Analysis And Identification Of Subtle Features Of Radar Radiation Signals

In today's world,with the advent of the information age,electronic science and technology have developed rapidly.Military operations are also increasingly relying on electronic information,so countries around the world are constantly updating their combat equipment.Among them,the most important is the development and update of radar.Under the background of the emergence of various new systems and new types of radars,the electromagnetic environment has become more and more complex and volatile,which has brought great challenges to electronic countermeasures.Aiming at the problem of accurate identification of radiation sources in complex electromagnetic environments,this paper will analyze the high-order spectrum and wavelet transform based on the refined features of phased array radar sources,mainly envelope characteristics and phase noise characteristics,and combine them There is a supervised learning method to complete the radar radiation source identification task.Finally,this paper also proposes a radar source identification method based on radar signal intermediate frequency data and one-dimensional convolutional neural network.The main content is divided into the following sections: 1.The structural composition of phased array radar is studied and mathematical modeling is performed.The analysis and modeling of the phased array radar's transmitting system and antenna system are emphasized,and then the radar transmitting signal model is given according to the signal generation process and phased array antenna scanning principle,according to the modulation,mixing,amplification,beam synthesis and other processes,And simulations of several common radar radiation source signals.Finally,a receiver signal is formed by superimposing the receiver noise.2.The refined characteristics of radar radiation sources are studied.This paper focuses on the detailed characteristics of individual radar emitters from the time,frequency and transform domains.Firstly,the generation mechanism and characteristic model of envelope feature and phase noise feature are analyzed,then the signal envelope feature is extracted by Morlet wavelet transform,phase noise feature is extracted by indirect method of nonparametric bispectral estimation and bounding line integration,and finally by The spectral wavelet transform extracts spectral features and constructs an eight-dimensional feature vector.3.The method of radar source identification is studied.Aiming at the refined features of the extracted radar signals,an eight-dimensional feature vector is constructed.Three classification algorithms,K nearest neighbor algorithm,support vector machine algorithm,and stacked autoencoder network are used to classify and identify different radar radiation individuals.There are advantages and disadvantages,but can still achieve better classification recognition results.With a signal-to-noise ratio of 20 d B,the recognition rate of various features has reached more than 90%.4.Aiming at the problem of low recognition rate under the condition of low signal-to-noise ratio,a one-dimensional convolutional neural network model was proposed for classification and recognition based on the one-dimensional characteristics of the intermediate frequency data of the radar radiator signal.By adjusting the network parameters,a relatively high recognition accuracy rate is obtained.With a signal-to-noise ratio of 0 d B,the recognition rate can reach more than 90%,which is better than the previous classifier recognition effect.In addition,the network shows good robustness and noise immunity in the simulation experiments of high SNR data training and low SNR data test,and has achieved a good classification and recognition effect.