中文摘要：

利用单个矢量阵元的阵簇估计提供的初始参数，对MUSIC噪声子空间谱进行迭代搜索谱峰，实现目标的方位估计．该方法减少了运算量，同时提高了目标方位估计的精度．为了检验算法的性能，进行了外场试验．利用3个矢量水听器组成了三元矢量阵，对比了矢量阵自初始化MUSIC算法和MUSIC空间谱估计以及常规波束形成的性能．试验结果表明，矢量阵常规波束形成的目标方位估计精度较差，MUSIC空间谱的估计性能较好，而迭代搜索MUSIC谱峰方法的定向精度最高．当空间严重降采样时，常规波束图的栅瓣高度接近主瓣高度，MUSIC空间谱表现出较强的栅瓣抑制能力，而自初始化MUSIC算法不受空间降采样的影响，总能给出正确的目标方位估计值．

英文摘要：

Higher precision direction of arrival （DOA） estimation can be obtained by iteratively searching the maximum of the MUSIC spatial spectrum using initial parameters obtained by array manifold estimation with a single vector hydrophone. A linear array was created using three vector hydrophones. A lake trial was carried out to test the validity and practicability of the self-initiating MUSIC algorithm. Experimental results showed that DOA estimation performance using conventional beamforming was worse, and estimation performance using the MUSIC spatial spectrum was better. The highest precision of DOA estimation was obtained using the self-initiating MUSIC algorithm. The height of the sidelobe of conventional beamforming was near the height of the mainbeam when the wavefield was undersampled. Sidelobes of the MUSIC spatial spectrum were depressed. The right angle estimation given by the self-initiating MUSIC algorithm was without the effects of the undersampled wavefield.