| With the continuous development of modern radar technology,the radar sources of various modulation modes are constantly updated and iterated,and the electromagnetic environment of today's battlefields is more complex and variable,so that the radar radiation source signals are diverse,various parameters are constantly changing,and conventional The radar emitter signal sorting method can no longer meet the modern and rapid electronic warfare battlefield situation,which makes the target reconnaissance identification increasingly difficult.Aiming at the problem of low accuracy and sensitivity to existing methods,a signal classification and recognition framework based on machine learning is constructed.A new radar signal sorting method is proposed to realize the radar source signal with low SNR.High correct rate sorting.The specific research work of this paper is as follows:1.The contribution of the data processing re-expression method to the radar signal sorting effect is studied.Because the radar pulse signal has different sampling widths and the pulse width parameters are different,the sampling points of each pulse are inconsistent,and the pulse data needs to be dimension-reduced.The improved re-expression algorithm is used to reduce the dimensionality of the time series signal.Therefore,the theoretical analysis of the symbolic aggregate approximation SAX method is needed.For the information loss problem of the method itself,the improved SAX symbolized BOP(Bag-of-Patterns)The combined method is used for theoretical research and analysis,and the influence of radar signal classification accuracy is verified by designing time series data before and after re-expression.2.The main research of this paper is the processing method of complex radar radiation source signal data intercepted in complex electromagnetic environment,that is,re-expression;in order to effectively select and characterize the radar signal after data processing to achieve accurate signal classification,A cross-validation experiment was carried out on the influence of the re-expression method proposed in this paper on the classification results,and the re-expression of the original data information loss and dimension reduction and denoising was described.At the same time,this paper extracts the two characteristics of sample entropy(SampEn)and approximate entropy(ApEn),and then selects the sample entropy as the radar signal classification feature based on the simulation results.This paper mainly studies the data processing method of complex radar radiation source signals intercepted in complex electromagnetic environment,that is,datare-expression;in order to effectively perform feature selection and feature extraction on radar signals after data processing,and to achieve accurate signal classification,The re-expression method proposed in this paper has done a cross-validation experiment on the impact of classification results,and the re-expression of the original data information loss and dimensionality reduction denoising is described.3.Fourier transform and normalization are performed on the received radar radiation source signal,then the pre-processed signal is re-expressed,and the sample entropy and power spectral entropy characteristics of the signal are extracted and compared with similar time series signals.Finally,the SVM is used to sort the 6 types of radar emitter signals.From the simulation results,it can be seen that when the signal-to-noise ratio is below 0dB,the average correct recognition rate of the six types of radar source signals is at least92.03%;when the signal-to-noise ratio is 10 dB,the six types of signals can be completely separated,verifying the The effectiveness and feasibility of the proposed method. |
