Research On Individual Recognition Technology Of Radar Emitter

Individual recognition of radar emitter is the key to obtain strategic layout information of non partners and quickly grasp the initiative of war.Nowadays,the complexity of the battlefield electromagnetic environment makes it difficult to guarantee the stability and reliability of the traditional recognition methods based on single feature and single classifier.How to obtain effective individual features and identify them accurately in the complex electromagnetic environment has become the primary problem of radar emitter individual identification.Therefore,this subject focuses on radar emitter modeling,feature extraction,feature optimization and classifier design,and designs a set of radar emitter individual identification system based on multi-dimensional feature extraction and decision-making level fusion.The main research contents are as follows:Firstly,from the structure of radar emitter,the principle of radar transmitter is introduced,the noise characteristics of frequency source and the nonlinearity of amplifier are analyzed in detail,and the main source of unintentional modulation is phase noise.The method of adding unintentional modulation to the signal is studied deeply,and the mathematical model of unintentional modulation of different modulation signals is established,and the correctness of the model is verified by simulation in time domain and frequency domain.Secondly,aiming at the problem of the stability and reliability of radar emitter's unintentional modulation feature,a multi-dimensional unintentional modulation feature extraction scheme is designed.Based on signal time-frequency analysis,the scheme combines energy spectrum,information entropy,image processing and deep learning theory.According to the differences of unintentional features in different dimensions,it extracts the unintentional features in multi dimensions,and establishes the unintentional modulation feature set with comprehensive characterization for the use of feature optimization and classification recognition.Then,aiming at the effectiveness and the heterogeneity of the features,an optimization method of radar emitter unintentional features is proposed.In this method,KPCA algorithm is used to extract features from high latitude features,in order to remove the redundant information with high correlation in the features,and to achieve the enhancement of the features' interoperability.Then,the Relief F algorithm is used to analyze the feature weight,and the effectiveness of feature set is improved by eliminating low weight features.Finally,according to the non-linear characteristics and small sample characteristics of radar emitter identification data,the theory of non-linear support vector machine is introduced.Combining with the wolf pack intelligent optimization algorithm,the efficient selection of non-linear support vector parameters is realized,and the classification and recognition system of non-linear support vector machine based on wolf pack algorithm is established.In order to overcome the shortcomings of single feature and single classifier in stability and reliability,the D-S evidence theory is introduced,and an individual recognition system based on D-S evidence theory radar emitter is designed,which realizes the fusion of decision information under different features,and improves the overall performance of the recognition system.