performs better to overcome the inherent contradiction of the FxLMS algorithm between convergence speed and stable error. For further verification, the CFxLMS algorithm is applied in high speed elevator cabin active noise control. Simulation results are also compared with the results of the FxLMS algorithm.

摘要：

本文提出了一种改进的CFxLMS(Correlation FxLMS)算法，该算法使用滤波参考信号与误差信号的相关函数控制步长更新。即随着相关函数的减小，步长也逐渐减小，从而提高了算法的实时性，较好地解决了传统FxLMS算法因步长固定带来的稳态误差大和实时跟踪能力弱的问题。为进一步验证改进算法的性能，本文将CFxLMS算法应用于高速电梯的轿厢内噪声主动降噪仿真中，并与FxLMS算法的结果进行了对比。

关键词: 声学, 主动噪声控制, 变步长, CorrelationFxLMS(CFxLMS)算法, 电梯噪声, 仿真计算