Intelligent Inference Technology Of Jamming Decision Of Radar Active Jammer

As an indispensable component of cognitive electronic warfare,radar anti-jamming plays an important role in modern warfare.With the increasing application of electronic information equipment in war,the battlefield needs more and more reliable,fast and adaptive anti-jamming methods.The research on anti-jamming methods and technologies has a large number of research results.Limited by the level of radar transmitting,receiving and signal processing,it is difficult to make breakthrough progress in the short term.Giving full play to the anti-jamming ability of existing anti-jamming methods has become a new breakthrough in radar anti-jamming.Fast and accurate anti-jamming decision-making method can directly affect the process of the war situation and grasp the initiative of the war.The existing anti-interference decision-making methods based on game theory or deep learning have made great progress in the accuracy and timeliness of decision-making.However,these decision-making methods depend on the processing results of jamming signals.In principle,radar anti-jamming can only respond passively,so that anti-jamming always lags behind jamming.Realize the intelligent inference of the jamming decision of the radar active jammer,and can predict the jamming decision of the other party first when transmitting the radar signal,so that the radar can prepare the anti-jamming strategy in advance and design the waveform according to the jamming decision information,so as to increase the antijamming means and improve the anti-jamming ability;It can retrieve the jamming decision-making system of the opponent’s jammer,obtain the information of the opponent’s jamming equipment,and optimize our radar equipment.There is no public research on the inference of jamming decision of radar active jammer at home and abroad.Firstly,by summarizing and analyzing the existing jamming decision-making process,jamming decision-making principle and jamming decisionmaking method of radar active jammer,this thesis summarizes that the jamming decisionmaking inference problem of radar active jammer is a nonlinear system identification problem.Secondly,through the analysis,it is concluded that the traditional mathematical modeling method is not suitable for the nonlinear system identification of radar jammer jamming decision-making,and the system identification method based on neural network deep learning is a possible direction.Then,it studies the whole process of radar jammer countermeasure,jamming decision system and jamming decision process,puts forward three equivalent models of radar active jammer,such as dialogue process model,multi-input and multi output nonlinear system model and Markov decision process model,and puts forward the evaluation method of jamming decision inference model.Finally,based on the analysis and modeling results,a problem-solving method based on natural language processing,nonlinear system inversion and inverse reinforcement learning is proposed.Using the radar jammer countermeasure signal samples generated by simulation,the jammer jamming decision-making system modeling based on dialogue system,sequence to sequence nonlinear system inversion and depth maximum entropy inverse reinforcement learning is realized,the jammer jamming decision-making process inversion is realized,the intelligent inference of radar active jammer jamming decisionmaking is realized,and the feasibility and effectiveness of the algorithm proposed in this thesis are verified.