多模雷达信号分选算法研究

发布时间：2019-01-04 19:57

【摘要】：雷达辐射源信号分选是现代电子对抗和未来信息战中的首要技术,也是电子情报侦察系统和电子支援系统中的瓶颈技术。只有雷达侦察设备正确分选的基础上,才能有效识别辐射源信号,判定被截获雷达的类型和威胁等级,进而选择合适的对抗策略。随着电子战、信息战的激烈对抗和现代雷达技术的快速发展,雷达信号分选所面临的电磁环境日益复杂密集,采用低截获概率、脉内波形变换、多参数捷变等技术,具有多种工作模式的先进体制雷达也在逐渐取代信号形式简单、信号参数不变或只是缓慢变化的传统雷达。针对复杂密集电磁环境下先进体制多模雷达辐射源信号处理中存在的关键理论问题,本论文进行了探索性研究,获得如下研究成果:(1)分析多模雷达辐射源信号特征的基础上,讨论传统未知辐射源信号分选模型处理多模雷达信号存在的问题,提出一种新的针对复杂电磁环境下未知多模雷达辐射源信号的分选模型,以更有效利用截获雷达脉冲序列。(2)将数据挖掘理论引入到雷达信号分选技术,通过数据场理论描述雷达辐射源信号样本以抑制复杂电磁环境中高强度的噪声干扰和离群数据干扰,根据数据场等势线的分布情况建立信号样本嵌套结构供后续的层次聚类多参数分选算法使用,规避需要辐射源信号先验知识才能完成的参数设定环节。(3)提出利用层次聚类算法进行多模雷达辐射源信号分选,层次聚类通过层次构架模式,递归地对信号样本进行合并或分裂,最终形成一种嵌套的类层次结构或类谱系图,进而将属于同一辐射源不同工作模式的雷达信号归纳为同一谱系,可有效减少多模雷达信号分选过程中“增批”现象的出现。(4)利用云模型理论可以从模糊、随机、不确定的小样本数据中提取出定性概念的固有特征,提出一种基于云模型理论的分选结果有效性评估算法。算法中将每个层次聚类分选结果视为一个云模型,根据提出的评价判定准则比较、处理不同云模型间的隶属度。该算法可有效解决聚类算法类间分离度较差的问题,同时可反馈优化多参数聚类分选过程。
[Abstract]:Radar emitter signal sorting is the most important technology in modern electronic countermeasure and future information warfare, as well as the bottleneck technology in electronic intelligence reconnaissance system and electronic support system. Only on the basis of the correct sorting of radar reconnaissance equipment can the emitter signal be effectively identified, the type and threat level of the intercepted radar can be determined, and the appropriate countermeasures can be selected. With the fierce confrontation of electronic warfare, information warfare and the rapid development of modern radar technology, the electromagnetic environment of radar signal sorting is becoming more and more complex and dense. The techniques of low probability of interception, in-pulse waveform transformation, multi-parameter agility and so on are used. The advanced system radar with various working modes is gradually replacing the traditional radar with simple signal form, constant signal parameters or only slowly changing signal parameters. In view of the key theoretical problems existing in the signal processing of advanced multi-mode radar emitter in complex and dense electromagnetic environment, this paper carries out an exploratory study. The main results are as follows: (1) on the basis of analyzing the characteristics of multi-mode radar emitter signal, the problems of traditional unknown-emitter signal sorting model for multimode radar signal processing are discussed. A new sorting model for unknown multi-mode radar emitter signals in complex electromagnetic environment is proposed in order to make more effective use of intercepted radar pulse sequences. (2) data mining theory is introduced into radar signal sorting technology. Radar emitter signal samples are described by data field theory to suppress high intensity noise interference and outlier data interference in complex electromagnetic environment. According to the distribution of the isopotential lines of the data field, the nested structure of the signal samples is established for the subsequent hierarchical clustering multi-parameter sorting algorithm. In order to avoid the parameter setting process which requires prior knowledge of emitter signal, the hierarchical clustering algorithm is proposed for multi-mode radar emitter signal sorting. The signal samples are merged or split recursively to form a nested class hierarchy or pedigree diagram, and then the radar signals belonging to different working modes of the same emitter are grouped into the same spectrum. It can effectively reduce the phenomenon of "increasing batch" in the process of multi-mode radar signal sorting. (4) using cloud model theory, the inherent characteristics of qualitative concepts can be extracted from fuzzy, random and uncertain small sample data. An algorithm for evaluating the validity of sorting results based on cloud model theory is proposed. In the algorithm, the clustering and sorting results of each level are regarded as a cloud model, and the membership degree among different cloud models is dealt with according to the comparison of the evaluation criteria proposed. The algorithm can effectively solve the problem of poor separation between clusters and can be used to optimize multi-parameter clustering process.
【学位授予单位】：哈尔滨工程大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN957.51

【参考文献】