基于分数阶傅里叶变换的线性调频信号估计与分离研究

发布时间：2018-04-14 23:07

本文选题：线性调频信号 + 分数阶傅里叶变换　；参考：《南京理工大学》2017年硕士论文

【摘要】：线性调频信号(LFM)是一种雷达和通信等信息系统常采用的信号波形。一方面,由于现代电子对抗中电磁环境更加复杂多变,使对LFM信号的分选与识别变得十分棘手。对LFM信号各项参数的检测,特别是在信噪比并不理想的条件下,完成对LFM信号的检测和参数估计,具有相当重要意义。另一方面,接收机会在同一时刻接收到不止一个信号,因此要从混叠信号中检测出各个信号的特征参数并将它们分离,成为电子信息对抗领域中一个亟待研究的重要课题。本文主要研究的是基于分数阶傅里叶变换(Fractional Fourier Transform,FrFT)的时频分析方法,包括对信号参数估计和分离的研究。在对LFM信号进行参数估计时,传统的时频分析方法如短时傅里叶变换、Wigner-Ville分布等方法中存在的分辨率低、交叉项干扰等问题,相比之下,FrFT就不存在这样的弊端。不过FrFT中涉及二维峰值搜索,搜索中涉及很大的计算量,影响对信号检测的实时性。本文针对上述问题,首先提出了对应的改进算法:基于FrFT插值法的改进算法和基于Nuttall窗的能量重心法。其中插值法的改进算法中利用对频谱进行插值处理的方法,在保证参数估计精度的条件下,加大搜索步长r,减小二维搜索的运算量;而基于Nuttall窗的能量重心法则是利用Nuttall窗的优良性能提高能量重心法估计调频率时的精度,利用调频率和FrFT阶数的关系,在小范围内进行搜索得到精确的调频率和起始频率。最后,对改进算法进行了 MATLAB软件仿真,验证了其性能。基于CLEAN思想的多分量信号分离方法,采用遮蔽已知信号的方法能够成功的将多分量LFM信号分离并且同时逐一完成参数估计。但这种分离方法无法应用于FrFT改进算法中的降维处理,使用受到了限制。本文介绍了一种改进的CLEAN分离方法,使其分离算法契合FrFT改进算法中的降维思想,同时进行了软件仿真。最后利用硬件系统对算法进行了实验测试,检验了算法在实验条件下的性能,实验结果与MATLAB仿真结果一致,证明了该方法的工程实用价值。
[Abstract]:Linear Frequency Modulation signal (LFM) is a kind of signal waveform commonly used in radar and communication information systems.On the one hand, the electromagnetic environment is more complex and changeable in modern electronic countermeasures, which makes the sorting and recognition of LFM signals very difficult.It is of great significance to detect and estimate the parameters of LFM signal, especially under the condition that SNR is not ideal.On the other hand, the receiver will receive more than one signal at the same time, so it is necessary to detect and separate the characteristic parameters of each signal from the aliasing signal, which is an important subject to be studied in the field of electronic information countermeasure.This paper mainly studies the time-frequency analysis method based on Fractional Fourier transform (FrFT), including the estimation and separation of signal parameters.In parameter estimation of LFM signal, the traditional time-frequency analysis methods such as short time Fourier transform (STFT) Wigner-Ville distribution have some problems such as low resolution, cross term interference and so on.However, 2D peak search is involved in FrFT, which involves a lot of computation, which affects the real time of signal detection.In order to solve the above problems, this paper first proposes the corresponding improved algorithm: the improved algorithm based on FrFT interpolation and the energy barycenter method based on Nuttall window.In the improved interpolation algorithm, the interpolation method is used to interpolate the spectrum. Under the condition of guaranteeing the precision of parameter estimation, the search step size is increased, and the operation amount of two-dimensional search is reduced.The energy barycenter rule based on Nuttall window is to improve the accuracy of the energy barycenter method when estimating the frequency by using the excellent performance of the Nuttall window. By using the relation between the modulation frequency and the FrFT order, the accurate tuning frequency and the starting frequency can be obtained by searching in a small range.Finally, the improved algorithm is simulated by MATLAB software and its performance is verified.The multi-component signal separation method based on the idea of CLEAN can successfully separate the multi-component LFM signal and complete the parameter estimation one by one by using the method of masking the known signal.However, this separation method can not be applied to the dimensionality reduction in the improved FrFT algorithm, and its use is limited.In this paper, an improved CLEAN separation method is introduced to fit the dimensionality reduction idea of the improved FrFT algorithm, and the software simulation is carried out at the same time.Finally, the algorithm is tested by hardware system, and the performance of the algorithm under the experimental condition is tested. The experimental results are in agreement with the results of MATLAB simulation, and the practical value of the method is proved.
【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】：TN911.23

【参考文献】