水下多目标回波稀疏表示及声成像方法研究

As an active research area, acoustic imaging of underwater multiple targets is an important problem in signal and information processing, underwater target recognition, underwater acoustics, etc. It is also urgently needed in the development of modern torpedo homing. But, in view of the essence of signal representation, current researches on underwater acoustic imaging methods and their engineering systems are usually based on representing signals by complete orthogonal basis. Such representations of signal are difficult to meet the needs of practical applications in torpedo homing and sonar fields because of their inherent disadvantages. Unlike conventional signal representation, sparse decomposition uses over-complete dictionary to represent signal. The great advantages it showed in many applications of signal processing provides theoretical reference for the researches of underwater multi-target acoustic imaging. Aiming to satisfy the significant requirement of modern torpedo homing under complex multi-target circumstances, this project utilizes advanced signal processing theory, signal sparse decomposition, to study the sparse representation of underwater multi-target echoes and acoustic imaging. The main research contents are as follows:.1. Study the sparse essential property of underwater multi-target echoes and reveal the sparse representation mechanism of the echo signals. Propose generalized sparse representation of the acoustic imaging echo data..2. Propose underwater multi-target acoustic imaging methods based on signal sparse representation theory. Extract multidimensional characteristic acoustic image containing time, frequency and spatial domain information of the underwater multiple targets..3. Put forward an anechoic tank experimental method to simulate the sparse representation and acoustic imaging of underwater multi-target echoes. Design a set of experimental system to demonstrate and verify the validity of the imaging mechanism..This project aims at breaking through the technique bottleneck in modern torpedo homing. And provide new theory and methods for the development of underwater multi-target acoustic imaging.

水下多目标声成像是信号与信息处理、水下目标识别、水声物理等学科的基本问题，也是现代鱼雷自导发展迫切需要解决的关键问题。但从信号表示的本质来看，目前主要研究的水声成像方法及水声成像系统都建立在信号完备正交基表示的基础上，其理论具有局限性，限制了在实际中的应用。信号稀疏分解为水下多目标声成像提供了新的理论。本项目针对复杂多目标环境中鱼雷自导发展的重大需求，应用信号稀疏分解的先进信号处理理论，研究水下多目标回波稀疏表示及声成像的理论和方法，主要研究内容有：①揭示水下多目标回波的稀疏性表征机理，提出声成像回波数据的一般性稀疏表示方法；②提出基于信号稀疏表示理论的水下多目标声成像方法，提取包含多目标时、频、空域多维度信息特征的声图像；③提出消声水池模拟实验方法，研制一套水池模拟实验系统，验证这种声成像机理的正确性。为突破现代鱼雷自导发展的技术瓶颈问题、推动水下多目标声成像技术发展提供新的理论和方法。

针对水声成像的正交变换理论在低信噪比、小阵列、少快拍、相干源等实际应用条件下分辨性能下降的缺陷和不足，该项目研究水下多目标回波稀疏表示及声成像方法，旨在解决实际应用中的关键问题，满足复杂水声环境中的成像需求。.主要研究内容有：①研究信号稀疏表示的理论，探索声成像回波数据的一般性稀疏表示方法；②研究基于信号稀疏表示理论的水下多目标声成像方法，提取包含多目标时、频、空域多维度信息特征的声图像；③研究消声水池模拟实验方法，研制一套水池模拟实验系统，验证这种声成像机理的正确性。.该项研究取得的主要研究成果包括：①系统研究了信号稀疏表示的FOCUSS方法、匹配追踪方法、凸优化类方法、稀疏贝叶斯学习方法以及稀疏表示固有的模型失配（off-grid）问题和自适应过完备字典设计问题，给出了水下多目标声成像的新理论和新方法。②研究了稀疏表示理论在水下多目标声成像中的应用问题，揭示水下多目标回波的稀疏性表征机理，提出了基于多目标回波空域稀疏性的高分辨DOA估计方法、基于多目标回波空时分布稀疏性的角度-多普勒声成像方法、基于多目标回波时频分布稀疏性的微多普勒频率估计方法，有效提取了包含多目标回波时域、频域、空域多维度信息特征及声图像。③研制了一套消声水池模拟实验系统，开展了理论和方法的实验研究，验证这种声成像机理的正确性。.该项研究突破了波束形成的方位分辨率，对双源的分辨率达到波束宽度的1/5，并可分辨出体目标2~9个亮点。运动平台的稀疏空时处理在混响背景中分辨出2节低速目标。稀疏重构STAP比传统STAP的改善因子(IF)降低约10dB。采用微多普勒原子正确估计了振动频率分别为30Hz、40Hz、50Hz、60Hz，振动幅度均为1cm的4亮点的瞬时频率、振动幅度和角度参数。.该项研究不仅有助于推动信号处理学科的发展，对水声装备声成像技术发展也具有重要意义。