相关噪声背景下声矢量阵目标方位估计

DOA Estimation under Correlated Noises Using Vector Hydrophone Array

传统的矢量水听器数据处理方法将单个矢量水听器的输出用一复数向量来表示,没有充分利用矢量水听器各振速通道的正交特性,为此提出了用复四元数对矢量水听器的输出数据建模,将矢量水听器输出的三个振速分量用复四元数的三个虚部表示,保留了矢量水听器各阵元的正交性,对相关噪声有较好的抑制能力。针对声源的非平稳特性,将观测数据分成多个子段,计算每个子段的协方差矩阵并进行向量化处理,通过一正交映射,消除干扰噪声,最后对去除噪声的伪协方差矩阵做特征值分解,利用MUSIC算法原理建立目标的方位估计公式。算法降低了对声源信号平稳性要求,对相关噪声有较强的鲁棒性。

In traditional acoustic vector data processing, the output of a single vector hydrophone is modeled as a complex column vector, which cannot hold the orthogonal structure of the hydrophone. According to this, the eomplex-quaternion data model of the vector hydrophone is proposed, and the velocity elements are put in the position of three imaginary parts, which hold the orthogonality of velocity sensors and is robust to coherent noises. As the non-stationary properties of the acoustic sources, the received data are divided into multiple frames. The covariance matrix and its vectorization of each frame are calculated. An orthogonal projection is employed to eliminate the interference noises. The noise free covariance matrix is used to estimate the DOA of the sources by taking use of the principle of MUSIC algorithm. The simulation result shows that the algorithm lower the smoothness requirement for the acoustic wave, and has better robustness to noises.