Research On The Effectiveness Evaluation Method Of Radar Emitter Signal Recognition

The result of radar radiator signal sorting and recognition directly affects the quality of radar countermeasure intelligence,which has received great attention from researchers in this field in recent years.The performance of this technology depends on two core elements,namely the identification feature parameters and the classifier.However,the effectiveness of the recognition system composed of these two core elements is unknown.Therefore,in-depth research on the evaluation of radar radiator signal recognition efficiency is very necessary and has practical significance.In view of the existing evaluation methods of radar radiator signal recognition effectiveness,there are common problems such as unreasonable index weight solution,limited evaluation methods,and high decision-making risks.In this paper,the identification feature parameters are fixed,and different identification classifiers are evaluated.From the perspectives of multi-attribute decision-making and machine learning,the core technologies involved in two different radar emitter signal identification effectiveness evaluation methods are discussed in depth.The research results are as follows:1.The effectiveness evaluation of radar emitter signal recognition based on multi-attribute decision-making mainly overcomes the difficulties of unreasonable index weight solution and high decision-making risk.In this method,firstly,the intuitionistic fuzzy number is introduced into the analytic network process method to weight the indicators,which reduces the risk of decision-making and solves the problem of interrelationship between indicators.Then,on the basis of interval hesitant fuzzy sets,the elimination and selection transformation method and the approximation ideal ranking method are combined,which considers the problem that the approximation ideal ranking method sets arbitrary positive and negative ideal solutions,and overcomes the problem of incomplete evaluation caused by the elimination and selection transformation method.Relevant experiments are designed,and the simulation results show that the method is effective and feasible,and the method can adjust the preference degree of the index according to the willingness of the decision maker,so as to obtain the evaluation results that meet the needs of the current scene.2.The effectiveness evaluation of radar emitter signal recognition based on neural network mainly overcomes the shortcomings of low applicability and inversion caused by the first method.In this method,the maximum information coefficient is used to describe the correlation between indexes,and the correlation matrix is obtained.Then the fast community detection algorithm is used to analyze the correlation matrix,and the number of community division indexes is obtained.Then,based on the above prior knowledge,a network evaluation model with binding and military significance is constructed,which adopts the Elman network structure based on the sparrow search optimization algorithm.Relevant experiments are designed,and the simulation results show that the maximum information coefficient and the fast community detection algorithm provide important reference value for the pre-setting of network parameters.Moreover,the Elman network optimized by the sparrow search algorithm has higher evaluation accuracy,stronger stability,better learning ability and faster convergence speed.Compared with the first method,it can be found that this method effectively solves the reverse order problem,and the application scope is wider.