毫米波线性调频系统普遍采用“差频–快速傅里叶变换(FFT)”进行信号处理，但是当目标与雷达之间存在相对运动时，差频信号包含了目标的距离和速度信息。利用传统的FFT 变换进行速度距离解耦合操作，效果并不理想。针对此问题，介绍了分数阶FFT 和基于这一变换的调频信号检测原理，提出一种改进的分数阶傅里叶域局部频谱细化方法。仿真结果表明，该频谱细化算法在相同的FFT 变换点数下，能够获得更高的频谱分辨率，而频谱分辨率相同时，该算法的计算量更少。

"Beat-FFT" is usually used in the Millimeter Wave(MMW) Linear Frequency Modulated Continuous Wave(LFMCW) system. The beat signal contains velocity and range information when there is relative motion between target and radar. The effect of decoupling by FFT transform is not ideal. In order to solve this problem, Fractional Fourier Transform(FRFT) is introduced as well as the principle of LFM signal detection based on FRFT. An improved local FRFT spectrum refinement method is put forward. Simulation results show that the frequency spectrum refinement algorithm can obtain higher spectral resolution under the same points of FFT transform; and it would take less calculation to get the same spectral resolution.