Research On Multi-Function Radar Working Mode Recognition Based On Machine Learning

Electronic Warfare(EW),also known as Electronic Counter-Measures(ECM),is an important component of modern warfare.Its importance has become increasingly prominent in modern warfare,and has become a key factor in the struggle for air and sea supremacy,even to a large extent determining the outcome of a war.Multi-function radars have high flexibility and agile parameter change capabilities,and are therefore widely used in modern warfare,greatly increasing the complexity of the electromagnetic environment on the battlefield,and bringing considerable challenges to reconnaissance in electronic countermeasures.To seize the initiative on the battlefield and provide support and basis for our own actions,effective recognition of radar working modes by the reconnaissance side has become an urgent task.Identifying the radar working modes is an important basis for our side to assess the threat on the battlefield and take appropriate countermeasures,which is conducive to firmly grasping the initiative on the battlefield,and is a key factor in winning the electronic warfare and even the entire war.This article focuses on the research of multi-function radar working mode recognition,the main research contents are as follows:1.The forward mechanism of multi-function radar is studied.The modeling of multifunction radar system is explored from the aspects of antenna system,spatial wave position arrangement,and scheduling.The transition rules of multi-function radar working modes are analyzed from the perspective of radar,which serves as the basis for the reconnaissance side to perform the working mode recognition task.The signal hierarchy of multi-function radar signals,including waveform units,tasks,working modes,and functions,is analyzed,and the switching criteria between radar working modes are elucidated.2.Based on clustering methods,the classification of multi-function radar working modes is studied in the absence of prior information.Firstly,common clustering algorithms are discussed from the aspects of principles and implementation steps.Secondly,the performance evaluation indicators of clustering algorithms,including internal evaluation indicators and external evaluation indicators,are explored,and the applicable scenarios of the two types of indicators are analyzed and discussed.Finally,in combination with the application background of multi-function radar working mode recognition,clustering algorithms are applied to the Pulse Description Word(PDW)dataset for clustering experiments,and the internal and external evaluation indicators of clustering algorithms are comprehensively compared and analyzed.3.Based on deep learning models,the recognition of multi-function radar working modes is studied under the condition of prior information.The structure composition and principles of convolutional neural networks and long short-term memory networks are discussed and analyzed,and the rationality of applying them to multi-function radar working mode recognition is explored.Finally,based on the above two network models,simulation experiments for multi-function radar working mode recognition are conducted at the pulse level and sequence level,respectively,and the effectiveness and robustness of the algorithm are verified.