分数阶Fourier域强弱LFM信号检测与参数估计

分数阶Fourier变换（FRFT）由于其特有的性质，非常适合处理线性调频（LFM）信号，尤其是，作为一种线性变换，可以克服多分量LFM信号之间的交叉项干扰。但是采用逐次消去法检测多分量LFM信号时，每检测一个LFM信号，都要对信号分别求旋转角 的FRFT，再进行二维搜索，计算量较大。为了提高FRFT对多分量LFM信号的检测效率，本文给出一种在分数阶Fourier域检测强、弱LFM信号的新方法。首先，分析了逐次消去法和聚类分析法检测多分量LFM信号的原理，以及它们的优缺点。提出一种聚类分析和逐次消去相结合的信号检测方法,利用平面截取信号在平面(u,α)上的尖峰，并引入基于广度优先搜索邻居（BFSN）的聚类算法，对截取的信号尖峰进行聚类分析，获得每个LFM信号对应的信号尖峰，实现多个较强信号的检测与参数估计，再利用逐次消去法实现弱信号的检测。该方法可以同时检测多个能量相近的LFM信号，提高了检测效率，以及次强信号的参数估计精度，并有效地抑制了强信号对弱信号的遮蔽影响。通过对信号进行平面切割处理，减少了BFSN聚类算法中输入集样本数量，大大降低了算法的计算量。最后，仿真验证了该方法的有效性。 

Detection and Parameter Estimation of Strong and Weak LFM Signals in the Fractional Fourier Domain

In order to improve the detection efficiency of the fractional Fourier transform for multicomponent LFM signals, this paper focuses on detection and parameter estimation of strong and weak LFM signals based on the fractional Fourier transform (FRFT). First, respectively introduce the detection and parameter estimation theories of multicomponent LFM signals based on the elimination one by one method and clustering analysis method. And analyze their advantages and disadvantages existing in the detection of strong and weak LFM signals. A novel detection method is presented combining elimination one by one method with clustering analysis method, and a clustering algorithm named broad first search neighbors (BFSN) is introduced to detect the multicomponent LFM signals. It can simultaneously detect several signals with approximative energy. It improves the detection efficiency and it also improve the parameter estimation precision of the stronger signals. The preprocessing by the flat cutting to reduce the points’ number of the input data-set is proposed for the BFSN clustering analysis method. It improves the algorithm’ computation efficiency. And then, use the elimination one by one method to eliminate the strong signals. So it can also eliminate the shading effects of strong signals on weak signals. Finally, simulations verify the effectiveness of the method.