| With the rapid development of science and technology,the traditional method of spatial target identification based on infrared features and scattering features of spatial targets is no longer sufficient for all scenarios in the complex modern battlefield.As a result,many domestic and foreign scholars have started to focus on another important feature of spatial targets-the micro-Doppler feature.This feature arises from the unique micro-motion of the target,so the detection radar can invert the motion characteristics of the target based on this feature,and thus classify and identify the target.Based on the above background,in order to provide an effective technical means for missile attack and defense simulation,this paper mainly studies the modeling of the trajectory and micro-motion characteristics of the space target in the middle part of the ballistic path,the reconstruction of dynamic radar scattering cross section(RCS)and the extraction of micro-Doppler features as well as the target identification method based on these features,and develops the relevant software simulation module.The main contents are as follows:(1)The ballistic trajectory model and the micro-motion simulation method of the mid-flight segment of the space target are studied.The ballistic trajectory and micro-motion characteristics are simulated for eight different types of structural target models,such as three types of warhead targets,four types of decoys and a kind of mother nacelle.(2)A dynamic RCS sequence reconstruction method for complex space targets is proposed.Unlike the traditional single cone warhead model,the partial space target model established in this paper is a non-rotationally symmetric structure,so it is necessary to combine ballistic trajectory prediction and target micro-motion characteristics to obtain the dynamic attitude angle sequence of the target during flight,and then reconstruct the dynamic RCS sequence of each target during flight from the static RCS scattering data table of each target for micro-Doppler analysis.In this paper,the six-degree-of-freedom vector of the target is solved using the form of advection superimposed on micro-Doppler,and the radar line-of-sight direction is determined by the line of sight from the radar deployment coordinates to the target center of mass,and the two can be combined to obtain the line-of-sight angle of the radar wave irradiation direction on the target coordinate system.(3)Based on the dynamic RCS sequence,the generation of spatial target micro-Doppler time-frequency map and micro-motion feature extraction are realized,and the target identification verification is carried out.Firstly,we analyze and compare the advantages and disadvantages of linear and nonlinear time-frequency analysis methods,and select the short-time Fourier transform of linear time-frequency analysis method to study the micro-Doppler time-frequency map of spatial targets;then we analyze and compare the micro-Doppler feature extraction methods based on peak detection method,empirical modal algorithm and inverse Jordan transform method,and select the inverse Jordan transform method to study the micro-motion feature parameter extraction,and the simulation results are correct.Finally,the micro-Doppler time-frequency map and the feature parameters obtained from the previous steps are used as the basis for testing the effect of target recognition.The results show that the micro-Doppler feature parameters extracted by the inverse Jordan transform method are feasible for recognition,but the stability needs to be enhanced;while the Resnet50 neural network,using the time-frequency map as the input data and combining with the popular deep learning method,can effectively discriminate various targets with high stability,and the recognition rate can reach 92.88% at a signal-to-noise ratio of 0 dB. |
PDF Full Text Request