【作者】 韩业强；

【导师】 陈耀武；

【作者基本信息】 浙江大学 ， 电子信息技术及仪器， 2013， 博士

【摘要】 实时相控阵三维成像声纳系统采用单频窄带声脉冲信号透射整个水下探测场景,通过大规模平面换能器阵列接收声纳回波信号,并且运用相控阵技术同时产生数以万计个波束强度信号,从而获得实时三维图像。在实现实时相控阵三维成像声纳系统过程中,为了克服波束形成计算量庞大和硬件系统复杂且成本昂贵的难题,本论文分别针对相控阵三维成像声纳系统的波束形成算法和阵列稀疏优化算法进行深入研究,并将理论与实践相结合,应用于实时相控阵三维成像声纳系统设计。第一章绪论部分首先阐述了研究背景与选题意义,然后对国内外研究现状以及相关技术进行了综述和总结,最后列出本文的主要研究内容。第二章研究并提出一种远场波束形成算法,分布式并行子阵(DPS)波束形成算法。全面阵被分解成两级分布式子阵,一级子阵和二级子阵之间采用流水线分布式计算结构,一级子阵波束形成采用并行计算结构。首先对DPS波束形成算法过程进行了阐述并给出数据通路；然后将该算法与直接(DM)波束形成算法和快速傅立叶变换(FFT)波束形成算法进行计算效率对比和分析；最后通过仿真测试结果验证：在减少参数存储空间和计算负载的前提下,DPS波束形成算法保持波束方向图性能不变。第三章针对近场波束形成算法计算量过大的问题,首先基于时延参数的菲涅耳近似表达式和泰勒级数展开优化了近场时延参数；然后结合优化后的时延参数,针对DPS波束形成算法进行相位补偿和聚焦,将远场DPS算法扩展到近场成像区域；最后基于Matlab进行仿真测试,结果可以验证：近场DPS波束形成算法的波束方向图可以保持主瓣宽度和抑制旁瓣能量,内存需求量和计算需求量也得到大幅度降低。第四章针对平面换能器阵列的阵元数目多和硬件成本高的问题,将DPS波束形成算法和模拟退火算法相结合,应用于接收换能器稀疏优化设计。首先基于DPS波束形成算法重新定义一个能量函数；然后以新能量函数为目标,采用模拟退火算法进行阵列稀疏优化；最后将优化后的算法应用于接收阵设计并进行仿真测试,实验结果表明：优化后的算法与其它算法相比较,采用更少的换能器数目和更低的权重系数比,获得了相同的波束方向图性能,大幅度降低了系统的硬件复杂度和成本。第五章将理论与实践相结合,基于DPS波束形成算法设计了一个实时相控阵三维成像声纳系统的原理样机。首先,通过权衡系统制造成本、波束形成计算量和成像性能,根据DPS波束形成算法的精度、参数取值范围和约束条件,配置接收换能器阵列参数；随后详细阐述信号处理机的硬件系统设计和控制逻辑设计方案；最后将研制开发的原理样机应用于湖试和海试,测试结果表明：原理样机可以实现高达每秒20帧的实时高分辨率水下三维成像。第六章总结了本论文的研究成果和创新点,并进一步展望未来该课题相关领域的研究方向和内容。更多还原

【Abstract】 To start the processing of the3D imaging, the real-time phased-array three-dimensional (3D) imaging sonar system transmits an acoustic pulse to insonify the scene of interest under narrow-band. A large scale planar transducer array gathers he backscattered signals and beamforms in more than then thousands of steering directions. This thesis investigates the research on new techniques of beamforming and sparse array to overcome the huge computational load and the complicated hardware system.Chapter1demonstrates the background of the research, presents the current status of the relevant techniques in different contries, and lists the main content of the thesis.Chapter2proposes a beamforming algorithm worked in far field:Distributed and Parallel Subarray (DPS) Beamforming. The full array is subdivided into two distributed subarrays and the parallel beamforming is implemented in two-stage subarrays. First, the DPS beamforming process is described and a data-path is illustrated. Second, the computational requirements are compared among DPS, DM and FFT beamforming. Third, the algorithm is simulated and the experimental results verify that the DPS beamforming achieves a similar beam pattern performance with lower computational and memory requirements.In Chapter3, the near-field time-delay parameters are optimized based on the Fresnel Approximation and the Taylor Series Expansion first. Second, the DPS beamforming algorithm is extended to the near field condition and the optimized time-delay is applied to compensate the phase shift and focus in a point. Finally, the near-field DPS beamforming algorithm is simulated in Matlab. The experimental results demonstrate that the algorithm can not only maintain the mainlobe width and the sidelobe energy, but also reduce the memory and computational requirements.In Chapter4, the DPS beamforming algorithm is combined with the simulated annealing algorithm in order to thin and weight the transducers of the receiving array. First, a new energy function is defined based on the DPS. Second, the simulated annealing algorithm with the new energy function is applied to design the sparse array. Finally, the optimized algorithm is employed on a target array and the experimental results demonstrate that:The optimized algorithm can achieve the similar beam pattern with less transducers and a lower CTR.Chapter5combines the theory and the engineering applications. A prototype is designed based on the DPS beamforming algorithm. First, considering the costs, computational requirements and the image quality, the parameters of the receiving array are chosen based on the approximations, constraints and the range validity of the coefficients in DPS. Second, the detail of the signal processor is described. Finally, the prototype is tested in lake and sea and it can image the scene with a high resolution and20frames per second.The last chapter concludes the innovation points of the research in this thesis. The prospect of the future research is also described.更多还原