【作者】 谭小刚；

【导师】 魏平；

【作者基本信息】 电子科技大学 ， 信号与信息处理， 2008， 博士

【摘要】 复杂雷达信号的检测和分类一直是电子战的关键技术。随着雷达发射机和波形种类的增加，采用到达方位、载频、到达时间、脉宽及脉幅的传统方法已经不能满足电子战信号处理的要求。为了有效地识别雷达信号，必须对雷达信号脉内细微特征进行分析和提取。本论文研究了雷达频率调制信号的脉内细微特征提取方法，对低信噪比下多分量频率调制信号的快速高精度参数估计和检测进行了深入的研究。论文内容包括：1．介绍了电子战的概念和组成结构，以及现代战争中的雷达信号脉内细微特征分析技术研究现状，总结了非平稳信号分析的内容、发展过程和现状。从信号表示的角度回顾了时频分析的起源，介绍了它的发展趋势。2．提出了平滑伪多项式Wigner-Ville分布(SP-PWVD)以减少低阶多项式Wigner-Ville分布(PWVD)表示高阶多项式相位信号(PPS)信号时出现的交叉项，总结了SP-PWVD的有关性质，利用最优化方法来设计高阶PWVD，采用快速傅立叶变换(FFT)插值来实现PWVD的高阶瞬时矩中非均匀采样信号相乘时的尺度变换。3．深入分析了交叉项和自适应扩散局部扩散行为的特点，利用差分法来去除自适应扩散无法去除的交叉项，采用Hesse矩阵代替梯度平方张量来获取局部扩散方向以提高时频分布可读性，给出了扩散算法的一种数字实现方案。4．采用FFT插值和二次插值来提高乘积高阶模糊函数(PHAF)估计精度，将逐次滤波和RELAX结合起来改进PHAF，以实现低信噪比(SNR)下多分量高阶PPS的高精度参数估计。讨论了用PHAF实现多分量高阶PPS检测时遇到的困难，对基于PHAF模的检测算法作了分析，提出了基于逐次滤波和PHAF模最大值的PPS检测方法，可以在较低SNR下实现多分量高阶PPS的检测。研究了PPS阶数和延时组的选择。5．分析了在低SNR下基于Radon-模糊变换(RAT)模的检测器存在的问题，利用二阶差分法来改进RAT以提高线性调频信号(LFM)的检测概率，采用二次搜索方法改进RAT来实现LFM的快速高精度参数估计，采用FFT插值和二次插值来提高解调法的估计精度，对Radon Wigner变换(RWT)、解线调、PHAF和乘积三次相位函数进行了比较，并重点分析了解线调方法和PHAF方法的运行时间、信噪比门限和样本个数之间的关系。6．对全文进行了总结，指出了频率调制信号参数估计中有待进一步研究的问题。更多还原

【Abstract】 The detection and classification of complex radar signal has been a keytechnology for electronic warfare.The traditional method employing direction ofarrival,carrier frequency,time arrival,pulse width and amplitude can not meet therequirements of application in signal processing of electronic warfare withever-increasing number of radar emitters and different radar waveforms.It isnecessary to analyze and extract the in-pulse fine features of radar signals forefficient recognition of radar signals.In this thesis,the extraction of in-pulse finefeature of frequency modulated signals in radar is investigated,and the fast andaccurate parameter estimation and detection of multicomponent frequencymodulated signals are studied extensively.The main results are as follows:1.The conception and construction of electronic warfare are introduced.Thepresent research of in-pulse fine feature analysis of radar signals in modern warfareis presented.And the content,evolution and present status of nonstationary signalanalysis are concluded.The origin of time frequency analysis in the point of signalrepresentation is reviewed.And the development trend of time frequency analysis isintroduced.2.Smoothed pseudo polynomial Wigner-Ville distribution (SP-PWVD) isproposed to reduce interference terms which present when higher order polynomialphase signal (PPS) is represented by lower order polynomial Wigner-Villedistribution (PWVD).The properties of SP-PWVD are concluded.High orderPWVD is designed by optimal method.The scale transformation of the non-uniformsampling of the signals in the high order instantaneous moment of PWVD isperformed by fast Fourier transform (FFT) interpolation.3.After making an in-depth analysis of the interference terms and localdiffusion behavior of the adaptive diffusion process,we combine the adaptivediffusion with difference method to eliminate the stubborn interference terms.Andwe substitute the Hesse matrix for the gradient square tensor to obtain the local diffusion direction,and thus improve the readability of the adaptive diffusiondistribution.A scheme of digital realization is provided.4.The FFT and quadratic interpolation are used to improve the precision ofproduct high order ambiguity function (PHAF).A method based on iterativefiltering and RELAX is proposed to provide accurate parameter estimation for highorder muticomponent PPS at low signal to noise ratio (SNR).Difficulties indetection of higher order and multi-component PPS by PHAF are discussed.Adetector based on iterative filtering and maximum magnitude of PHAF is proposedafter an analysis of detector based on PHAF magnitude.And the selection of orderand delay sets is studied.5.A detector based on second order difference of Radon ambiguity transform(RAT) is proposed to improve detection probability of linear frequency modulatedsignal (LFM) after analysis of problems with detector based on magnitude of RATin low SNR.A secondary search strategy of RAT is proposed for fast and accurateparameter estimation of LFM.Dechirp based on FFT and quadratic interpolation isproposed to analyze LFM at low SNR.Comparisons of Radon Wigner transform,dechirp,PHAF and product cubic phase function are performed and the emphasis isplace on time collapsed,signal to noise ratio threshold and number of samplesbetween dechirp and PHAF.6.At last,the whole dissertation is concluded.And some further researchfrontiers of frequency modulated signal detection and parameters estimation arepointed out.更多还原