Research On Target Simulation And Recognition Of Midcourse Ballistic Radar In Complex Scene

With the increasing complexity of modern battlefield environment,the improvement of information technology is very important.At present,many countries have developed advanced ballistic missile penetration technology.Taking the middle of the trajectory as an example,the missile wi ll release a variety of decoys and form a complex scene with a variety of micro motion characteristics.The main task of antimissile system is to identify the real warhead and decoy accurately and timely,so that the air defense missile can intercept accur ately.Radar is the equipment of the anti missile system to detect the target.The content of this paper is to simulate the process of the anti missile system radar receiving the echo signal,and distinguish the different types of targets accurately.The m ain work is as follows:(1)In this paper,the modeling of middle trajectory and motion in complex scene is studied.Eight types of true and false targets are established,and the middle course motion process of ballistic missile is simulated,including trajectory and fretting characteristics of different targets.(2)The generation of RCS sequence signal and the extraction of characteristic parameters are studied.Firstly,the static RCS library is established for eight kinds of targets,then the RCS sequences of all targets are gener ated by the simulation of trajectory motion characteristics in the middle of trajectory,and 10 kinds of features of RCS sequences are extracted for target recognition based on feature parameters.(3)The generation of HRRP signal and the extraction of its characteristic parameters are studied.The full angle HRRP data of eight kinds of targets are obtained by CST software simulation.HRRP needs energy normalization.Combined with RCS sequence generation,a GUI software for generating target signal database is developed.PCA and LDA are used to extract feature parameters for target recognition.(4)The target signal recognition based on feature parameters and depth learning is studied respectively.On the one hand,the target recognition based on the characteristic parameters mainly uses three classification recognition algorithms: logistic regression(LR),random forest(RF)and support vector machine(SVM).The target recognition experiments of RCS sequence and HRRP signal are carried out under different sample numbers and different signal-to-noise ratios.The experiments show that the more samples,the higher the signal-to-noise ratio,the higher the recognition accuracy.On the other hand,deep learning technology is a hot direction in speech recognition,character recognition,image recognition and other fields.Radar signal is similar to speech signal,so we can try to use deep learning algorithm to do target recognition in this scenario.This paper studies the use of MLP,GRU,Bi-GRU,CNN,FCN,Res Net18 and other network algorithms.Among all the recognition algorithms,Res Net18 has the highest average accuracy for radar signal classification.