Research On Reconnaissance And Jamming Decision Method In Electronic Warfare Environment

Due to the rapid development of multi-field technology,radar has gradually developed from a single system to a more complex multi-function system in the modern battlefield.The flexible working state makes it have more anti-jamming measures and stronger anti-jamming ability,while the traditional electronic countermeasures are difficult to adapt to more complex countermeasures.Cognitive electronic countermeasures is proposed to cope with the complex battlefield environment and changing radar operating mode.The jamming system based on cognitive electronic countermeasures mainly includes four steps: reconnaissance,analysis and evaluation,jamming decision making and jamming.Through reconnaissance,the jammer can dynamically perceive the battlefield environment and the change of the working state of the countermeasure radar,and provide reliable information basis for jamming decision.According to the information obtained by dynamic reconnaissance,the jamming decisionmaking link can formulate more reasonable and effective jamming strategy when dealing with dynamic multi-functional radar,and achieve better jamming effect.Therefore,reconnaissance and jamming decision is an indispensable part of cognitive electronic countermeasures.The thesis mainly studies from the two links of reconnaissance and interference decision.First of all,for the reconnaissance link,this article explains the pulse description word measurement principle,analyzes the pulse signal sorting algorithm,the pre-sorting algorithm based on hierarchical clustering and the main sorting algorithm of sequence difference,and simulates the sorting algorithm to verify Therefore,it still has good sorting performance under certain pulse loss and noise conditions.On this basis,the FPGA+ARM hardware architecture is used to realize the reconnaissance function,the pulse description word measurement is realized in the FPGA,the sorting algorithm is realized in the ARM part,the function verification is carried out on the PYNQ evaluation board,and the communication with the upper computer is realized.Uploading the results to the host computer for display is of reference significance for engineering applications.Secondly,in order to further obtain the working state information of multi-functional radar in reconnaissance,the working mode identification of multi-functional radar should be carried out after sorting.In this paper,the multi-function radar model is constructed,the hierarchical structure from pulse signal to working mode is proposed,and the method of wave position arrangement is analyzed.The four common working modes are analyzed and simulated from three aspects: function description,wave potential scanning mode and pulse signal characteristics.Through SVM method and BP neural network method,the recognition of working mode is realized respectively.When SVM is used for recognition,aiming at the kernel function selection problem,experimental simulation is designed to compare the performance of four kinds of kernel functions under the same conditions,proving that gaussian kernel function is more suitable for this scene.Because the parameter setting has great influence on the classification recognition effect,the parameter optimization methods are analyzed,including the classical grid search algorithm and heuristic algorithm,and the performance of several common heuristic algorithms in parameter optimization is compared,which verifies that the genetic algorithm has better performance than other algorithms.The performance comparison of SVM and BP neural network under different measurement errors proves that BP neural network has better anti-interference ability.Finally,in the process of interference decision making,the method of cognitive interference decision making based on reinforcement learning is studied,and the selection of interference target,interference pattern matching and interference resource allocation are introduced.Then the principle of reinforcement learning algorithm is introduced and the model is analyzed.The similarity between the interference decision system and the reinforcement learning model is illustrated by analogy.The Q learning algorithm in reinforcement learning is used to realize the cognitive interference decision system.A confrontation scenario is constructed and the effect of q-learning-based interference decision system is verified.The optimal strategy is obtained through multiple learning iterations,and the obtained optimal strategy is analyzed to prove the rationality of the method.