面向雷达信号分选的自约束搜索密度聚类算法  

Self-constrained search density based clustering algorithm for radar signal sorting

作　　者：嵇志康 周子楠 李煊鹏 JI Zhikang;ZHOU Zinan;LI Xuanpeng(School of Instrument Science and Engineering,Southeast University,Nanjing 210096,China)

机构地区：[1]东南大学仪器科学与工程学院,江苏南京210096

出　　处：《系统工程与电子技术》2025年第1期62-69,共8页Systems Engineering and Electronics

基　　金：国家自然科学基金(61906038)资助课题。

摘　　要：针对雷达信号分选过程依赖先验知识、参数适配调优困难等问题,提出一种基于自约束搜索密度聚类的参数自适应信号分选方法。该方法在点序识别聚类结构(ordering points to identify the clustering structure,OPTICS)算法生成可达距离序列的基础上,引入一种启发式的自约束搜索机制,该机制能够自动分析数据集的内在结构,根据其数据特性自适应划分簇。通过自动调整超参数,该算法能够有效处理不同参数分布的脉冲描述字(pulse description word,PDW)数据。仿真实验表明,在无先验知识依赖情况下,所提算法在雷达信号的分选准确率和抗干扰能力方面均优于传统方法,干扰脉冲比例不高于60%的复杂电磁环境中雷达信号分选准确率达到98%以上。In response to the challenges of dependency on prior knowledge and difficulty in parameter adaptation and tuning in the radar signal sorting process,a parameter self-adaptive signal sorting method based on self-constrained search density clustering is proposed.This method leverages the reachable distance sequence produced by the ordering points to identify the clustering structure(OPTICS)algorithm and introduces a heuristic self-constraining search mechanism.This mechanism is capable of autonomously analyzing the intrinsic structure of a dataset and adaptively partitioning clusters based on their data characteristics.With the capability to automatically adjust hyperparameters,the algorithm efficiently processes pulse description word(PDW)data with diverse parameter distributions.Simulation experiments demonstrate that,without the dependency on prior knowledge,the proposed algorithm outperforms traditional methods in terms of accuracy and anti-interference capability in radar signal sorting,achieving an accuracy rate of over 98%in complex electromagnetic environments with interference pulse ratios not exceeding 60%.

关 键 词：信号分选 密度聚类 自约束搜索 参数自适应 

分 类 号：TN957.51[电子电信—信号与信息处理]