【作者】 徐春光；

【导师】 谢维信；

【作者基本信息】 西安电子科技大学 ， 信号与信息处理， 1999， 博士

【摘要】 时频分析是分析处理非平稳信号的有力工具，它将信号表示为时间和频率的联合函数，揭示了信号的时变谱特征。本文围绕时频分析的有关理论和应用进行了研究，主要研究工作及取得的研究成果包括： 1.分析了短时傅立叶变换的时频分辨性能和时间-时延-多普勒函数的性 质，提出了一种采用线性调频窗函数的自适应短时傅立时变换方法，利 用时间-时延-多普勒函数和Radon变换技术估计信号的瞬时调频指数， 自适应地调整窗函数的参数。 2.分析对比了时频分布中交叉项的抑制方法，提出了一种最优相位核时频 分布；讨论了最优相位核时频分布的性质、自主项区域的确定和附加其 它时频分布性质的最优相位核时频分布；给出了离散化最优相位核时频 分布的数值计算方法；利用最优相位核时频分布对蝙蝠信号进行分析， 获得高度的时频分辨率与精确的时频结构。 3.将局部Radon变换引入时频分布领域，提出了局部Radon-Wigner变换， 给出了其定义表达式和一种快速算法，分析了其性质以及与Radon- Wigner变换、Wigner分布、模糊函数的关系，讨论了其与广义边缘条 件时频分布的内在联系。 4.分析了局部化时频分布的性质，提出了互局部化时频分布，并提出一种 基于互局部化时频分布的迭代瞬时频率估计方法；以解析方法分析了无 噪声情况下该方法的收敛性，讨论了初始参考信号的估计方法，用计算 机仿真方法分析了在噪声背景中与其它方法的性能对比；利用该方法分 析雷达测速系统的多普勒信号，准确地估计了被测目标的加速情况。 5.分析了编队机群低分辨雷达回波多普勒信号的时频结构特性，提出了编 队目标雷达回波信号的数学模型，并用实测雷达数据的时频分析结果验 证了该模型的正确性；提出了自适应Chirp信号表示方法；利用该方法 对实测编队目标雷达回波数据进行分析，有效地实现了多目标分辨。更多还原

【Abstract】 ABSTRACTTime frequency analysis is an effect tool to analyze and process non-stationarysignals, which maPs a one-dimensional signal into a two-dimensional function of timeand frequency and reveals the time-varying spectra. This dissertation fOcus on therelated theories and aPplications of time-frequency analysis, and the main researchworks and results are as follows:l. The time-frequency resolution of short-time Fourier transfOrm (STFT) andthe properties of time-lag-Doppler function (TLDF) are analyzed. AnadaPtive STFT with linear frequency modulation window is presented andthe instriboeous frequency modulation rate is estimated by the TLDF and -Radon technique.2. Based on the comparison of different interference restraining methods, anoptimal phase kernel time-frequency distribution is put forwarded. Theproperties of optimal phase kemel time-frequency distribUtion, thedetermination of auto-terms region, and the optimal phase kemel time-frequency distribution with additional properties are discussed. Then analgorithIn of discrete-time, discrete-frequency optimal phase kemel time-frequency distribution is proposed. Analyking bat signals by the optimalphase kemel time-frequency distribution, the high time-frequency resolutionand accurate time-frequency structure is obtained.3. By introducing the localized Radon technique into time-frequency analysisarea, the localized Radon-Wigner transfOrm (LRWT) is suggested. First, thedefinition of LRWT and a fast algorithm is consimcted. Then the propertiesof LRWT are analyzed and the relationship of LRWT with Radon-Wignertransform (RWT), Wigner distribution (WD) and Ambiguity function (AF)is discussed. Finally, the relationship of LRWT and generalized marginaltime-frequency distribution is studied.4. The properties of localized time-frequency distribution (LTFD) are analyzed,then an iterative instantaneous frequency estimation algorithIn is presented.The convergence of this algorithIn is examined analytically, the estimationof initial frequency is discussed. and the perfOrmance in noise is comparedwith other algorithIns by computer simulation. Finally, the Doppler signal ofa radar velocity analyzing system is studied by means of the method in this paper and the speed variation of the observed object is clearly described.5.Based on the analysis of the time-frequency structure of Doppler signals from multi-object by traditional Radar, a mathematical modal of a nonstationary signal is proposed, whose validity is demonstrated by real Radar echoes. In allusion to this modal, an adaptive Chirp representation (ACR) is presented, by which multi-object classification is easily realized.更多还原