WeakTarget Detectionfor Aircraft-warning Radar

At present homeland air defense of China are faced with serious challenges fromthe stealth targets and small targets. So this paper goes from this tough reality problemand closely around the core purpose of the detection for weak high-speed moving target.Firstly the traditional detecting methods commonly used for radar target is reviewedsystematically and summed up as five groups of detection methods: matched filteringmethod, the coherent accumulation method, the non-coherent accumulation method,image processing method and the transform domain method. On the basis of this review,this paper holds fast to the root of the target detection and around improving the SNRbefore detection as much as possible all method is further classified into two categories:one is as far as possible improve the accumulation of the target signal energy, the otheris to reduce the energy of the noise as much as possible. According to this, the paperputs forward two categories, a total of four kinds of new improved accumulation ofweak target detection method: the first major category is directed at that a long timeaccumulation for a high-speed motion target would result in range migration and giveout three kinds of improved detection method (a coherent accumulation detectionmethod using keystone transform to compensate the first-order and the second-orderrange migration, a non-coherent accumulation detection method combined the secondorder correction with multi-channel Radon filter group, a non-coherent accumulationdetection method combined the second order correction with Houghtransformation);The second category is to use the feature that higher order spectrum canmake gaussian noise be zero to eliminate the noise so as to improve the SNR greatly.According to this, this paper takes out another new method for weak signal targetdetection by comparing the mutual double spectrum of echo signal and radar’stransmitting signal with the double spectrum of echo signal. In this article, matlabsimulation is carried out in uniform simulation parameters, and the simulation resultsproved that the proposed four methods can all significantly improve the SNR beforedetection by different degrees. Simulation in the fifth chapter further put the results ofall different accumulation detection methods of this paper into comparison. Theseresults are calculated by constant false alarm of the unit average (CA-CFAR) indifferent input SNR conditions and be drawn into the comparative graph of detectionprobabilities(Pd). From this graph we can see different detection ability of the various methods and sum up three laws:1.The coherent accumulation to improve signal to noiseratio is better than that of the non-coherent accumulation;2. Multiple pulseaccumulation is superior to the single pulse (the time-frequency analysis method andpulse pressure are all single pulse detection, so improving ability not better than othermethods).3. Any items in the range migration correction are better to improvesignal-to-noise ratio than no correction, and two items correction is better than only oneitems correction. These outcomes lay the foundations for the later improvement of theaccumulation detection method.