【作者】 田晓；

【导师】 唐斌；

【作者基本信息】 电子科技大学 ， 信号与信息处理， 2013， 博士

【摘要】 综合电子干扰是雷达的四大威胁之一，特别是随着数字射频存储器（DRFM）技术的产生和应用，由它构成的欺骗干扰机可以产生灵活多样式、对抗性强的多种欺骗干扰，对军用雷达构成了严重的威胁。此外，现有的雷达装备虽然都有一定的抗干扰措施（ECCM），但是由于不具备干扰感知功能，无法根据干扰环境尽快选取最优的抗干扰手段。为了提高复杂电磁环境下雷达的预警探测能力，本文结合欺骗干扰的产生机理及其对雷达的作用机理，研究雷达有源欺骗干扰的感知方法，为后续的雷达有源欺骗干扰抑制提供先验信息，以便更有针对性地调度雷达抗干扰手段进行雷达反对抗。本文首先分析了干扰的产生机理，然后通过从信号处理、信息处理、波形分集和干扰机指纹特征几个方面研究了欺骗干扰的综合感知方法。论文的主要工作和研究成果如下：（1）介绍了本文的研究背景，介绍了目前常用的抗干扰方法，对近年来国内外的雷达有源欺骗干扰感知技术研究现状进行了较为全面的综述。介绍欺骗干扰的产生机理和信号模型。首先介绍了DRFM欺骗干扰的产生机理，并进行了DRFM欺骗干扰的特性分析，然后介绍了基于干扰过程的拖引干扰的产生机理，主要包括距离拖引干扰、速度拖引干扰和距离-速度联合拖引干扰，最后研究了其它类型的欺骗干扰。根据干扰的产生机理提出了欺骗干扰的模型，为后面干扰感知算法研究奠定了基础。（2）通过干扰的作用过程建立针对跟踪雷达的典型拖引欺骗干扰的信号模型，利用多尺度分解理论进行欺骗干扰的感知算法研究。提出了基于小波分解和基于经验模式分解的欺骗干扰感知算法。第一种算法中利用小波分解得到高频细节分量和低频逼近分量，提取各阶分解向量的能量比作为特征参数，从而实现各种拖引干扰的识别和分类。第二种算法处理流程和前一种方法类似，通过经验模式分解得到本征模式函数分量，提取分解分量的特征参数实现三类欺骗干扰的感知。（3）通过频域和慢时域二维联合处理，提取二维积谱的非负矩阵分解向量特征和二维频谱的纹理特征，进行拖引干扰的识别。首先建立频域和慢时域的二维积谱矩阵，利用非负矩阵分解理论，提取分解矩阵中各阶向量的特征参数，用于欺骗干扰的感知。将图像处理中的纹理特征引入到欺骗干扰的感知算法中，提取二维频谱灰度图像的纹理特征进行欺骗干扰的感知。（4）通过信息域进行欺骗干扰的存在性检测，提出了2种基于幅度起伏特性的欺骗干扰感知算法，包括基于拟合优度的欺骗干扰检测算法和基于粒子滤波的欺骗干扰检测算法。第一种算法通过分析干扰机在释放干扰时雷达接收波束内的信号幅度起伏差异，建立了欺骗干扰存在性的检测流程，提出了两种拟合优度的欺骗干扰检测算法，并通过仿真比较了各种条件下的Anderson-Darling（AD）检测和修正的AD（MAD）检测算法的性能。第二种算法利用拖引干扰在捕获期内，目标与干扰的在同一个波束内不可分辨，导致接收信号混叠到相邻匹配滤波采样点，基于这一特性建立了欺骗干扰的信号模型，提出了基于似然比的检测流程，通过粒子滤波检测器实现了欺骗干扰的检测。（5）研究了基于波形分集理论的欺骗干扰感知算法，包括基于随机线性调频斜率的波形分集信号的欺骗干扰感知算法和基于混沌调相波形分集信号的欺骗干扰感知算法。针对线性调频信号和相位编码信号，分别采用广义似然比检测（GLRT）算法和Holder系数的特征提取方法，对欺骗干扰的检测和识别性能进行了数字仿真验证。（6）针对欺骗干扰在生成过程中干扰机存在二次调制，利用Volterra模型建立干扰机的非线性失真模型，提取了干扰机的功放指纹特征，研究了基于子空间和基于粒子群优化稀疏分解方法的欺骗干扰识别方法。更多还原

