Radar Signal Intelligent Sorting And Threat Level Evaluation

The electronic countermeasure reconnaissance environment has become much more complicated.On the one hand,there are a large number of radar emitters in the time-space domain and frequency domain cause serious overlap of pulse signals;on the other hand,the interval pulse modulation parameters are complex and changeable and the radar function can be flexibly switched in electronic countermeasure environment.A new challenge has been raised for radar signal sorting.Constant innovation and development are needed by radar signal sorting to adapt to the future electronic countermeasure.In view of the problems of signal sorting in electronic reconnaissance environment,the main work of this paper is as follows.Firstly,a method for measuring the importance of pulse description feature is proposed.It adaptively selects highly separable features for radar sorting to improve the traditional radar pre-sorting algorithm.Secondly,deep learning network combines with pulse description words for feature extraction,and known radar database and semi-supervised theory are used to achieve a new radar sorting systems.Thirdly,the threat level of the selected radiation source results is evaluated.Finally,the algorithm of feature importance estimation sorting and threat level evaluation are implemented based on DSP6678 platform,data indicators meet the requirements of engineering design.From a large number of overlapped pulse signals in the environment,traditional presorting adopts fixed feature sequences for multi-parameter pre-sorting.In this paper,the feature entropy and mutual information values of pulse description words are calculted to estimate the degree of influence of features on sorting,and high-importance features are adaptively selected for K-means.And then the main-sorting algorithrm based on PRI is used to complete pulse deinterleaving,and the result of sorting and recongnition shows that the sorting method is effective.From the problem traditional pulse description words can not be combined with deep learning networks,and it is difficult to extract deep features of radar emitters.In this paper,traditional pulse description words are studied,and the matrix of single-pulse is proposed.The known radar database combine with the semi-supervised theory for feature matching to label the known radar pulse,and the importance of pulse description words are estimated to obtain highly separable features.Then this is reconstructed to form single-pulse feature matrix.The sparse autoencoder encodes is used for processing single-pulse features matrix,after that it is performed semi-supervised clustering for the encoded feature.Finally,the sorting result is added to the sorting radar database.Simulation results show that the scheme can effectively complete radar signal sorting.After sorting of the radar emitters pulse signal,it is necessary to evaluate the threat level of the radar emitters quickly and accurately,and processing the high-threat radar.In this paper,the traditional fuzzy expert evaluation mode can only express the two decision states.So the theory of intuitionistic fuzzy set is introduced to evaluate the threat level of the radar emitters.Based on the original analytic hierarchy process,the intuitionistic fuzzy number evaluation is performed on indicators and criteria,and a membership function is constructed to evaluate and analyze the threat level of the radar emitters parameters.Finally,the simulation of threat level of the radar emitters verify the effectiveness of the evaluation method.From the problems signal sorting needs to be processed in real-time.In this paper,the signal sorting algorithm,threat level evaluation algorithm,data interaction and interface control and communication is implemented based on DSP6678 platform.Finally,the system testing meet the engineering requirements.