Research On Radar Signal Separation And Interference Signal Recognition Technology

At present,modern electronic warfare is becoming increasingly fierce,and radar plays an important role in electronic warfare.With the rapid development of information processing technology,the role of radar system is becoming more and more significant.However,there are endless forms of interference against radar in actual electronic warfare.It is difficult to obtain information that is beneficial to us,so it is still necessary to adopt a targeted interference identification method to classify and identify interference patterns.At the same time,due to the increasingly complex electromagnetic environment,it is easy to form time-domain overlapping electromagnetic signals.In order to obtain useful and accurate information,it is necessary to separate the time-domain overlapping radar signals.In order to solve these two problems,this article focuses on the separation of radar signals in the time domain and the recognition of radar interference signals.The main research contents of this paper are as follows:Firstly,this paper proposes a single-channel chirp signal separation method based on variational modal decomposition for the time-domain aliased single-channel radar signals received by single-antenna reconnaissance aircraft on the electronic battlefield in recent years.This thesis uses the variational modal decomposition algorithm to successfully construct a virtual multi-channel signal,and combines the variational modal decomposition algorithm with the fast fixed point iterative algorithm to reconstruct the mixed single-channel chirp signal.The simulation results show that,compared with the empirical mode decomposition algorithm,the algorithm proposed in this paper can determine the number of separated signals,reduce the time for chirp signal reconstruction,and improve the performance of radar signal separation.Finally,this paper analyzes the influence of the chirp signal frequency on the separation performance of the variational modal decomposition algorithm.Secondly,this paper proposes an interference identification method based on random forest algorithm to solve the problem of low interference signal recognition rate.Analyze and select appropriate signal features and input them into the random forest algorithm to realize the classification and recognition of interference signals.The signals selected in the simulation of this paper are suppressed interference signals,smart interference signals and intermittent sampling interference signals.The time domain features of the above interference signals are extracted,and the time domain features with tags are extracted as the training data set for training.The model analyzes the test data set with unknown labels.The random forest algorithm can select samples independently,reducing the probability of erroneous data,and at the same time,it can prioritize the selection of the best interference characteristics for classification.Finally,the interference recognition performance obtained by the random forest algorithm and the support vector machine algorithm is compared,and the simulation is verified the effectiveness of the algorithm.Finally,because the manually extracted signal features are subjective and incomplete,this paper starts with the recognition of interference signals that does not require manual feature extraction,and proposes a deep learning method based on residual neural network to achieve radar interference.Supervised learning of signals.In this thesis,time-frequency analysis of radar interference signals is performed,and the time-frequency diagram of the interference signal is obtained.The Res Net50 network is used to train and test the time-frequency diagram of the interference signal,and the interference signal identification is completed.It can also achieve a good recognition effect under the conditions.Since both the algorithm and the random forest algorithm identify interference signals,the advantages and disadvantages of the two algorithms are finally explained.