【作者】 刘云涛；

【导师】 张文旭；

【作者基本信息】 哈尔滨工程大学 ， 信息与通信工程， 2023， 硕士

【摘要】 在雷达有源欺骗干扰技术得到迅猛发展的今天,其中的间歇采样转发式干扰在战场上得到了广泛的应用,诸如近两年内的战争,间歇采样转发式干扰就在战场中起到了至关重要的作用。间歇采样转发式干扰具有多假目标欺骗的效果,是一种很好的电子对抗的手段。为了让雷达系统在这种情况下能够正常的检测到真实的目标,就必须在雷达系统中加入干扰识别的功能,使其能够在收到假目标干扰时正确识别出真假目标,让雷达能够拥有出色的识别和对抗干扰的能力,使作战方在战场上占据主动,获得先机。因此,为了能更好的实现对干扰的识别,需要了解间歇采样转发式干扰的原理、实现以及种类等先验知识,为干扰识别提供依据。本文以间歇采样转发干扰为研究对象进行识别方法研究,提出了一种基于矩形积分双谱的信号特征提取并结合基于高效通道注意力机制(Efficient Channal Attention,ECA)的一维残差神经网络的识别方法,并进行了大量的仿真验证,验证特征提取方法的有效性,主要工作如下:首先,阐述间歇采样转发干扰识别的意义和背景,对该方向的国内外发展现状进行详细的介绍和总结。对基于数字射频存储技术(Digital Radio Frequency Memory,DRFM)的有源干扰进行详细的介绍,然后开展针对基于DRFM的间歇采样转发式干扰的原理研究。对基于DRFM的间歇采样转发式干扰产生原理进行研究,并构造间歇采样转发干扰的模型,最后进行基于FPGA硬件平台的工程实现。其次,针对目前对于间歇采样转发干扰下的回波信号干扰特征提取方法的弊端,提出了基于积分双谱的特征提取方法。该方法首先对回波信号求取了双谱,由于双谱是三维特征,为了降低计算量,在对双谱进行二次特征提取,根据特定的积分路径求取积分,将数据从三维降到一维,在保留信号重要特征的同时,最大限度上减小数据总量,从而减小运算量,方便简化识别网络结构。再次,针对不同干扰参数下的间歇采样转发干扰,对目标信号和干扰信号同时进行积分双谱的特征提取。以间歇采样重复转发式干扰为研究对象代表,对不同干扰参数下的干扰信号进行积分双谱的特征提取,同时对真实目标回波进行积分双谱的特征提取,将二者的积分曲线进行对比,并将积分双谱特征与时频特征进行了对比,说明此方法理论上的优势。最后,对一定参数范围内干扰进行了数据集构建,然后采用基于ECA的一维残差神经网络对数据集进行处理和分类,对干扰信号识别对比分类识别的各项指标,实现对间歇采样转发式干扰的分类和识别。更多还原

【Abstract】 Today,with the rapid development of radar active deception jamming technology,the interrupted sampling and forwarding jamming has been widely used in the battlefield,such as the Russian-Uzbekistan war in the past two years.The interrupted sampling and forwarding jamming has played a vital role in the battlefield.The interrupted sampling and forwarding jamming has the effect of multiple false targets deception,and is a good means of electronic countermeasures,In order to enable the radar system to detect the real target normally in this case,it is necessary to add the jamming recognition function in the radar system,so that it can correctly identify the true and false targets when receiving the false target jamming,so that the radar can have excellent anti-jamming capability,so that the combat side can take the initiative in the battlefield and gain the first chance.Therefore,in order to better realize the recognition of jamming,it is necessary to understand the principle,implementation and type of interrupted sampling and forwarding jamming,and provide a basis for jamming recognition.This article focuses on the research of recognition methods for interrupted sampling and forwarding jamming.A signal feature extraction method based on integral bispectrum and a one-dimensional residual neural network based on Efficient Channel Attention(ECA)are proposed,and a large number of simulations have been conducted to verify the effectiveness of the feature extraction method.The main work is as follows:Firstly,the significance and background of interrupted sampling and forwarding jamming recognition are elaborated,and the current development status at home and abroad in this direction is introduced and summarized in detail.Provide a detailed introduction to active jamming based on Digital Radio Frequency Memory(DRFM)technology,and then conduct research on the principle of interrupted sampling and forwarding jamming based on DRFM.Research the principle of interrupted sampling and forwarding jamming generation based on,construct a model of interrupted sampling and forwarding jamming,and finally carry out engineering implementation based on FPGA hardware platform.Secondly,the feature extraction method based on integral bispectrum is introduced for the radar received signal under interrupted sampling and repeater jamming.This method calculates the integral bispectrum of the radar received signal after pulse compression processing,and classifies and identifies different target signals by using the difference of the integral spectrum curve between different signals.Thirdly,for the intermittently sampled and forwarded jamming,the target signal and jamming signal are extracted by the integrated bispectrum feature.Taking the interrupted sampling and repetition-forwarding jamming as the representative of the research object,the integrated bispectrum feature extraction is carried out for the jamming signal under different jamming parameters,and the integrated bispectrum feature extraction is carried out for the real target echo.The integration curves of the two are compared to reflect the feature difference and form a data set.Finally,the method of machine learning or deep learning is used to process and classify the data set to achieve the classification and recognition of jamming signals and target signals.By comparing the various indicators of classification and recognition,the classification and recognition of interrupted sampling and forwarding jamming is realized.更多还原