压缩感知雷达中感知矩阵自适应构建理论与关键技术研究

This project will investigate the theory and key technique of the adaptive construction of the sensing matrix in compressed sensing radar (CSR) to optimize its performance for observing sparse target scene. First, to slove the computation and storage problem of the large dimensional matrix caused by a redundant dictionary in the fast adaptive construction of the sensing matrix, the low dimensinal optimization criterion will be designed, and the radar transmit waveform and measurement matrix will be optimized by the hybrid optimization technique. Then, for severely decreased recovery performance in colored noise enviornment, the adaptive construction criterion of the sensing matrix, which has good orthogonaltiy and the capability of whitening colored noise, will be considered. The sensing matrix will be optimized by the convex programming technique to decrease the target scene recovery error in colored noise. Finally,we note that the CSR system can not effectively observe the target scene with wide dynamic range. The corresponding improvement criterion based on prior target scene information will be given and the reasonable sensing matrix design method will be proposed to enhance detection performance of the weak target. The research of this project will optimize the target information obtainment in colored noise enviornment and target scene with wide dynamic range for CSR, and plays an important role in improving the system performance of this new radar and extending its application fields.

本项目研究压缩感知雷达中感知矩阵自适应构建的理论和关键技术以优化稀疏目标场景的观测性能。首先，针对高维冗余字典下感知矩阵整体自适应构建面临的高维矩阵运算与存储问题，设计低维度的感知矩阵构建准则，采用混合优化技术解决雷达发射波形和测量矩阵的联合设计问题，减小所需运算和存储负荷；然后，针对有色噪声引起重构算法性能严重下降的问题，研究同时具有良好正交性和有色噪声白化能力的感知矩阵构建准则，采用凸规划技术完成其优化，降低在有色噪声环境下的场景重构误差；最后，针对压缩感知雷达无法有效对大动态范围场景观测的问题，建立基于稀疏场景先验信息的压缩测量空间动态范围改善准则，提出可行的感知矩阵设计方法，提升对弱小目标的检测能力。通过本项目的研究，完善现有的感知矩阵构建理论和方法，对于提升新型雷达系统性能、拓展其应用场合具有重要意义。

本项目研究压缩感知雷达中感知矩阵自适应构建的理论和关键技术以优化稀疏目标场景的观测性能。首先，针对高维冗余字典下感知矩阵整体自适应构建面临的高维矩阵运算与存储问题，设计低维度的感知矩阵构建准则，采用混合优化技术解决雷达发射波形和测量矩阵的联合设计问题，减小所需运算和存储负荷；然后，针对有色噪声引起重构算法性能严重下降的问题，研究同时具有良好正交性和有色噪声白化能力的感知矩阵构建准则，采用凸规划技术完成其优化，降低在有色噪声环境下的场景重构误差；最后，针对压缩感知雷达无法有效对大动态范围场景观测的问题，建立基于稀疏场景先验信息的压缩测量空间动态范围改善准则，提出可行的感知矩阵设计方法，提升对弱小目标的检测能力。 .本项目针对上述三个研究要点均已按照预定设想和目标进行了细致研究，对三个研究内容中的各个优化问题实现了优化算法的设计和计算机仿真分析；通过大量的计算机仿真实验表明了该项目在理论可行性和实践可行性完全符合预期设想。在研究过程中，我们还发现上述优化问题的建模和求解存在计算量超大的问题，我们还引入了GPU技术，用于加速优化问题的建模和求解。.通过本项目的研究，完善了现有的感知矩阵构建理论和方法，在一定程度上提升了现有压缩感知雷达的性能。