A Study On Behavior Analysis And Identification Technology Of Multifunction Phased Array Radar

Multifunction Phased Array Radar(MPAR)has the beam and the signal characteristics of agility,strong adaptability,which requires that the new Radar countermeasures system must also have adaptive and intelligent confrontation ability as the main characteristics of cognition.It is of great significance for the formation of a real-time adaptive jamming countermeasure strategy to develop a radar radiation source behavior-level intelligent countermeasure and accurately perceive the dynamic change of radar behavior state.According to the characteristics of MPAR multi-task parallel,high data rate and high intelligence in the future battlefield,this paper studies the MPAR behavior identification technology for the needs of radar behavior identification in cognitive electronic warfare.The main research contents of this paper are as follows:1.Combined with MPAR behavior characteristics and work scheduling rules,the syntactic structure model is optimized to add a layer that can represent the law of radar task execution,namely "radar section",and the corresponding relationship between each layer and radar behavior(or state)is analyzed;This paper summarizes the behavior characteristics of MPAR,analyzes the mapping relationship between radar characteristics and radar behavior,and provides theoretical support for subsequent MPAR behavior identification.2.To solve the problem of MPAR signal agility,a Bayesian change point detection algorithm is proposed.The joint probability of MPAR pulse characteristics is used to divide the function of the pulse train to correspond to the current behavior of the radar.In order to solve the problem that the traditional pulse partitioning algorithm needs complete and accurate pulse information and may disturb the original time series relation,the algorithm has certain effect.Simulation results show that the algorithm can accurately classify both conventional and unconventional pulse sequences.3.For "small sample" unlabeled radar signal data,an unsupervised C-means clustering algorithm based on principal component analysis and mean shift is proposed.Principal component analysis is introduced to determine the main pulse parameters,reduce the data volume,optimize the clustering center through data preprocessing and mean shift,and realize the fast and effective classification of MPAR pulse train.Experimental results verify the effectiveness of the algorithm.According to the logical mapping relationship between MPAR behavior and signal characteristics,a method of constructing behavior-feature matrix is provided to infer radar behavior.4.Considering radar based on "data" driver behavior intelligent identification requirements,artificial extraction oriented MPAR sequence and intercepting MPAR signal pulse parameters are proposed based on BP neural network and the convolution of the neural network which has the supervision behavior identification method respectively.The behavior of different composition radar data sets are constructed which can be applied under different conditions of intercepted radar signal analysis and processing and achieve good recognition effects effectively.