基于任意声矢量传感器阵列的角度和频率估计算法

Novel joint angle and frequency estimation algorithm based on arbitrary acoustic vector array

提出了一种基于任意声矢量传感器阵列下联合角度和频率估计新方法。将矢量传感器阵输出的信号进行建模分析表明此信号具有平行因子四线性模型特征。利用四线性分解的唯一性条件,从分解得到的矩阵中联合估计出声源的参数。该算法首先应用平行因子四线性分解算法估计出频率矩阵和方位矩阵,然后根据频率矩阵和方位矩阵的结构特点及最小二乘方法进行声源参数的计算。该方法与传统算法相比,无需多维谱峰搜索及参数配对,并具有更高的估计精度,在工程上有一定的应用价值。计算机仿真和实验结果验证了该方法的有效性。

A novel joint angle and frequency estimation algorithm for acoustic vector sensor array is presented in this paper. The output signal of the acoustic vector array is analyzed and the result shows that it has parallel factor quadrilinear model characteristics. The source parameters can be estimated from the resultant matrices via low-rank decomposition that utilizes the uniqueness of PARAFAC decomposition. The algorithm firstly uses parallel factor quadrilinear decomposition algorithm to estimate the direction and frequency matrices; secondly the frequencies and direction of arrivals （ DOAs）, are estimated according to the structural characteristics of the matrices and least square principle. Compared with traditional algorithm, the method which doesn＇ t require multidimensional searching spectral peak or parameter pairing, has higher estimation precision of the parameters, and has certain application value in engineering. Simulation and experiment results verify the effectiveness of the proposed algorithm.