水下无人平台小尺度高性能仿生感知研究

Intelligence of underwater unmanned platform is hotspot of the current research. In the face of severe marine security situation, it is urgent to carry out in-depth and continuous research in the field of intelligent sensing. This project is aimed at small scale high performance intelligent perception. Based on animal sonar, a new underwater unmanned sensing system is proposed. First of all, starting from the influence of animal sonar performance space-time-frequency factors, theoretical and experimental research is carried out after modeling of animal sonar. Correctly understanding of the animal sonar sensing performance is strived. Secondly, aimed at small scale and the high resolution requirements of unmanned underwater sensing system, the bionic design of waveform and array manifold is studied in order to obtain high range and azimuth resolution. Then, based on in-depth analysis of animal sonar echo processing model, combined with modern signal processing methods for biomimetic environment perception and target identification, also improvement of performance by continuous cycle will be done. Finally, waveform adaptation, three-dimensional imaging, fine structure identification experiments are designed to experimentally verify and optimize the bionic design, perception and identification methods. The research on the theory and engineering technology of the small scale bionic perception of underwater unmanned platform is of great significance and broad application prospects.

水下无人平台的智能化是当前的研究热点,面对严峻的海洋安全形势，急需在智能感知领域开展深入、持续的研究。本项目致力于小尺度高性能智能感知，借鉴动物声呐，开展新型体制的水下无人感知研究。首先，通过对动物声呐建模，从影响动物声呐性能的空-时-频因素入手，开展理论、仿真和实验研究，力求正确理解动物声呐感知性能。其次，针对水下无人感知系统的小尺度、高分辨需求，借鉴动物声呐开展信号形式、阵列流形为核心的仿生设计以获得高距离、方位分辨能力。再次，在深入分析动物声呐接收处理模型的基础上，结合现代信号处理方法，寻求仿生环境感知和目标辨识方法，并通过连续周期的处理提高目标辨识性能。最后，设计波形适应性、三维成像、目标精细结构识别等实验，对仿生设计、感知和辨识方法开展实验验证与优化。水下无人平台小尺度仿生感知的理论和工程技术研究，具有重要的基础研究意义和广阔的工程应用前景。

动物声纳的优越性为实现高精度目标定位提供了科学依据，本项目致力于水下无人平台的智能感知，通过仿生设计和处理获得小尺度条件下的目标定位和辨识。首先，从理论、仿真和实测实验入手，深入分析了蝙蝠声呐的分辨能力，发现了蝙蝠通过增加频带解决小孔径条件下的时间、空间分辨的问题，提出并由3D打印蝙蝠外耳实测数据初步验证了基于稀疏表达的动物声呐接收模型。其次，借鉴蝙蝠体制，在小尺度约束条件下开展了多谐波发射波形和稀疏随机阵列为核心的系统仿生设计。再次，通过理论推导和数值计算深入研究了距离、方位、俯仰等字典和Benchmark模型角度字典的构建，分析了发射波形、阵列流形、采样率等对字典相干性的影响。最后，结合压缩感知理论，提出了基于稀疏表达的动物声呐接收模型的目标定位方法和角度辨识方法，提出了可降低运算量的基于距离-方位积分字典的目标定位方法，通过仿真、水池以及外场试验验证了可实现多目标定位（优于SCAT模型）的基于稀疏表达的动物声呐接收模型，验证了小尺度约束（时间空间强欠采样）条件下的目标定位和角度辨识能力。总之，项目研究原理性的实现了一种新型轻便高分辨仿生目标探测体制，具有重要的理论和工程应用价值。