一种基于半监督高斯混合聚类的雷达信号侦察结果正确性评估方法  

An Evaluation Method for Validity of Radar Signal Reconnaissance Results Based on Semi-Supervised Gaussian Mixture Clustering

作　　者：李悦[1] 甘荣兵[1] LI Yue;GAN Rongbing(Southwest China Research Institute of Electronic Equipment,Chengdu 610036,China)

机构地区：[1]中国电子科技集团公司第二十九研究所,成都610036

出　　处：《电子信息对抗技术》2025年第2期32-41,共10页Electronic Information Warfare Technology

摘　　要：随着电子对抗的发展,各种新型电子对抗设备的应用使电子对抗中的电磁环境愈发复杂,也对电子侦察设备形成正确侦察结果的能力提出了更高要求。电子侦察能够获取敌方雷达辐射源的参数和位置等信息,是进一步实施电子干扰等对抗手段的基础。现有电子侦察效能评估方法难以评估非合作雷达侦察结果的正确性,为实现对非合作雷达侦察结果正确性的评估,进一步提升电子侦察效果评估能力,提出一种基于先验知识和应用半监督高斯混合模型(Semi-Supervised Gaussian Mixture Model,Semi-GMM)的侦察结果正确性评估方法。该方法假设能够通过一组高维高斯分布拟合来自不同种类被侦察雷达的雷达信号,并以雷达信号数据在各个高斯分布上的最大后验概率作为该雷达信号数据的评估结果。仿真结果表明,相较于侦察设备形成的雷达数据本身的侦察结果和基于最大综合相似度的评估方法所形成的评估结果,所提出方法形成的评估结果在真实来源上的准确率分别提升了23.70%和10.16%。With the development of electronic countermeasures,the application of various new electronic countermeasures equipment has made the electromagnetic environment in electronic countermeasures increasingly complex,and has also put forward higher requirements for the ability of electronic reconnaissance equipment to form correct reconnaissance results.Electronic reconnaissance can obtain parameters and location information of enemy radar radiation sources,which is the basis for further implementing electronic interference and other countermeasures.The existing electronic reconnaissance effectiveness evaluation methods are difficult to evaluate the correctness of non-cooperative radar reconnaissance results.To achieve the evaluation of the correctness of non-cooperative radar reconnaissance results and further enhance the evaluation ability of electronic reconnaissance effects,a reconnaissance result correctness evaluation method based on prior knowledge and the application of semi-supervised Gaussian mixture model(Semi-GMM)is proposed.This method assumes that radar signals from different types of reconnaissance radars can be fitted through a set of high-dimensional Gaussian distributions,and the maximum posterior probability of the radar signal data on each Gaussian distribution is used as the evaluation result of the radar signal data.The simulation results show that compared to the reconnaissance results of the radar data formed by the reconnaissance equipment itself and the evaluation results formed by the evaluation method based on maximum comprehensive similarity,the accuracy of the evaluation results formed by the proposed method on real sources has increased by 23.70% and 10.16%,respectively.

关 键 词：半监督学习 高斯混合模型 电子侦察 效能评估 

分 类 号：TN971.1[电子电信—信号与信息处理]