Research On Radar Emitter Signal Feature Extraction And Identification Technology

Receiving different radar signals and identifying their intra-pulse modulation is a very important issue in modern electronic warfare.As the electromagnetic environment continues to deteriorate and the complex system radar gradually dominates,the feature extraction and identification of radar emitter signals are also facing severe tests.Extracting effective radar signal characteristics and designing high-performance classifiers is also an urgent problem to be solved.In view of the above problems,this paper studies the intra-pulse analysis technology of radar signals and neural network classifiers.The specific work content has the following aspects:The model of radar emitter signals of different modulationisestablished,and its characteristics are introduced.For the intra-pulse analysis technology,several main time-frequency analysis methods are described,and the simulation analysis is carried out to provide a basis for subsequent radar signal feature extraction.The feature extraction of the ambiguity function main ridge and the identification method of the radar signal intra-pulse modulation method based on GRNN are studied.Using the relationship between fractional autocorrelation ambiguity function to extract the ambiguity function main ridge of radar signal and rotation Angle,the symbolic time series analysis is introduced at the same time.In order to solve the problem of symbolic method parameter selection difficulties,this paper proposes a particle swarm optimization ambiguity function main ridge symbol entropy feature extraction method,the extracted symbol entropy and rotation angle constitute the characteristic parameters,finally analyzes the learning method of GRNN,the network is used as a classifier to classify andidentify characteristic parameter.Simulation results show that particle swarm optimization feature extraction method can effectively extract the symbol entropy of ambiguity function main ridge,the symbol entropy has good anti-noise performance and improves the recognition rate of radar signal intra-pulse modulation under the condition of low SNR.The extraction of time-frequency image features of radar signals and the identification of radar signal intra-pulse modulation based on convolution neural network are studied.The time-frequency image of radar signal is extracted by pseudo Wigner-Ville distribution and Choi-Williams distribution,the image processing method is used to reduce the dimension and noise of time-frequency image,time-frequency images of radar signals with different modulation modes are obviously different,which can be distinguished by time-frequency images.Convolutional neural network as a deep neural network has been widely used in the field of image recognition,in this paper,the convolutional neural network is introduced into the radar signal recognition field,and the convolutional neural network is used as a classifier to recognize time-frequency images of radar signals.The relationship between the training parameters and the recognition performance of the network is analyzed by simulation,the results show that the intra-pulse modulation method based on the convolutional neural network has better anti-noise performance and higher recognition rate when the SNR is low.