雷达有源欺骗干扰多维特征提取与识别技术研究

发布时间：2019-01-01 15:00

【摘要】：雷达作为现代战争的“眼睛”,其性能好坏直接影响战争的胜负,随着电子频谱的日益复杂,多种干扰的出现严重影响其工作性能,雷达的生存环境受到严峻挑战,而其中又以有源欺骗干扰为甚。雷达抗有源欺骗干扰研究已经成为当前军事研究的重要热点。雷达抗有源欺骗干扰一般是通过提取干扰信号的特征参数,经过分类识别,得到干扰信号类型,然后有针对性的采取抗干扰措施。本文以常见雷达有源欺骗干扰在多维变换域的特征提取与识别为研究内容,主要进行了如下的工作:(1)根据雷达工作原理,分析了常见雷达有源欺骗干扰信号产生过程并建立数学模型进行仿真分析,为进一步对干扰信号的特征提取做准备。(2)分析七种不同有源欺骗干扰信号在频域、时域、图像域等多维变换域的差异,根据这些干扰信号特点,利用时频变换、小波变换、波形分析等手段,提取了一系列具有区分性大、噪声敏感度低的特征参数,并通过仿真,给出了对每种特征参数在不同干扰噪声比下的分布特点。(3)在提取多维特征参数的基础上,运用逆云模型、统计判决树、神经网络的方法实现对干扰信号的有效识别,对识别结果进行仿真分析,得出不同识别算法优劣。统计判决树在低干噪比的条件下识别率低,逆云模型由于采用了最小距离计算方法,对有些干扰的识别率较低,神经网络的识别率高,但需要大量的先验信息。针对不同的识别算法特点,利用F参数、类内类间距离、特异性参数等特征评价指标,进行特征选择,优化干扰信号的识别率,取得了一定效果。(4)正确识别出干扰类型后,以RVGPO干扰为例,有针对性地进行抗干扰研究。分析回波信号与干扰信号在时频域的差别,利用模态分解的方法对干扰信号分解重构,并在此基础上,分别采用PCA和小波变换的方法,改进RVGPO的抗干扰性能,并给出仿真结果分析,取得较好的抗干扰效果。
[Abstract]:Radar, as the "eye" of modern war, its performance directly affects the success or failure of war. With the increasing complexity of electronic spectrum, the appearance of multiple jamming seriously affects its working performance, and the survival environment of radar is severely challenged. And among them is active deceptive jamming is very much. The research of radar anti-active deception jamming has become an important focus of military research. Radar anti-jamming is generally by extracting the characteristic parameters of the jamming signal, after classification and identification, the type of jamming signal is obtained, and then anti-jamming measures are taken. In this paper, the feature extraction and recognition of common radar active deceptive jamming in multidimensional transform domain are studied. The main work is as follows: (1) according to the principle of radar, The generation process of radar active deceptive jamming signal is analyzed and the mathematical model is built to prepare for further feature extraction of jamming signal. (2) analyze seven kinds of active deceptive jamming signals in frequency domain, time domain, According to the characteristics of these interference signals, a series of characteristic parameters are extracted by means of time-frequency transform, wavelet transform, waveform analysis and so on. The distribution characteristics of each characteristic parameter under different interference noise ratio are given. (3) on the basis of extracting multi-dimension characteristic parameters, the method of inverse cloud model, statistical decision tree and neural network is used to realize the effective recognition of interference signal. The recognition results are simulated and analyzed, and the advantages and disadvantages of different recognition algorithms are obtained. The recognition rate of the statistical decision tree is low under the condition of low signal-to-noise ratio. The inverse cloud model has a low recognition rate for some disturbances and a high recognition rate for neural networks due to the minimum distance calculation method. However, a large amount of prior information is needed. According to the characteristics of different recognition algorithms, using F parameter, inter-class distance, specificity parameter and other characteristic evaluation indexes, the feature selection is carried out to optimize the recognition rate of interference signal. (4) after correctly recognizing the type of interference, taking RVGPO interference as an example, the anti-jamming research is carried out. The difference between echo signal and interference signal in time-frequency domain is analyzed. The interference signal is decomposed and reconstructed by mode decomposition method. On the basis of this, the anti-jamming performance of RVGPO is improved by using PCA and wavelet transform, respectively. The simulation results are analyzed and good anti-jamming effect is obtained.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TN974

【相似文献】