基于时频分布的多分量信号提取与重建技术研究

发布时间：2018-05-05 07:10

本文选题：时频分析 + 信号分离　；参考：《哈尔滨工业大学》2017年硕士论文

【摘要】：随着战场电磁环境的日趋复杂,战场中截获到的雷达信号也日益繁杂,不仅调制种类繁多,而且叠加进入接收机的分量个数也在加剧。在复杂的截获信号中得到各分量的类型和参数,进而对各辐射源进行正确识别和高效干扰是制定战略决策的重要因素。要正确的分析截获的多分量雷达信号,将其包含的分量提取和重建是一个不可避免的过程,本文重点基于信号的时频分布特征对多分量信号的提取和重建技术进行探讨。首先,总结了常见雷达信号类型的时频特征和稀疏特征,并进一步研究了多分量雷达信号在时频分布中存在的规律。研究发现计算时频分布时不可避免的需要在干扰项抑制和信号项模糊之间均衡选择,因此引入自适应方向核的二次时频分布。该时频分布根据信号项和干扰项在模糊域的特征,通过自适应选择模糊域方向核,在干扰项抑制和信号项模糊之间达到较为理想的均衡。然后,为了获取信号的分量结构研究了信号的瞬时频率估计算法,研究发现已有算法不适用于存在交叉分量的信号,因此引入了梯度旋转方法来增强时频分布图像,并提出了基于端点梯度的片段连接和拟合算法,不仅消除分量瞬时频率跟踪错误,也降低了估计误差。最后,在瞬时频率估计的基础上采用时变滤波方法提取和重建各个分量信号,分析发现时变滤波在分量交叉处存在较大的畸变,因此引入幅度校正算法并提出基于时变阶的短时分数阶傅立叶变换时变滤波算法。所提算法大幅度的提升了信号提取和重建的准确性,特别是针对非线性调频信号的提取和重建。论文针对多分量信号时频分析中存在交叉项干扰与信号项模糊相矛盾问题,引入了自适应方向核的时频分布算法获得了较理想的时频分布图像。进一步在该图像上使用旋转梯度增强和瞬时频率连接拟合算法,取出了各分量的瞬时频率。最后采用基于时变阶的短时分数阶傅立叶变换的时变滤波算法,提取并重建出各个分量的时域波形。将信号的提取和重建分解为信号建模、时频分析、瞬时频率提取和时变滤波四个步骤进行,形成了一套完整有效的多分量雷达信号提取和重建的方案。
[Abstract]:With the increasing complexity of the electromagnetic environment of the battlefield, the radar signals intercepted in the battlefield are becoming more and more complicated. Not only there are many kinds of modulation, but also the number of components superimposed into the receiver is increasing. It is an important factor to make strategic decision to obtain the types and parameters of each component in the complex intercepted signal, and then identify the emitter correctly and interfere with each other efficiently. In order to correctly analyze the captured multi-component radar signal, it is an inevitable process to extract and reconstruct the components contained in it. This paper focuses on the multi-component signal extraction and reconstruction technology based on the time-frequency distribution characteristics of the signal. Firstly, the time-frequency characteristics and sparse features of common radar signal types are summarized, and the existence of multi-component radar signals in time-frequency distribution is further studied. It is found that when computing time-frequency distribution, it is necessary to choose between interference suppression and fuzzy signal term, so the quadratic time-frequency distribution of adaptive direction kernel is introduced. According to the feature of signal and interference term in fuzzy domain, the time-frequency distribution adaptively selects the direction kernel of fuzzy domain, and achieves a more ideal balance between interference term suppression and signal term ambiguity. Then, in order to obtain the component structure of the signal, the instantaneous frequency estimation algorithm is studied. It is found that the existing algorithm is not suitable for the signal with cross components, so a gradient rotation method is introduced to enhance the time-frequency distribution image. A segment connection and fitting algorithm based on endpoint gradient is proposed, which not only eliminates the instantaneous frequency tracking error, but also reduces the estimation error. Finally, based on the instantaneous frequency estimation, the time-varying filtering method is used to extract and reconstruct each component signal, and it is found that the time-varying filter has a large distortion at the intersection of the components. Therefore, an amplitude correction algorithm is introduced and a short-time fractional Fourier transform time-varying filtering algorithm based on time-varying order is proposed. The proposed algorithm greatly improves the accuracy of signal extraction and reconstruction, especially for nonlinear FM signal extraction and reconstruction. In order to solve the problem that there is a contradiction between the crossover interference and the ambiguity of the signal in multi-component signal time-frequency analysis, an adaptive direction-kernel time-frequency distribution algorithm is introduced to obtain an ideal time-frequency distribution image. Furthermore, the instantaneous frequency of each component is extracted by using rotation gradient enhancement and instantaneous frequency connection fitting algorithm on the image. Finally, the time-varying filtering algorithm based on short-time fractional Fourier transform is used to extract and reconstruct the time-domain waveforms of each component. The signal extraction and reconstruction are decomposed into four steps: signal modeling, time-frequency analysis, instantaneous frequency extraction and time-varying filtering, which form a complete and effective scheme for multi-component radar signal extraction and reconstruction.
【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN957.51

【参考文献】