CAJ | 학술논문

方位超分辨是国内外雷达界长期以来持续探索的一项技术难题，针对求解过程中所遇到的病态性，本文深入研究了范数正则化方法。利用 L 2范数正则化方法对方位超分辨模型进行求解，针对 L 2范数正则化方法的不足并考虑到目标信号的稀疏性质，建立了 L 1范数正则化模型。为提高求解的计算效率，将其转化为半定规划模型，用预测校正原对偶路径跟踪法进行求解。针对不同信噪比情况进行了计算机仿真，初步结果表明，两种方法都能用于雷达方位超分辨，且在相同条件下 L 1范数正则化方法分辨性能更好，具有较强的噪声适应能力，在信噪比低至0 dB 时，分辨力提高1．7倍。
방위초분변시국내외뢰체계장기이래지속탐색적일항기술난제，침대구해과정중소우도적병태성，본문심입연구료범수정칙화방법。이용 L 2범수정칙화방법대방위초분변모형진행구해，침대 L 2범수정칙화방법적불족병고필도목표신호적희소성질，건립료 L 1범수정칙화모형。위제고구해적계산효솔，장기전화위반정규화모형，용예측교정원대우로경근종법진행구해。침대불동신조비정황진행료계산궤방진，초보결과표명，량충방법도능용우뢰체방위초분변，차재상동조건하 L 1범수정칙화방법분변성능경호，구유교강적조성괄응능력，재신조비저지0 dB 시，분변력제고1．7배。
The azimuth super-resolution is a technique problem which has been explored continuously in the field of radar.The L 2 norm regularization method is used for the solution to radar azimuth super-resolution problem,because of the ill-condition encountered in the solution process.The L 1 norm regularization model is also established for the shortage of the L 2 norm regularization method and the sparseness of the target signal.In order to improve the computational efficiency,the L 1 norm regularization model is transformed into a semi-defi-nite programming model which is solved by the predictor-corrector primal-dual path-following method.The computer simulation for different signal-to-noise ratio (SNR)is made.The preliminary findings show that both methods can be used in radar azimuth super-resolution,and the performance of the L1 norm regularization method is better under the same conditions,with a strong noise adaptive ability.It shows when SNR is not less than 0 dB,the resolution is increased by 1.7 times.