空间目标监视是利用探测设备对空间目标进行探测和跟踪，获取目标尺寸、轨道等信息，并掌握空间态势的过程。雷达成像技术作为实现空间目标监视的重要手段之一，以逆合成孔径雷达(ISAR)为代表的二维成像技术已较成熟，而雷达三维成像能够获取目标形状、尺寸、姿态等信息，成为空间目标监视技术的重要发展方向。目前空间目标三维成像技术仍处在前沿理论研究阶段，提出的干涉ISAR和序列ISAR等成像方法仅适用于卫星等具有稳定轨道的慢机动目标，受到目标机动特性对信号相参性的影响，不具备对高机动目标进行三维成像探测的能力。针对这一问题，本文深入分析空间目标三维成像机理，提出了面向高机动目标的稀疏阵列雷达三维成像技术思路，从稀疏阵列雷达优化设计理论和稀疏阵列三维成像算法等方面展开系统性研究。论文的主要贡献和创新性成果如下： 提出了稀疏阵列雷达瞬态三维成像技术体制，利用阵列雷达空间采样代替传统成像雷达时间采样，仅用几个甚至单个发射脉冲即可对目标进行三维成像，从而减少相参处理时间内目标机动对成像的影响，实现对高机动目标的三维成像能力。建立了基于空间谱理论的稀疏阵列三维成像性能分析模型，为成像分辨率和系统参数的优化设计奠定了基础。 针对稀疏阵列雷达阵元位置优化的问题，提出了基于观测矩阵优化理论的稀疏阵列构型优化算法，从压缩感知观测矩阵角度对阵元位置进行优化设计，能够降低稀疏观测矩阵互相关系数，提升三维成像质量。在此基础上，建立了通道间存在幅度误差、相位误差以及阵列定标误差时的回波信号模型，通过理论推导和仿真分析，总结了各项系统误差对于稀疏阵列成像质量的影响规律。 针对稀疏阵列雷达三维成像方法的问题，提出了一种基于稀疏贝叶斯学习的三维成像算法，距离向利用宽带信号获得一维距离像，俯仰向和方位向利用基于贝叶斯估计的超分辨算法实现二维成像，最终重建目标三维模型，理论分析表明，算法结果达到克拉美-劳界。 同时，提出了一种基于弹性网络和贝叶斯推断的超分辨成像算法，解决了传统干涉ISAR成像中等效散射中心无法分辨的问题，能够自适应估计等效散射中心散射点个数并实现超分辨成像，成功将该算法应用于阵列ISAR成像系统并重建目标点云模型。 综上，本文提出了稀疏阵列雷达三维成像技术方法，为空间目标三维成像技术的发展做出一定贡献，对后续的三维成像研究提供借鉴意义。 关键词：稀疏阵列，瞬态三维成像，阵列ISAR，弹性网络，稀疏贝叶斯学习

 Space target surveillance is the process of detecting and tracking the space target to obtain the target size, orbit and so on, in order to realize space situational awareness. The radar imaging technology is one of the important means to realize space target surveillance. The 2D imaging technology, represented by the synthetic aperture radar, is rather mature. The radar 3D imaging technique can obtain the information of the target shape, attitude and so on and it becomes an important direction of space target surveillance technology. At present, 3D imaging technology of space targets is still at the advanced theoretical research stage, and the interferometric ISAR and sequence ISAR imaging methods are only suitable for slow maneuvering targets with stable orbits, such as satellites and so on. Due to the influence of targets’ mobility on signal coherence, it is not able to realize 3D imaging of highly maneuvering targets. In order to solve this problem, this paper analyzes the 3D imaging mechanism of space target, and puts forward the 3D imaging technology based on sparse array for highly maneuvering target. The optimization design theory and 3D imaging algorithms of sparse array radar are systematically studied in this paper. The main contributions and innovations of the paper are as follows: A new transient 3D imaging technique based on sparse array is proposed, in which the array radar spatial sampling is used instead of the traditional time sampling, and the 3D reconstruction can be fulfilled with only a few or even single pulse. Thus, the influence of targets’ high mobility on imaging is reduced in the coherent processing time, and the 3D imaging capability of highly maneuvering target is realized. A 3D imaging performance analysis model of sparse array based on spatial spectrum theory is established, which lays a foundation for the optimal design of system parameters. To solve the problem of optimal elements position of sparse radar array, an optimization algorithm based on the observation matrix optimization theory is proposed. The element position is optimized from the view of sensing matrix optimization in Compressed Sensing. This algorithm can reduce the cross-correlation of the sensing matrix and achieve a superior performance of 3D imaging. On this basis, the echo signal model is established when there are amplitude error, phase error and array calibration error. The effects of system errors on the imaging quality of sparse array are summarized. To solve the problem of 3D imaging based on sparse array radar, a 3D imaging algorithm based on sparse Bayesian learning is proposed. The resolution ability in the range direction is realized by transmitting wideband signal which that in the pitch and azimuth directions are realized by a super-resolution algorithm based on Bayesian estimation. The 3D model of target can be reconstructed based on this algorithm and theoretical analysis shows that the Cramer-Rao bound is achieved. A super-resolution imaging algorithm based on elastic net and Bayesian inference is proposed, which solves the problem of synthetic scatterers in the conventional interferometric ISAR. It can effectively estimate the number of scatterers of the synthetic scatterer and realize super-resolution imaging. The algorithm is successfully applied to array ISAR imaging system and the point cloud model of target is reconstructed. In summary, 3D imaging method based on sparse array radar is proposed in this paper, which contributes to the development of 3D imaging technology of space target. These findings contribute in several ways to our understanding of space targets threedimensional imaging and provide a reference for further research. Key Words: Sparse array, Transient three-dimensional imaging, Array ISAR, Elastic Net, Sparse Bayesian learning