摘要
箔条干扰是一种典型的无源干扰措施,其扩散过程复杂且特征信息多变,现有的对抗方法无法兼顾准确率和普适性。针对这一问题,基于距离-多普勒(range-Doppler,R-D)二维图的距离、频率分布特征,提出了一种更为有效的抗箔条干扰方法。首先,采用均值漂移聚类算法分离目标与箔条的点集。然后,提取频偏和等新的特征信息辅助机器学习分类器完成整个扩散过程的对抗识别。最后,该方法被应用于某相参末制导雷达的大量抗箔条干扰实测数据。数据处理结果展示了箔条弹自打出到完全扩散整个过程中各干扰特征的变化情况,在此基础上讨论了特征的稳定性和对抗方法的可靠性。理论分析和实测数据处理结果都表明所提抗箔条干扰方法在整个扩散过程中能够准确地辨识箔条,因此所提方法抗干扰性能卓越、环境适应能力强。
A typical passive interference measure,the chaff interference,has a complex diffusion process and changeable characteristic information.The existing countermeasures do not take both the accuracy and the universality into account.To address this problem,a more effective anti-chaff jamming method based on the distance and frequency distribution features of the range-Doppler(R-D)diagram is proposed.Firstly,a mean-shift clustering algorithm to separate the point sets of target and chaff is employed,and then new feature information such as frequency offset sum and other new features to assist the machine learning classifiers in completing the confrontation recognition of the diffusion process is extracted.Finally,the proposed method is applied to a large amount of anti-chaff jamming real-life data of a certain coherent terminal guidance radar.The data processing results show the changes of each feature of the jamming during the whole process of chaff bombs from firing to complete spreading,and based on this,discuss the stability of the features and the reliability of the anti-jamming method.Both the theoretical analysis and the measured data processing results show that the proposed anti-chaff jamming method can accurately recognize the chaff during the whole diffusion process,and therefore the proposed method has remarkable anti-jamming performance and good environmental adaptability.
作者
王湖升
陈伯孝
叶倾知
WANG Husheng;CHEN Baixiao;YE Qingzhi(National Laboratory of Radar Signal Processing,Xidian University,Xi’an 710071,China)
出处
《系统工程与电子技术》
EI
CSCD
北大核心
2023年第7期2010-2021,共12页
Systems Engineering and Electronics
基金
国防科技基础加强计划(2019-JCJQ-ZD-067-00)资助课题。
关键词
箔条干扰
实测数据分析
特征提取
目标识别
机器学习
chaff jamming
measured data analysis
feature extraction
target recognition
machine learning