Enhanced Radar Target Recognition for Terminal-Phase High-Dynamic Flying Objects via XGBoost

This paper introduces a novel radar target recognition method for terminal-stage high-dynamic flying objects, employing the XGBoost machine learning model. By leveraging dynamic Radar Cross Section (RCS) data, the proposed approach effectively classifies high-dynamic targets and distinguishes them from pseudo targets during the terminal phase. Key features are extracted from the dynamic RCS sequences, providing a robust framework that significantly outperforms traditional classifiers, including SVM, KNN, LR, and Naive Bayes, in terms of classification accuracy and reliability. Experimental results demonstrate that the XGBoost model achieves an accuracy of 99.8%. This study enhances radar target discrimination capabilities and has promising applications in defense systems.