| With the multi-dimensional expansion of war,the working environment of radar is getting worse and worse.Especially with the gradual improvement of Digital Radio Frequency Memory(DRFM)technology,the new radar jamming technology embodies the characteristics of flexible and changeable styles and more and more accurate jamming,which makes the radar lose the necessary functions of active detection and other battlefield functions.Therefore,it is particularly important to identify the radar jamming type,which is the prerequisite for suppressing and eliminating jamming.In this paper,the algorithm of radar jamming signal recognition is studied,hoping to improve the living environment of radar in the battlefield.The classification of radar jamming needs to solve the following two problems: the extraction of characteristic parameters and the selection of classifier.The main research work is as follows:Firstly,the mathematical models of noise convolution signal,noise product signal,intermittent sampling direct transmitting interference signal,intermittent sampling repeated transmitting interference signal and intermittent sampling cyclic transmitting interference signal are built.Their interference mechanism to radar is analyzed.Their time domain,frequency domain and pulse pressure information are analyzed,and their simulation diagrams are given.Secondly,respectively from the time domain,frequency domain and fractal domain three Angle to extract features: extracting the time domain on the time-domain envelope fluctuation degree,time-domain envelope around the similarity,waveform entropy and four characteristic parameters of LZC complexity similarity to extract the frequency domain and frequency domain and frequency domain entropy norm two characteristic parameters,the fractal domain to extract the box dimension and information dimension of two characteristic parameters.Then,a recognition method based on multi-feature joint PCA-SVM is proposed.Then,a method of denoising self-coding network based on spectral characteristics is proposed,and the feasibility of the method is verified by simulation,and the strong generalization ability of the network is proved.Finally,this paper introduces the concept of singular value decomposition algorithm and kurtosis criterion,SSD algorithm of adaptive parameter selection is given,and proved the feasibility of this kind of based on kurtosis criterion parameter selection,this paper discusses the method of singular value decomposition of strong noise non-stationary nonlinear signal strong resolving power,and gives the best number of decomposition and best decomposition component search method.The definition and properties of 1.5 dimensional spectral estimation are given,and the suppression of white Gaussian noise is demonstrated.Finally,a new radar jamming recognition method based on adaptive SSD algorithm and 1.5D spectrum is proposed.This method has a high recognition rate even in low signal noise,and is compared with other methods in literature,which proves the superiority of this method. |
