Research On Detection And Tracking Method Of Rotor UAVs

The rapid development of UAV industry and the abuse of UAVs caused by low access threshold bring new threats to social stability and public airspace security.Effective detection and tracking is an important part of UAV supervision.Compared with other detection methods,radar system has the advantages of small weather impact and long detection distance,which is more suitable for automatic detection and tracking of such small low-flying targets.With the development of digital signal processing technology,linear frequency modulated continuous wave(LFMCW)radar would replace the pulse radar because of simple structure,higher resolution,etc.,which can be used as an important means to detect low altitude UAV in medium and short range.Therefore,the research on radar detection method of UAVs is helpful to monitor the unauthorized UAV,ensure the safety of production and living environment,which also has important theoretical significance and practical value,and has broad application prospects in lowaltitude security and other fields.There are usually one or more blades for drones,causing the micro-doppler effect of radar echo signal,from which the rotating characteristic parameters can be extracted for detection and recognition.Based on the theory of LFMCW radar,this paper studies the methods of target detection,parameter estimation and tracking of UAV,including the following three parts:(1)Firstly,the LFMCW radar signal principle is analyzed,and the echo of UAV is modeled,based on which the micro Doppler effect of rotating blades is analyzed.Then the relationship between the echo spectrum broadening and rotor parameters would be obtained,which provides theoretical basis for subsequent processing.(2)According to the periodic frequency characteristics of UAV echo signal,the detection method based on the scattering transformation and the variational mode decomposition is designed to separate and reconstruct the main body echo and fretting components for UAV target detection,and estimate the motion state,parameters of rotor speed,and blade length,etc.(3)In view of the situation that the main body of the target is located in the Doppler blind area for clutter suppression,according to the distribution of the extended traces generated by blade motion,combined with the principle of data association algorithm,a JPDA tracking method based on clustering division and multiplicative noise observation model is proposed for the UAV target tracking,which effectively avoids the problems of missing detection and track fracture.