论文部分内容阅读
线性调频(LFM)信号在某个阶次的FRFT域中具有能量聚集性,根据这一特性采用分数阶傅里叶变换(FRFT,Fractional Fourier Transform)来分离多个未知任何先验参数的LFM信号,通过在FRFT域搜索峰值点来检测并分离出LFM信号,并用相关系数对分离效果进行了评价。针对二维搜索峰值点的计算复杂性,提出了曲线拟合优化技术来进行FRFT模值检测,把二维搜索转化为一维曲线拟合问题,并介绍了高斯混合模型近似FRFT模值分布。计算机仿真表明,这一方法极大地降低了计算复杂度,对多分量LFM信号进行了有效的分离及参数估计。
Linear frequency modulation (LFM) signals have energy aggregation in a certain order FRFT domain. Fractional Fourier Transform (FRFT) is used to separate LFM signals of unknown parameters The LFM signal was detected and separated by searching the peak points in the FRFT domain, and the correlation coefficient was used to evaluate the separation efficiency. Aiming at the computational complexity of two-dimensional search peak points, a curve fitting optimization technique is proposed to detect the FRFT modal values. The two-dimensional search is transformed into a one-dimensional curve fitting problem. The approximate FRFT modulus distribution of the Gaussian mixture model is introduced. Computer simulation shows that this method greatly reduces the computational complexity and effectively separates and estimates multi-component LFM signals.