水下运动阵列小样本被动定位问题研究

At present , Unmanned Underwater Vehicle(UUV) has been received national attention by all over the world. It locates the underwater objects using small moving array. In locally stationary process ,there are small-sampling datas, so we need to solve localization problem utilizing the small-sampling data with moving array..Passive Acoustic Localization with moving array based on small-sampling data in noise background environment is the goal of the project, and there are three ideas to solve it. At first, how to construct sparse base matrix and observation matrix which are suitable for passive localization should be studied, and then compressed sensing and sparse reconstruction method are intended to be used for high-resolution direction of arrival(DOA) estimation with small samples. At second ,aperture extending methods are studied.Passive synthetic aperture and sparse aperture extending method are used to solve small-scale problems. Third, after aperture extension, weighted subspace method for high-resolution DOA estimation will be proposed based on large dimensional random matrix theory spectrum analysis .Combining compressed sampling with aperture extending is also used for high-resolution DOA estimation..The proposed DOA estimation method will be test by water tank experiments and lake experiments. It is expected that the theoretical basis could be established for target passive location applications for UUV . Achieving effective technical means for passive DOA of UUV are also expected.

目前无人水下航行器的研究得到各国的高度重视，由于尺度的限制及其在运动时局部平稳过程中得到的样本较少，因此利用小样本实现目标的水下定位成为无人水下航行器研究的关键问题之一。.本项目研究的目标是解决在一定噪声背景下，运动小尺度阵列的小样本被动定位问题，即采用无人水下航行器携带的小尺度阵对水下目标实现被动定位。首先，深入研究如何构造适合于水下声信号被动定位的稀疏基和观测矩阵，采用压缩感知与稀疏重构的方法解决小样本的高分辨定位；其次，研究孔径扩展方法，拟采用被动合成孔径方法与有限孔径条件下的稀疏孔径扩展方法解决小尺度阵孔径不足的问题；最后，在孔径扩展的基础上，基于大维随机矩阵谱分析理论实现样本协方差矩阵子空间加权的高分辨目标定位，将压缩采样与孔径扩展相结合实现基于虚拟大孔径的小样本目标高分辨定位。

本项目以UUV为应用背景，研究了小尺度运动阵列目标被动方位估计技术。针对小尺度阵列存在孔径不足等问题，研究了两类阵列孔径扩展技术，包括均匀布阵的被动合成孔径和稀疏布阵的互质孔径扩展，提出了基于主特征空间、正交匹配追踪和互质阵的多种改进方法，横向对比了所提方法的性能，验证了它们的有效性；针对运动阵列方位估计时的快拍不足等问题，研究了四种稀疏重构小快拍方位估计方法，其中L1-SVD算法优势明显，在此基础上，提出的基于被动合成孔径的稀疏重构方位估计方法和基于主特征空间的小快拍方位估计方法，能够准确分辨相邻目标，同时解决了小尺度运动阵列的孔径不足与快拍不足问题，提高了方位估计的分辨率；提出了基于样本协方差矩阵特征值极限性质的小快拍方位估计方法和采用噪声及干扰空间谱加权的稀疏约束Capon波束形成方法，进一步抑制噪声并增加方位估计精度；利用样本特征向量信息，提出了一种基于协方差矩阵特征向量夹角联合概率密度函数的信源个数估计方法，以解决方位估计前的信源个数确定问题；结合小尺度阵列波束形成和被动定位的一系列理论与应用研究，搭建了实验系统，对上述方法进行了水池与湖上试验验证，结果表明所提方法可有效解决运动声纳目标方位估计问题，具有显著的应用前景。