Research On Squint Imaging Algorithms For SAR Mounted On Maneuvering Platforms And The Parallel Computing Of The GPU

The Synthetic Aperture Radar(SAR)systems mounted on maneuvering platforms has the advantages of high maneuverability,early observation and real-time imaging.The research on the squint imaging of the SAR mounted on maneuvering platforms and the parallel algorithm of the GPU can promote the SAR systems mounted on maneuvering platforms to realize fast real-time imaging and target detection and recognition functions.The research can be applied to the final guidance section of a new type of anti-enemy strike weapon,to improve the ability to accurately and effectively strike core targets,and to optimize the overall performance of the weapon system.There are an important application value and a research significance in national defense and military affairs.The main work of the thesis is as follows:Aiming at the curved motion trajectory of the missile platform,the radar seeker working mode is planned,and the system parameters such as signal bandwidth and pulse repetition frequency are designed according to imaging requirements.In the squint strip-map imaging mode of the maneuvering platforms,a curved trajectory SAR imaging algorithm based on azimuth nonlinear scaling is proposed.First,according to the motion parameters of the curved trajectory of the maneuvering platform,a fourth-order approximate slant range model of squint SAR imaging is established.Then,the range walk correction and range pulse compression are performed on the echo signal in the range frequency domain and azimuth time domain to reduce the cross coupling of the echo signal in the range and azimuth directions.Range bending correction is performed in the two-dimension frequency domain to correct the range bending caused by nonlinear terms in range migration,and to improve the range focusing performance of the scattering point target.Doppler center compensation and zero padding are performed in the time domain.Nonlinear scaling factors are used to correct the residual phase error in the echo,weakening the spatial variability of the Doppler parameters.The azimuth dimension is focused through SPECAN operation.Finally,geometric distortion correction is performed to obtain SAR images without distortion after focusing.In order to meet the real-time imaging requirements of the signal under the bomb platform,it is necessary to study the rapid realization method of the algorithm in this paper.In this paper,under the GPU-based CUDA architecture,a parallel acceleration algorithm for SAR strabismus imaging is programmed.Aiming at the hardware model and memory mode,the CUDA parallel algorithm design and optimization methods are analyzed from the aspects of synchronous and asynchronous communication between threads,memory access optimization and instruction optimization.In the CUDA parallel computing programming model,methods of memory management and thread allocation are proposed,such as shared memory access conflict avoidance,global memory aligned access and delay hiding.It lays the foundation for the CUDA parallel programming and optimization of the SAR imaging simulation system of the maneuvering platform.On the GUI platform of the Matlab environment,the interactive application software design of the missile-borne SAR imaging simulation system is realized,which makes the working mode of the radar seeker and the imaging simulation process modular and visual.Combined with actual ballistic trajectory parameters,squint SAR imaging simulation experiments were carried out on the lattice target and the ship target under the curved trajectory,and the imaging results were quantitatively and qualitatively analyzed to verify the effectiveness of the algorithm.