Research On Allocation Tactics For Radar Jamming Based On Reinforcement Learning

In the current battlefield environment,with the rapid development of related software and hardware technology,radar is gradually evolving towards the direction of multi-function,and even has reached the requirements of multiple radar detection targets.Different from the fixed working mode of traditional radar,today's multifunctional radar is more inclined to adopt multiworking mode,and adjust the relevant parameters in real time according to the working environment and task requirements of the current radar itself.By comparison,the traditional interference system follows a fixed interference strategy,which has low real-time performance and can not meet the needs of modern war.In this paper,based on the background of penetration operations,the interference resources are allocated adaptively in different radar environments.The main work of the paper is as follows:Firstly,the relationship between the signal-to-noise ratio(SNR)and the detection probability is established by studying the radar action distance and detection probability.The radar signal processing model is established to explore the basic principles of the modules that affect the radar signal-to-noise ratio(SNR),and to quantitatively calculate the changes of the SNR of these modules.The working mode of multifunctional radar is analyzed,which lays a foundation for the establishment of radar confrontation environment.Secondly,several typical radar active jamming styles are introduced,which can be divided into suppression jamming and spoofing jamming according to different effects of radar system and simulation of jamming.The power criterion and probability criterion are used as the evaluation indexes of suppression interference and deception interference respectively.Combined with radar signal processing model,the influence of suppression jamming on signalto-noise ratio of different radar systems is given,and the detection probability of radar at different distances and suppression jamming is compared by simulation,so as to establish the evaluation model of deception jamming based on radar anti-spoofing probability.Finally,reinforcement learning algorithm is applied to radar intelligent jamming decision,radar environment and jamming resource are modeled,radar status,jamming maneuver and training reward are designed.The real-time allocation of jamming resources with the change of navigation distance and state of various multifunctional radars is simulated by using reinforcement learning algorithm.Combined with prior knowledge to further improve the efficiency of decision-making,DQN algorithm and PK-DQN algorithm are compared from the convergence speed and the penetration success rate of convergence,indicating that reinforcement learning algorithm can effectively complete the task of disturbance resource allocation.