Intelligent Detection And Sorting Technology Of LPI Radar Signals In Complex Electromagnetic Environment

With the development of large-scale integrated circuit technology and the world-wide application of modern active phased array radar,LPI radars are carried by novel weapon platforms in various countries.Therefore,nowadays,LPI radar signal reconnaissance is becoming an important research filed of radar electronic countermeasures.The widespread use of LPI radar has forced radar reconnaissance to face the challenges of low SNR,agile waveform,and complex modulation types.Conventional statistics LPI radar signal detection and sorting techniques often require conditions such as a large number of pulses,high radar signal-to-noise ratio,constant pulse repetition interval,narrow radar signal bandwidth,and no hopping of radar signal carrier frequency.Therefore,conventional techniques face the problem that the detection probability and the sorting success rate are not high on LPI radar signals.At the same time,with the continuous progresses of computer technology,artificial intelligence and deep learning have become research hotspots recently.In order to avoid the limitations of conventional statistical radar signal detection and sorting algorithms in LPI radar signal reconnaissance,this thesis focuses on the research of LPI radar signal intelligent detection and sorting algorithms based on deep learning.This thesis analyzes the typical LPI signal.Researched the LPI radar intelligent detection method based on visibility graph and FCF convolution network.Researched the intelligent recognition method of intra-pulse modulation type based on link prediction and deep residual network.Finally,signal sorting method based on long and short-term memory deep learning network is discussed.The main work and contributions of the paper are as follows:1.The time-frequency domain characteristics of typical LPI radar signals are analyzed.Enumerated the limitations of traditional detection methods.An intelligent detection architecture for LPI radar signals is given.An intelligent detection method for LPI lightning signals based on visibility graph is introduced.With the density and eigenvalues of the visibility graph adjacency matrix as input,the FCF convolutional neural network is proposed.2.Considering the limitations of the traditional statistics method of identifying intra-pulse modulation type,the characteristics of the visibility graph of different types of intra-pulse modulation in LPI radar are studied.The link prediction of the nodes and edges of the graph domain is introduced,and a various methods of connection predictions are discussed.An intelligent identification method of LPI radar signal intra-pulse modulation type based on link prediction matrix+deep residual network is proposed,and the performance of different connection predictions is compared.3.The intelligent sorting method of LPI radar signal based on long and short-term memory deep learning network is studied,and the influence of the number of pulse description words loss,pulse loss rate and parameter estimation error on the algorithm is analyzed and discussed.The effectiveness of all the above algorithms has passed computer simulation,which verifies the effectiveness of the algorithm and gives the performance of the algorithm in different situations.The intelligent detection of LPI radar signal,the intelligent classification of LPI radar signal intra-pulse modulation and the intelligent sorting of LPI radar signal methods are proposed.