Strong Method Of Weak Target Detection Under The Clutter Background Research

With the development of stealth technology, the RCS magnitude of radar hasdecreased for one to two times and the target detection is greatly reduced. Theseproblems has brought serious challenges to the radar. In order to detect the weak targetin the environment of strong clutter, the research of clutter suppression and weak targetdetection need to be done. In the dissertation sponsored by an important science funds,the basic principle of radar target detection and clutter model and the suppressiontechniques have been analyzed. Then some exploratory research about space-timeadaptive processing (STAP) and Multi-Frame Coherent Integration TBD (MFCI-TBD)has been done. The dissertation’s main content comprises:1. The basic principle of radar target model and detection is studied. TheDetect-Before-Track(DBT) algorithm and the Track-Before-Detection(TBD) algorithmare analyzed. Then the MFCI-TBD algorithm is introduced to solve the shortcomingthat multi-frame incoherent integration in the existing TBD algorithm. At last, thetechnology of CUDA based on GPU which is used in parallelize processing isintroduced.2. Clutter suppression methods are investigated in detail. Firstly, several classicclutter statistical models are introduced and then a typical clutter suppression method isstudied: Optimal STAP. Based on the eigenvalue analysis in the cultter covariancematrix after demension reduction, a Joint-Domain Localized STAPmethod based oneigenvalue analysis is studied. The computational complexity is reduced with a smallloss of the performance of clutter suppression in the Joint-Domain Localized STAPmethod based on eigenvalue analysis. At last, simulations verify the validity of thismethod.3. The feasibility of MFCI-TBD is Researched. Then three methods of MFCI-TBDis given.1) A MFCI-TBD method based on KEYSTONE transforming whit phasecompensation is researched: this method uses KEYSTONE transforming to correctrange migration and a method to correct the Doppler spectrum migration. At last, multi-frame coherent integration is implemented for target detection.2) A MFCI-TBD method based on fast KEYSTONE transforming is proposed: atfirst, we verify that fast KEYSTONE transforming can be used for correcting rangemigration of continuous echoes. Then the proposed method combines fast KEYSTONEtransforming and phase compensation method to detect the weak target whose echoesare discontinuous in multi frame echoes.3) A MFCI-TBD method based on dynamic programming is proposed: the idea ofdynamic programming is used in the model of MFCI-TBD. In this method, coherentintegration is implemented along the tracks which may be the target’s tracks to detectweak target.The thesis discussed the processes and principles of these methods in detailed.Thesis tested and verified that these methods through simulation and compared theperformance of these three methods through simulation experiment. At last, thetechnology of CUDA is used to optimize the methods and reduce the running time ofthe methods. The real-time performance of the methods is improved by using CUDA.