基于深海大深度声场互相关的水面水下目标分类研究

With the evolution of ocean security situation, long-range detection of underwater target in deep ocean has become a hot research area, and how to effectively distinguish the underwater target from the surface one by making full use of sound propagation characteristics is a key scientific problem that needs to be solved. The cross-correlated acoustic field at great depth of deep water consists of multi-dimensional acoustic characteristics information including target’s depth and velocity, further study on which is of great significance for improving the long-range detection capability of deep ocean sonar system.. Focusing on the interference phenomena of cross-correlated field, this research comprehensively uses numerical simulation, theoretical derivation, and experimental data analysis to ensure the accomplishment of the following points. First, this research will be devoted to the formation mechanism and variation law of interference fringes and the extracting of acoustic characteristics information that can be used for estimating source depth. Second, without a prior knowledge of target range, this research will focus on the source depth discrimination method of surface and underwater target based on the union of multi-dimensional acoustic characteristics information. Third, by analyzing the experimental data, the interference theory of cross-correlated acoustic field will be developed and the robustness of the source depth discrimination method by the optimization of algorithms will be improved. This research will promote the fundamental research of deep ocean sound propagation theory and provide a new idea for surface and underwater target classification.

随着我国海洋安全形势日趋严峻，深海目标远程探测成为当今水声界的研究热点。如何利用水声传播特性实现水面水下目标分类是其中亟需解决的关键科学问题。在深海大深度声传播信道条件下，声场互相关包含目标深度、速度等多维目标声场特征信息，深入开展基于声场互相关的水面水下目标分类研究，对深海声纳目标远程探测具有重要的科学意义和应用价值。. 本项目围绕深海大深度声场互相关干涉现象，综合运用数值仿真、理论推导和实验数据分析，首先研究声场互相关干涉条纹的形成机理和变化规律，并提取可用于表征声源深度的声场特征量；然后在缺乏目标距离先验信息条件下，研究基于目标运动补偿和声场干涉特性长时间相干累积的稳健声源深度分类方法；最后通过分析海上实验数据，完善声场互相关干涉现象的理论解释，并通过算法改进提高分类方法的稳健性。研究成果将丰富深海大深度声传播的基础研究，同时为水面水下目标分类提供一种新思路。

顾及水声物理场的水声信号处理方法是提升声呐探测能力的重要途径。其中，水面水下目标分类能力是其中的关键环节。深海大深度接收声场及互相关声场包含的目标位置、速度等信息，隐含在声场干涉结构中。在理清多途声传播规律及机理基础上，本项目开展了扎实的研究工作，主要创新成果如下：. （1）目标参数估计：提出了一种基于垂直线列阵互相关函数延迟求和的多途时延稳健估计方法，把阵元间的互相关函数作为输入，借鉴常规波束形成原理，通过延迟求和可以稳健获得不同阵元间的直达波时延和海面反射波时延，以及参考位置处的直达波与海面反射波时延。. （2）目标深度分类：在前述多途时延估计方法基础上，提出一种利用多途时延比的目标深度区间估计方法。由于不严格满足平面波传播假设条件，目标深度真实值大于深度区间估计值上限。通过定性和定量分析发现，利用目标深度区间估计值可以实现近距离（5 km内）目标的测距与定深，以及远距离（10 km外）目标的距离和深度稳健二值判别。. （3）信道冲击响应估计：射线盲解卷积方法（RBD）是一种广泛应用的水声信道冲击响应（CIR）估计方法，其主要优势在于计算量小、实际应用方便。在深海大深度环境下，当接收垂直阵孔径较小时，有限带宽内的常规波束形成由于角度分辨率不足，往往无法区分海面附近声源的直达路径和海面反射路径，进而导致RBD方法性能下降。针对这一问题，基于劳埃德镜原理，提出了一种适用于短孔径垂直阵的深海盲卷积方法（DW-RBD），进一步完善了深海CIR估计理论与方法。