Research On Radar Target RCS Modeling And Recognition Technology

With the promotion of modern information warfare,radar automatic target recognition(RATR)technology has been developed greatly.The research on the electromagnetic scattering characteristics analysis of radar targets and recognition technology has attracted wide attention.In this paper,according to the requirements of national defense project,the radar target RCS modeling and RATR technology based on high range resolution profile(HRRP)are studied.The main work of this paper is as follows:Aiming at the complex or dense radar target,the radar target RCS modeling method is studied.The RCS synthesis of long-distance target is introduced.A RCS fitting method of close-distance target is proposed.The target is modeled by Matlab and FEKO,and trained by radial basis function networks.The modeling analysis of radar target can reduce the complexity and provide data samples and new features for subsequent recognition.The radar target recognition method based on feature weighting and manifold learning for HRRP is studied.The method of obtaining HRRP and simulation platform are introduced.Three characteristics of HRRP and its corresponding pretreatment method are discussed.A feature extraction method combining feature weighting and manifold learning is proposed.The feature weighting is performed by using the weight of the feature sensitivity,and further feature extraction is achieved by using the adaptive neighborhood preserving discriminant projection algorithm.The proposed method can reduce the feature dimensions effectively,and also achieves good recognition performance and robustness.In order to further improve the recognition performance,the radar target recognition method based on multi-classifier fusion is studied.The common classifier learning algorithm and fusion algorithm are introduced.A multi-classifier fusion algorithm based on AdaBoost is proposed.Using K-nearest neighbor,correlation matching and support vector machine for the weighted fusion,and AdaBoost algorithm is used to fuse the classifiers.Finally,a two-level weighted voting method is used to construct the final fusion classifier.The proposed method can improve the recognition performance of single classifier effectively.