Effectiveness Evaluation Of Radar Jamming Based On Machine Learning

With the rapid development of modern radar system technology,radar countermeasure systems need to develop intelligent technology that combines adaptive reconnaissance,intelligent interference effectiveness evaluation,self-determination and adaptive interference.Interference effectiveness evaluation plays an important role in intelligent electronic warfare systems,and is the key factor for the system to achieve cognitive goals and intelligent interference technology.Therefore,it is necessary to study advanced and intelligent interference effectiveness evaluation methods.The traditional radar interference effectiveness evaluation methods have defects that contain too many human factors.Targeted introduction of machine learning algorithms and other methods is the main research direction of current interference effectiveness evaluation methods.The dissertation focuses on the problem of radar interference effectiveness evaluation and studies the combination of random forest algorithm and neural network algorithm,and studies the evaluation method based on vector similarity weighting.It mainly comprises:1)Aiming at the problem of radar interference effectiveness evaluation,the interference effectiveness evaluation and its key technologies in adaptive countermeasure model are analyzed.The necessity of applying machine learning algorithm to radar interference effectiveness evaluation is clarified.2)Aiming at the problem of evaluation indicator selection,from the aspects of radar reconnaissance information,radar interference pattern characteristics and radar anti-interference measures,after analysis of indicators that can reflect changes in radar performance,Interference purpose and changes in radar transmit signal parameters,the indicator vector of the evaluation algorithm input is obtained.3)Aiming at the problem of weight calculation based on vector similarity weighting evaluation method,the similarity is applied to effectively characterize the real-time changes of radar transmitter parameters.The cross entropy is introduced to represent the ambiguity of the similarity distribution space,and the stochastic gradient descent algorithm is used to solve the similarity fuzziness minimization in autumn,and the corresponding index weights are obtained.It effectively solves the difficulty of removing the human factors and the reasonable assignment of index weights.4)Aiming at the problem of effectiveness evaluation of radar jamming based on machine learning algorithm,the characteristics of the feature selection ability of the random forest algorithm are combined with the strong regression ability of the neural network algorithm for evaluation.This method combines the random forest algorithm and the neural network algorithm effectively,and has better evaluation performance than the single neural network algorithm evaluation model.The vector similarity weighting evaluation method is simulated.The simulation results show the effectiveness of the method for interference performance evaluation.The random forest-neural network algorithm interference performance evaluation is simulated.The simulation results show that the evaluation performance and the accuracy of the results are slightly improved compared with the single neural network algorithm evaluation model.The simulation compares the stability and generalization of the algorithm of AHP with that of stochastic forest-neural network,and verifies the validity and accuracy of the stochastic forest-neural network algorithm.The simulation compares the performance of the two evaluation methods with less data.The simulation results show that the performance of the random forest-neural network algorithm is stable,and the performance based on the vector similarity weighting evaluation method is poor.