Research On Key Technology Of Radar Signal Intra-pulse Features Analysis And Recognition

Radar signal sorting and recognition,as a key link in radar electronic warfare,plays an important role in the electronic warfare reconnaissance system.With the rapid development of information technology,the competition in the electronic countermeasures field has become increasingly fierce,and the electromagnetic environment of the battlefield has become increasingly complex,especially in the large-scale equipment application of various new system radars represented by Low Probability of Intercept(LPI)radars,which bring many problems and challenges in radar signal sorting and recognition.This paper focuses on the key technology of radar signal intra-pulse features analysis and recognition,proposes three radar signal modulation recognition algorithms,and designs a verification system of radar signal analysis and recognition.The main work includes:1.Aiming at the shortcomings of traditional intra-pulse modulation feature extraction algorithm,such as high computational complexity,poor feature parameters universality and low recognition accuracy,a radar signal modulation recognition algorithm based on fractional order domain is proposed.Firstly,Fractional Fourier transform is used to obtain the fractional-order domain waveforms of radar signals under different orders.Then,multiple sets symmetric Holder coefficients features are extracted.Finally,the automatic modulation and recognition of radar signals is realized via using support vector machine classifier.Simulation results show that the proposed algorithm is simple in calculation,accurate in recognition accuracy and easy to implement.2.In order to further improve the recognition accuracy and anti-noise performance of the composite modulated radar signal,a radar signal modulation recognition algorithm based on multidomain joint feature is proposed.Firstly,the time-frequency image of the radar signal is obtained by the introduced Multi-synchro-squeezing transform.Then the image processing technology is used to preprocess the time-frequency image,and multi-domain feature parameters,such as texture features,moment features,power spectrum parameter features and square spectral complexity features,are extracted from the time domain,frequency domain and time-frequency domain.Finally,the automatic modulation and recognition of radar signals is completed via using support vector machine classifier.Simulation results show that the proposed algorithm has excellent antinoise performance and strong anti-aliasing performance while dealing with composite modulated radar signals.3.Aiming at the problems of limited feature representation,insufficient versatility,poor recognition rate under low signal-noise-ratio(SNR),while deep learning model exists the problems of complex parameters,high requirements on data samples and long training time.A radar signal modulation recognition algorithm based on transferred deep learning is proposed.Based on the GoogLeNet and ResNet pre-training models,the algorithm re-offline training of time-frequency image of radar signals by means of transfer learning,which is used for automatic online recognition of radar signals after fine-tuning network parameters.Simulation results show that the proposed algorithm owns excellent anti-noise performance,good generalization ability and brilliant recognition performance under small samples.4.Aiming at the shortcomings of heavy theoretical simulation and light engineering application existed in radar signal analysis and recognition algorithms,a radar signal analysis and recognition verification system is designed.The system utilized AWG arbitrary waveform generator and broadband acquisition and analysis equipment is designed as a platform to realize various radar signal generation,pulse sorting,pulse analysis,data acquisition and modulation recognition.Experiment results show that the system has stable performance,good recognition effect and strong scalability,and has certain engineering application value.