Radar Multi-frame Joint Detection,Tracking And Classification Technology

With the needs of national security and the development of national defense science and technology,modern warfare has higher and higher requirements on the ability of radar system information acquisition.On the basis of high-precision and robust detection,a certain target classification capability is an urgent requirement of modern radars.The joint tracking and classification technology establishes a feedback mechanism between the tracker and the classifier by mining the dual dependence between the target state and the target type,so as to achieve the effect of mutual promotion between the two,thereby improving the system detection ability and system classification ability..However,the existing framework has two limitations,such as easy loss of weak target signals(such as stealth aircraft,small unmanned aerial vehicles,etc.),and limited dimension of target type feature information incorporated.The multi-frame joint detection,tracking and classification(MF-JDTC)technology improves the target signal-to-noise ratio by combining the number of multi-period samples,and uses target feature resources such as amplitude,speed,height,and maneuverability to build a higher-dimensional target type feature space,fully mining the dual dependencies between states and types in multiple consecutive periods,establishing a dual correlation feedback mechanism,forming a tightly coupled configuration of detection,tracking,and classification,and finally realizing the integration of detection,tracking and classification,which can effectively improve the radar system's detection and classification ability for the weak targets.The current MF-JDTC technology has problems such as the lack of theoretical framework and fast implementation algorithms.Based on this,this thesis mainly focuses on the research on radar multi-frame joint detection,tracking and classification technology.The main research work and contributions are as follows:1.According to the difference of RCS of different types of targets,a type-related radar echo measurement model is established,which lays a model foundation for the establishment of subsequent theories.2.The theory of echo-level optimal multi-frame joint detection,tracking and classification is studied,and two ways of expressing the uncertainty relationship between the state and type of unknown target multi-frame from different angles are proposed,and the detection statistics of the detection statistics are deduced.The closed expansion provides a theoretical foundation for the design of the subsequent MF-JDTC algorithm.3.The multi-frame joint posterior ratio JDTC technology is proposed,and the advantages and disadvantages of serial processing and parallel processing are analyzed.Aiming at the parallel processing mode,a multi-frame joint posterior ratio JDTC technology based on sliding window stepping and sliding window iteration is proposed,which realizes the approximate estimation of the multi-frame joint state-class posterior probability ratio.The proposed algorithm has robust detection ability and high reliability recognition and classification ability.4.The multi-frame joint feedback JDTC technology is proposed,which constructs the mutual feedback mechanism of state and class in multiple scanning periods,and realizes the accumulation of value functions to the greatest extent.Compared with the multi-frame joint posterior ratio JDTC technology,the multi-frame joint feedback JDTC technology has further performance gains in weak target detection and classification.The above algorithms have been verified by simulation,and the results have confirmed the effectiveness of the proposed algorithm on weak targets.