车内道路噪声有源控制的虚拟参考信号分组构建方法

Method of Constructing Virtual Reference Signals Utilizing Grouping for Active Road Noise Control in Vehicles

摘要:

针对道路噪声有源控制系统参考信号数量较多导致系统运算负担较大，而直接构建的虚拟参考信号与车内目标噪声相干度不高的问题，提出了一种虚拟参考信号的分组构建方法。首先进行整车道路试验，选取4个车轮位置附近采集到的和车内噪声相干性较高的振动加速度信号作为原始信号，基于奇异值分析法确定构建的虚拟参考信号数量。然后基于车轮位置将关联度较高的原始信号分为一组，对每组原始信号通过奇异值分解法构建转换矩阵，分组构建虚拟参考信号。最后搭建基于归一化参考信号滤波最小均方误差（NFXLMS）算法的道路噪声有源控制模型，分析分组构建的虚拟参考信号在降低计算复杂度与降噪效果方面的表现。与原有虚拟参考信号构建方法相比，该方法在降低计算复杂度和提高降噪效果方面均有较大改善，验证了该方法的有效性。

Abstract:

Aimed at the problem that the large number of reference signals for the road noise active control system leads to a heavy computational burden， and the directly constructed virtual reference signals are not highly coherent with the target noise in the vehicle， a method for constructing virtual reference signals utilizing grouping is proposed. First， the vehicle road test is conducted， the vibration acceleration signals collected near the four- wheel positions are selected as the original signals， and the number of virtual reference signals is determined based on the singular value analysis. Then， the original signals with high correlations are divided into a group based on the wheel position， the conversion matrix is constructed by the singular value decomposition for each group of original signals， and the virtual reference signals are constructed utilizing grouping. Finally， a road noise active control model based on the normalized filtered-x least mean square （NFXLMS） algorithm is built to analyze the performance of the virtual reference signals constructed utilizing group in reducing computational complexity and noise reduction. Compared with the original virtual reference signal construction method， this method has greatly reduced the computation complexity and improved the noise reduction effect， which verifies the effectiveness of the method.