汽车车内噪声主动控制变步长NFB-LMS算法

为规避最小均方（Least Mean Square，LMS）算法不能同时提高收敛速度和降低稳态误差的固有缺陷，以及已有变步长LMS算法存在收敛速度慢和稳态误差估计精度差的问题，文中提出了一种基于变步长归一化频域块（Normalized Frequency-domain Block，NFB） LMS算法的汽车车内噪声主动控制方法。为了比较，应用传统的LMS算法、基于反正切函数的变步长LMS算法和变步长NFB-LMS算法分别进行实测汽车车内噪声的主动控制。结果表明，与其他两个算法相比，变步长NFB-LMS算法的收敛速度提高了70%以上，稳态误差减小了90%以上。变步长NFB-LMS算法在处理车内噪声信号时具有很高的效率，为进行汽车车内噪声主动控制提供了一种新方法。

A variable step-size NFB-LMS algorithm for active vehicle interior noise control

The LMS algorithm has an inherent shortcoming that the convergence speed can not be increased simultaneously with reducing the steady-state error. For the existing variable step-size LMS algorithm the convergence rate is low and the accuracy of estimating steady-state residual error is poor. To avoid such disadvantages, an active control method of vehicle interior noise based on variable step-size NFB-LMS algorithm is presented in this paper. The traditional LMS algorithm, the variable step-size LMS algorithm based on arctangent function and the variable step-size NFB-LMS algorithm are respectively applied to the active control experiments of the measured vehicle interior noise for comparison. The results show that the convergence speed of the variable step-size NFB-LMS algorithm is increased by 70% and the steady-state error is reduced by more than 90%, compared with the other two algorithms. Therefore, the variable step-size NFB-LMS algorithm has high efficiency in processing the vehicle interior noise signals, and provides a new method for active control of vehicle interior noise.