雷达目标的微多普勒特征建模与分析技术研究

发布时间：2018-05-17 15:39

本文选题：微多普勒 + S变换　；参考：《南京理工大学》2017年硕士论文

【摘要】：随着军事科技的高速发展,各种有威胁的行进目标以及小型的无人侦察机对传统雷达的识别能力提出了严峻的挑战。基于微多普勒特征提取的雷达目标识别技术存在时频分析频率分辨率较低,无法精确识别出目标等问题。因此,如何提高微多普勒信号分析的频率分辨率是目标识别领域的一大研究热点。本文主要研究高频率分辨率的微多普勒特征提取方法。主要工作如下:(1)研究各类目标的微运动,并建立运动模型。通过运动模型得到微多普勒回波信号,为后文分析微多普勒特征提取方法奠定基础。(2)针对常规微多普勒特征提取方法由于时窗长固定、时频聚焦性等导致的时频分析频率分辨率较低这一问题,采用S变换处理,使时窗长与信号频率成反比,克服了时窗长固定这一缺陷。仿真表明该方法在一定程度上提高了时频分析的频率分辨率。(3)针对S变换中由于信号截断造成的信号能量流失以及时频聚焦性较低这一问题,提出线性正则变换联合S变换方法。将信号映射入线性正则域,结合信号自身特性对线性正则矩阵参数进行调节,使信号频率在线性正则域中够压缩或拉伸,实现等效窗长与信号频率分量完美适配。仿真表明该方法极大的提高了时频分析的频率分辨率。(4)针对实际微多普勒信号分析时信号淹没于噪声这一问题,提出多重自相关法提取信号。在将信号从噪声中分离的同时,对信号进行加强。仿真表明,该方法有效的从噪声中分离出微多普勒信号并显著增强了信号。时频结果显示从含噪信号中提取出的微多普勒特征明显,实现了在低信噪比环境中目标微多普勒特征的提取。
[Abstract]:With the rapid development of military science and technology, a variety of threatening targets and small unmanned reconnaissance aircraft pose a severe challenge to the recognition ability of traditional radar. The technology of radar target recognition based on micro-Doppler feature extraction has some problems such as low frequency resolution of time-frequency analysis and the inability to identify target accurately. Therefore, how to improve the frequency resolution of microDoppler signal analysis is a hot topic in the field of target recognition. In this paper, we mainly study the feature extraction method of micro-Doppler with high frequency resolution. The main work is as follows: 1) the micro motion of various targets is studied and the motion model is established. The micro-Doppler echo signal is obtained by the motion model, which lays the foundation for the analysis of the micro-Doppler feature extraction method in the future. (2) for the conventional micro-Doppler feature extraction method, the time window length is fixed, and the time window length is fixed. Because of the low frequency resolution of time-frequency analysis caused by time-frequency focusing, the S-transform is used to make the time-window length inversely proportional to the signal frequency, which overcomes the limitation of fixing the time-frequency window length. Simulation results show that the method improves the frequency resolution of time-frequency analysis to a certain extent. A joint S-transformation method for linear canonical transformation is proposed. The signal is mapped into the linear canonical domain, and the parameters of the linear canonical matrix are adjusted according to the characteristics of the signal itself, so that the frequency of the signal is compressed or stretched in the linear canonical domain, and the equivalent window length is perfectly adapted to the frequency component of the signal. Simulation results show that this method greatly improves the frequency resolution of time-frequency analysis. When the signal is separated from the noise, the signal is strengthened. Simulation results show that the method can effectively separate the microDoppler signal from the noise and significantly enhance the signal. The time-frequency results show that the micro-Doppler features extracted from noisy signals are obvious, and the target micro-Doppler features can be extracted in low signal-to-noise ratio (SNR) environment.
【学位授予单位】：南京理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN957.51

【相似文献】