【作者】 李悦丽；

【导师】 梁甸农；

【作者基本信息】 国防科学技术大学 ， 信息与通信工程， 2008， 博士

【摘要】 采用合成孔径雷达(SAR)技术能够获得全天候、全天时、远距离的高分辨雷达图像,是提高精确制导武器打击精度的有效途径。然而,SAR成像末制导需解决以下问题:1)SAR图像的方位分辨率在沿平台航向的前斜视区域会迅速下降;2)导弹的机动飞行偏离理想运动轨迹,带来严重的运动误差使图像质量恶化;3)SAR实时成像需要大运算量和存储量,而弹载平台信号处理机难以满足要求。针对上述问题,本文围绕弹载SAR成像技术开展了以下研究工作:1、研究了弹载SAR大斜视角高分辨成像问题,提出了改进的方位向非线性CS算法,解决了时域线性距离走动校正(RWC)带来的聚焦深度问题,分辨率1m时成像斜视角可达50°以上。基于大斜视角SAR成像几何关系,研究了RD类和CS类算法,分析了瞬时斜距模型的近似误差和回波频域解耦合的残余相位误差,指出提高算法性能的关键在于:三次距离偏移量的补偿和二次距离压缩(SRC)的精度。通过理论推导证明:时域线性RWC可减小解耦合误差,但校正到同一距离门的目标存在随方位偏移线性变化的调频率误差。在此基础上,提出先补偿三次距离偏移,再引入改进非线性扰动方程补偿调频率误差的算法,仿真结果表明:ANCS算法成像分辨率高,聚焦深度和成像处理角更大。2、研究了匀加速平台的SAR成像及运动补偿方法,提出了一种二维频域补偿匀加速度的改进RD算法,提高了目标分辨率和峰值旁瓣比;并结合对比度最优法,给出了通用的匀加速平台SAR成像和运动补偿流程。首先,基于考虑三方向匀加速度的瞬时斜距模型和回波信号的多普勒历程,指出匀加速运动补偿的重点应为航向速度误差和视线位移误差;随后,给出了改进的RD和SPECAN算法,分别在距离徙动校正和方位聚焦处理中修正滤波函数,并在SPECAN成像后补偿了匀加速度带来的几何失真,仿真结果表明:算法简单,有效;在距离频域补偿视线位移误差,结合对比度最优法,给出了通用的成像运动补偿流程,并通过机载SAR飞行试验数据进行了成像验证。3、研究了弹载雷达的前视成像技术,基于多通道解卷积原理,提出一种单脉冲雷达解卷积前视成像新方法,仿真试验表明:在DBS失效的航向附近,解卷积图像的角分辨率比实孔径图像提高约10倍。针对条带式SAR图像的方位分辨率在前斜视区域迅速下降的问题,给出了导弹俯冲段DBS成像的信号处理参数选择准则,分析了方位分辨率的变化趋势;在DBS成像盲区,利用单脉冲雷达和差通道的准互质性提高前视图像角分辨率,并提出考虑天线方向图截断形状的解卷积器设计方法,可有效降低信噪比损失。4、研究了弹载SAR信号处理机的设计技术,设计了适宜弹载SAR成像的多DSP信号处理机结构,基于通用DSP芯片TS201,提出了SAR成像流程映射和算法优化的方法。根据典型通用DSP的特点和成像需求,设计了三种弹载SAR多DSP信号处理机结构,以匀加速平台SAR成像流程为例,基于TS201给出了从算法到结构的映射方法,提出了实时成像中关键步骤优化的具体方法;最后,基于主从式信号处理机结构,实现了弹载DBS成像信号处理机。更多还原

【Abstract】 Synthetic aperture radar (SAR) can achieve long range, high resolution imagery in all weather conditions, day and night, which contributes to an improvement of the orientation performance on precision guided weapons in strike. However, there are three problems must be resolved in missile-borne SAR applications: 1) the cross range resolution of SAR is degraded near the direction of flight due to the decrease in the Doppler bandwidth; 2) Serious motion errors are induced by the trajectory deviations from the straight flight track and forward velocity variations under maneuvers; 3) the large computational and memory requirements in real time SAR imaging are difficult to fulfilled on missile platforms. All the problems are researched on in this dissertation and the main achievements are as follows:1. The high resolution imaging algorithms of highly squinted missile-borne SAR are investigated. A modified azimuth nonlinear CS (ANCS) algorithm is proposed to resolve the limitation of the focus depth, which caused by linear range walk correction (RWC) in time domain. Simulation results are presented showing that it can be used toprocess data with up to approximately 50°angle with 1m resolution. Based on the squinted SAR geometry, the RD and CS algorithms are studied. A detail study is preformed on the Taylor extension of the range model and the decoupling errors of the signal expression in 2D frequency domain. It is concluded that the cubic range cell migration correction (RCMC) and secondary range compression (SRC) are significant in highly squinted SAR imaging. The investigation on RWC demonstrates that: the decoupling phase errors can be decreased after RWC; however, the focus depth of map is limited due to the azimuth variations of the target’s Doppler rate. So the modified ANCS algorithm is developed to resolve this problem. After the cubic RCMC, an azimuth nonlinear phase term is introduced to equalize the Doppler rate in the same range cell. The modified algorithm can process highly squinted data and extend the focus depth without deterioration of the image quality.2. The SAR image formation and motion compensation (MOCO) algorithms for constant accelerations platform are investigated intensively. A refined RD algorithm is proposed to compensate the constant accelerations in 2D frequency domain, which can improve the cross range resolution and PSLR greatly. In combination with the contrast optimization auto-focus (COA) approach, a generalized block diagram is given for data processing. Firstly, according to the range model and the return signal expression, it is concluded that the velocity error along the flight path and the displacement along the line of sight (LOS) must be compensated in data processing. Secondly, two algorithms including refined RD algorithm and spectral analysis (SPECAN) algorithm are derived to compensate the motion errors effectively. Geometric correction is implemented simply in the SPECAN algorithm too. In a different way, all the displacement along the LOS can also be compensated in range frequency domain. Using the COA approach to correct the residual phase errors, a generalized procedure for imaging and MOCO of the constant accelerations platform is provided. Finally, this procedure is tested by using data from an airborne flight experiment.3. The forward looking imaging techniques are investigated. A multi-channel de-convolution imaging method applied to mono-pulse radar is proposed to sharper the angular resolution in the direction of flight. Simulation results show that: about 10 times of improvement in angular resolution over real aperture image can be achieved. To enhance cross range resolution around the boresight, the Doppler beam sharpen (DBS) technique is used. As an unfocused SAR technique, the main specification and processing parameters of DBS in diving phase are deeply analyzed and the characteristics of cross range resolution are tested by simulation. To eliminate the DBS blind sector in the boresight, the return signals of the sum beam and the difference beam are deconvolved to sharpen the angular resolution. With the consideration of the truncation effect, the de-convolution operators are refined to restrain the noise caused by de-convolution effectively.4. The techniques to implement real time DSP systems in missile-borne SAR are investigated. Using multiple generalized DSPs as the processing units, the structures of real time missile-borne SAR processor are given. The mapping of SAR imaging procedure to a multi-DSPs system is presented and optimized. At first, according to the characteristics of the generalized DSPs, three SAR processors are proposed to adapt different missile-borne SAR applications. Subsequently, the mapping method of the SAR imaging procedure of constant accelerations platform to a processor formed by multiple TS201s is analyzed. Meanwhile, the key steps in the imaging algorithm are optimized in details. Finally, based on the master-slave structure processor, a missile-borne DBS imaging processor is implemented.更多还原