Moving Target Detection And Classification Method In Passive Radar

Passive radar is a kind of bistatic radar that uses non-cooperative opportunistic radiation sources to detect targets.It has significant advantages in anti-jamming signals,anti-stealth targets,anti-radiation attacks,anti-low-altitude penetration,etc.It has wide and critical applications in both military and civil fields.In scenarios such as air defense and airport surveillance,the detection and classification of high-speed maneuvering targets and UAVs are very important,and passive radars also need to undertake related tasks.However,due to the structural characteristics of passive radar system,the target echo signal is often weak in energy.How to achieve effective accumulation of target energy under low signal-to-noise ratio and achieve robust detection and accurate classification of the target has become a major problem.In the detection of high-speed maneuvering targets,it is necessary to improve the detection capability of the targets by long time accumulation,however,due to the maneuvering characteristics of the targets,distance migration and doppler spreading problems will occur in the accumulation,resulting in energy loss,which is detrimental to the robust detection of targets.Therefore,in order to improve the target detection capability as much as possible,the target echo distance migration and Doppler spreading need to be compensated.In addition,when classifying a typical low,slow and small micro-motion target such as a rotary UAV,its micro-Doppler effect can reflect the motion and structural size characteristics of the target,so the differences in micro-Doppler characteristics of different types of rotor UAVs can be fully exploited for classification.However,the traditional manual feature extraction method requires huge workload,and the subjective factors of experts can also reduce the efficiency and accuracy of classification.Therefore,there is a need to combine the microdynamic target classification problem of passive radar with artificial intelligence techniques to achieve accurate classification of UAV targets by extracting their microdynamic features at low signal-to-noise ratios.In this paper,with funding from the CST special project and the Innovation Special Zone project,the highspeed maneuvering target detection method and the rotating-wing UAV target detection and classification method based on passive radar are studied with Digital Video BroadcastTerrestrial(DVB-T)signal as the radiation sources respectively.The main work is summarized as follows.1.Firstly,the geometric model of passive radar system is established,and the DVB-T system model and signal modulation principle are introduced.Secondly,the traditional crossambiguity function and its equivalent fast algorithm are studied.Finally,the power of each signal component in the echo channel signal in passive radar is analyzed,and the principle and implementation method of direct wave cancellation based on the Extended Cancellation Algorithm(EC A)is studied.Finally,the simulation verifies the effectiveness of the EC A method and provides a theoretical basis for subsequent research.2.Aiming at the high-order migration compensation problem of high-speed maneuvering targets,a high-order migration compensation method based on passive radar is proposed.Firstly,the reasons and effects of high-order migration are analyzed according to the echo signal model of the uniformly accelerated moving target.Secondly,a method based on Keystone transform and search matching compensation function is proposed to compensate high-order migration,and three realization methods of Keystone transform and the realization method of matching compensation function are studied.Then,the principle of constant false alarm detection is introduced.Finally,the effectiveness of the proposed compensation method is verified by simulation experiments,and the improvement of the detection SNR by the compensation is analyzed combined with the ambiguity function.3.Aiming at the difficulty of detection and classification of slow targets such as rotor UAVs,a classification method based on Passive Radar Squeeze-and-Excitation Convolutional Neural Network(PR-SE_CNN)is proposed.Firstly,the signal model of the rotor UAV in passive radar detection scene is established,and the different time-frequency domain features of UAVs with different numbers and lengths are analyzed.Secondly,in order to enhance the classification performance of the network for UAVs under low signal-to-noise ratio,the Squeeze-and-Exciataion module is embedded in Convolutional Neural Netword(CNN)to extract the features of different channels.Finally,the effectiveness of the proposed PR-SE_CNN for UAV classification is verified by comparing the simulation with traditional Support Vector Machine(SVM)and CNN.The experimental results show that the proposed method can effectively classify UAVs even with lower SNR and smaller number of samples.