A Study On The Structure Characteristics Of Canonical Bodies Based On The Clustering Of Scattering Center In Wide Angle Domain

The use of new weapons can achieve unexpected effects in modern war.It is particularly important to process radar data in time,accurately and quickly.The target scattering center model is an important tool to describe the electromagnetic scattering characteristics of radar targets.It is also the research basis of analyzing the target characteristics.The target scattering center distribution in the wide angle domain can reflect the precise geometric information of the target.The structure characteristics of the unknown target can be analyzed more intuitively through the cluster analysis and processing of th e target scattering center data,so as to identify and judge the unknown air target quickly.However,the traditional clustering method is often not ideal for radar data processing.Meanwhile,the manifold learning algorithm is a kind of nonlinear dimensionality reduction algorithm developed in recent year s,which has unique advantages in high-dimensional nonlinear data processing.The deep learning algorithm is relatively fast in data processing,and the adoption of the deep learning algorithm to automatically learn data features can also improve the clustering accuracy.In this paper,the manifold learning method and the clustering algorithm ar e combined to achieve the dimensionality reduction and clustering of the sc attering center data in the wide angle domain.In addition,two kinds of deep learn-based clustering algorithms of scattering center data are proposed and applied to the structural characteristics of canonical reflector,aiming to realize high performance association algorithm of scattering center wide-angle domain and provide technical support for parametric modeling of target scattering characteristics wide Angle domain.The main research contents are as follows:(a)In order to study the scattering characteristics of typical scattering centers,this paper analyzes several existing scattering centers model and reviews the research status of parametric model of typical scattering center.The shortcomings of the model are pointed out and the characterization ability of the model is improved based on the typical scattering center model.The analytical expressions of equilateral dihedral angle and equilateral trihedral a ngle are presented to perfect the applicable range of the typical scattering center model.(b)In view of the non-linear manifold structure of radar data,this paper adopts manifold learning method to reduce the dimensionality of radar scattered center data,and then analyzes it in combination with traditional clustering algorithm,which can represent the internal parameters such as the position,attitude and scale of the data.The experimental results show that the fusion algorithm is superior to the traditional clusteri ng method in the dimensionality reduction processing time of the wide-angle scattering center radar data,and the performance is also better.(c)In view of the high dimension and large amount of data of radar data,deep learning algorithm has more advanta ges in extracting data features,combined with the data-driven premise,this topic put forward two kinds of clustering methods based on the deep learning,which are the DEC algorithm and the N2 D algorithm using manifold learning method and indirect depth c lustering algorithm.The scattering center clustering algorithm based on deep learning can represent the detailed structural characteristics of canonical reflector more clearly and improve the algorithm performance.The experimental results show that the clustering algorithm based on deep learning can represent the data structure of high-dimensional radar data well and has better performance in clustering effect.