The Siganl Detection Technique Of Multi Weak Targets In Low SNR

In recent years, as the indepth research of stealth technology, the stealth ability oftarget are enhanced so greatly that make the detection ability of radar detection systemdecrease immensely and brings severe test to radar. TBD method can detect and trackweak target, but existing TBD methods only use target’s amplitude have low detectionefficiency in low SNR. Thus, in order to improve the detection ability, phaseinformation should be used into TBD method.Aiming at weak targets in strong noise, the TBD techniques have been analyzed inthe dissertation which is funded by an important aviation science funds. The mainresearch and innovation contents are as follow:The research trends of weak target detection technique, DBT method and TBDmethod have been illustrated. Then the existing incoherent TBD algorithms areanalyzed and aiming at the shortage of low detection probability without using phaseinformation in low SNR, coherent TBD algorithms are introduced. In addition, theparallel computing technique based on CUDA is introduced to improve the real-time. Atlast, based on the sparsity of target, the technique of compressive sensing is introduced.Focus on the coherent TBD algorithm based on KT, the target motion model and theradar echo model are build and the principle of traditional KT is analyzed. Then, giventhe range formula and azimuth formula, the range resolution formula and azimuthresolution formula are deduced. At last, TKTA-TBD is illustrated and the excellentdetection ability is demonstrated by the simulations.FKT is proposed and used into TBD algorithm. Aiming at the problem oftraditional KT which has slow calculating speed by using sinc interpolation to havescale changed in time domain, FKT which is deduced from the KT formula improvesthe calculating speed greatly by using scale change formula, FFT and IFFT to any oneperiod of signal through the IDFT formula. What’s more, FKT can obtain interesteddetection range by choosing the appropriate period of signal in target detect area. ThenFKTA-TBD is proposed to improve the calculating efficiency and get the suitabledetection range in the condition that FKTA-TBD guarantees the same detection performance as TKTA-TBD. At last, FKTA-TBD is optimized using parallel computingtechnique based on CUDA. As a result, the real-time is improved further.CSKTA-TBD is proposed. It uses compressive sensing to get rid of false targetsand improve target’s resolution and the detection performance greatly by utilizingFKTA-TBD to improve SNR firstly. On one hand, aiming at the problem that there aremany false targets formed by the integrated strong noise after FKTA-TBD,CSKTA-TBD can remove false targets and improve SNR. On the other hand, aiming atthe problem that the false targets formed by high sidelobes which are caused by thediscontinuous echo in azimuth direction, FKTA-TBD, CSKTA-TBD can remove thesidelobe and improve the target resolution and the multi-target detection ablity greatly.A time-domain coherent track before detect method based on improved dynamicprogramming and back projection is proposed: an improved dynamic programmingalgorithm using target’s prior information and the information of programmed frames toget the estimated tracks is proposed; then back projection algorithm is utilized toaccumulate the energy along the possible motion tracks of target. Compare with existingTBD algorithms, it improves the detection and tracking ability greatly.