| 摘要: |
| 传统雷达多次回波分类方法容易受到目标类型和幅度起伏特性等因素的影响，其泛化性和准确性难以满足雷达装备实际需求。针对该问题，提出了一种采用集成装袋树的雷达多次回波分类方法。该方法首先对雷达多脉冲回波数据进行幅度对数变换和相邻脉冲幅度补齐预处理操作，然后利用决策树算法从标注的训练数据中学习雷达多次回波在脉冲维的幅度起伏特征，最后通过多个分类器的集成实现对雷达多次回波的准确分类。实测雷达数据验证结果表明，所提方法分类准确率达到了95.9%，可有效提升雷达多次回波的分类性能，并且不依赖于经验门限的特性，增强了其泛化能力。 |
| 关键词: 雷达信号处理 多次回波分类 集成装袋树 |
| DOI：10.20079/j.issn.1001-893x.220912001 |
|
| 基金项目:国家自然科学基金资助项目(62201455);陕西省教育厅科研计划项目(22JK0566); 陕西省科学技术协会青年人才托举计划项目(20230112) |
|
| A Radar Multiple Echoes Classification MethodBased on Ensemble Bagging Trees |
| PAN Meiyan,CAI Xingyu,ZANG Huikai,XUE Jian |
| (1.Xi揳n Electronic Engineering Research Institute,Xi揳n 710100,China;2.School of Communication and Information Engineering,Xi揳n University of Posts and Telecommunications,Xi揳n 710121,China) |
| Abstract: |
| The traditional radar multiple echoes classification method is easily affected by target types and amplitude fluctuation characteristics,and its generalization and accuracy are difficult to satisfy the actual needs of radar equipment.For this problem,a radar multiple echoes classification method based on Ensemble Bagging Trees is proposed.The method firstly performs amplitude logarithmic transformation and adjacent pulse amplitude complementing preprocessing operations on the radar multi-pulse echo data,and then uses the Decision Trees Algorithm to learn the amplitude fluctuation characteristics of the radar multiple echoes in the pulse dimension from the labeled training data.Finally,accurate classification of the radar multiple echoes is realized through the ensemble of multiple classifiers.The verification results of measured radar data show that the proposed method搒 classification accuracy reaches 95.9%,it can effectively improve the classification performance of radar multiple echoes,and its generalization ability is enhanced because of its independence on empirical threshold. |
| Key words: radar signal processing radar multiple echoes classification ensemble bagging trees |