【Abstract】 With the generation and applicationof the digital radio frequency memory (DRFM)technology, the synthetical electronic counter measure (ECM) is one of the so-called“four threats” to the radar.The deception jammer consists of the DRFM, which cangenerate multiple modes, fleasibility and stronger opposability jamming. It forms themost severe threat to military radar. Moreover, although the existing radars are equippedwith some electronic counter-countermeasures (ECCM), they cannot choose theoptimum anti-jamming measurement automatically according to the jammingenvironment, as a result of the absence of the jamming mode sensing function.In order to enhance the radar early-warning and detecting performance in thecomplicated electromagnetic environment, the generation mechanism of jamming andthe mechanism of jamming on radar are studied. On the bases, the sensing methods ofactive deception jamming are researched, which can provid the priori information forthe subsequent radar deception jamming suppression. Then the radar will choose thebest ECCM methods against the decptive jamming.First, the generation mechanism ofthe deception jamming is analyzes detailedly.Then the deception jamming integratedsensing algorithms are studied from the signal processing, information processing,waveform diversity and fingerprint characteristics of the deceptive jammer. The maincontents and results are listed as follows.(1) The research background is ascertained, and the conventional ECCM isintroduced. The domestic and overseas research status of the radar active deceptionjamming sensing methods is addressed detailedly. Firstly, the generation mechanism ofthe radar deception jamming based on DRFM is introduced, and then the generationmechanism of the pull jamming based on the process is presented, including range gatepull jamming, velocity gate pull jammingand range-velocity gate pull jamming. Finally,other type of deception jamming is studied. The deception jamming model is thefoundation forthe further jamming sensing algorithms.(2) The typical gate pull deception jamming signal model of the tracking radar isestablished in the releasing process of the jamming, multi-scale decomposition theory isused to study the sensing algorithm of deception jamming. Two types of multi-scaledecomposition methods of deception jamming sensing algorithms are proposed, including those based on wavelet decomposition algorithm and an algorithm based onempirical mode decomposition. In the first algorithm, wavelet decomposition of thenormalized radar received signal is used to get the high-frequency detail componentsand the low-frequency approximation components. Normalized power ratio is extractedas the multi-scale decomposition coefficients of the radar received signal. Base on thoseabove, the identification algorithm of the three towing interference is achieved. In thesecond algorithm, the processing method is similar to the former. The intrinsic modefunction (IMF) components are derived directly from the empirical modedecomposition. The features of the decomposed components are extracted to achieve thesensing algorithm of three types of jamming.(3) Non-negative matrix factorization (NMF) vector characteristics of productspectrum and texture characteristic parameters of two-dimensional spectrum areextracted by joint frequency domain and slow-time domain processing. Based on these,identification algorithms of pull jamming are proposed. In the first algorithm, productspectrum matrix (PSM) of the frequency domain and slow-time domain is established atthe beginning. Then, NMF theory is applied to the PSM, and features are extracted fromeach vectors of the decomposed matrx. Finally, a deception jamming sensing algorithmis proposed. In the second algorithm, the texture features of image processing areintroduced into deception jamming sensing algorithm. The gray image texturecharacteristics of two-dimensional spectrum are extracted. The sensing algorithm ofdeception jamming is presented using texture parameters.(4) Detection algorithms of the presence of deception jamming based oninformation domain processing are proposed, which are two kinds of deceptionjamming sensing algorithms based on amplitude fluctuation features perception. Thedetection algorithms are the goodness of fit (GOF) and particle filter (PF), respectively.In the first algorithm, when the deceptive jamming appears, the amplitude fluctuation ofthe received signal within the radar beam is different from echo. The existence ofdeception jamming detection is established, and two types of GOF detection algorithmsare proposed. Finally, the performance of Anderson-Darling (AD) detection andmodified Anderson-Darling (MAD) detection is compared by the simulationexperiments under different conditions. In the second algorithm, the echo and jammingwithin the same beam is unresolved in the pull jamming capture period. Consideringthat the energy of received signal will spill over to adjacent matched filter sampling points, the signal model is established. Base on those above, a likelihood ratio testmodel is developed. Finally, the detection algorithms of the presence of pull jammingare implemented, which uses particle filterdetector.(5) The sensing algorithms of deception jamming based on the waveform diversitytheory is studied, including those of random linear frequency modulate ratio signal(RLFMR) and chaotic phase modulation (CPM) signal. For linear frequency modulate(LFM) signal and the phase-coded signal, the generalized likelihood ratio test (GLRT)algorithm and Holder coefficient feature extraction method are presented, respectively.The detection and identification performance for deceptive jamming is verified bynumerical simulation.(6) Based on the second modulation of jammers in the deception jamminggeneration process, the nonlinear distortion model of jammer is established usingVolterra model. The fingerprint feature of jammer is extracted. The identificationmethod of deception jamming is studied based on subspace and particle swarmoptimization (PSO) in sparse decomposition method.更多还原