Research On Methods Of Radar Signal Intra-pulse Modulation Recognition

Radar signal intra-pulse modulation recognition is a crucial technology in radar electronic warfare,and it plays an important role in radar reconnaissance system.However,with the rapid development of radar technology,the new system radars represented by the low probability of intercept radar are constantly appearing,which not only makes the modulation methods of the radar signals more and more diversified,but also makes the signal-to-noise ratio(SNR)threshold of the normal operation of the radar signal lower and lower.It results in the invalidation of some traditional algorithms which only aim at a few kinds of signals and have poor anti-noise ability.This poses a new challenge to the algorithms of radar signal intra-pulse modulation recognition.Therefore,in this paper,the problem of radar signal intra-pulse modulation recognition is deeply studied,and three different algorithms of radar signal intrapulse modulation recognition are proposed,including:First,an algorithm radar signal intra-pulse modulation recognition based on singular value entropy and fractal dimension is proposed.First,the time-frequency image of the signal is obtained by Choi-Williams distribution(CWD),and the singular value entropy of the timefrequency image is extracted.Then the box dimension and the information dimension of the signal spectrum are extracted to form the three-dimension feature vector.Finally,a classifier based on support vector machine(SVM)is used to realize the classification and recognition of radar signals.The simulation results of eight typical radar signals show that the proposed method is robust to noise and has a high recognition rate.Second,an algorithm of radar signal intra-pulse modulation recognition based on convolutional neural network is proposed.The algorithm firstly extracts the time-frequency image of the radar signal,and then through a series of image processing including twodimensional Wiener filtering,bilinear interpolation and Otsu method,the time-frequency image of the signal is changed into a binary image,and finally passed.A convolutional neural network is designed to realize the intra-pulse modulation recognition of radar signals.The simulation results show that the proposed algorithm can effectively identify twelve kinds of radar signals in low SNR.In addition,this paper introduces a new kernel function for Cohen class timefrequency distribution(CTFD).Simulation analysis shows that the CTFD with new kernel function is more suitable for extracting time-frequency images of radar signals.Finally,an algorithm of radar signal intra-pulse modulation recognition based on convolutional denoising autoencoder is proposed.In this algorithm,a deep convolution neural network is designed to classify the time-frequency images of signals by combining convolution denoising autoencoder and Inception network module.The algorithm has a simple process and does not require too much pre-processing.It can directly use the simple processed timefrequency image as the input of the deep convolutional neural network to realize the radar signal intra-pulse modulation recognition.Although the convolutional neural network used in this algorithm has many layers,it has fewer parameters and good classification performance.The simulation results of twelve kinds of radar signals show that the proposed algorithm has strong anti-noise performance and good generalization ability.