本文采用傅里叶变换(简称傅氏变换)的近似算法(WFFT)[1]做通过采样获得的时间序列的功率谱密度估计,形成了一种新的经典谱估计方法。文章叙述了这种谱估计方法的原理和特点,并通过举例计算,把新的谱估计方法与经典谱估计中较好的Welch法进行了比较。计算表明,在一般情况下,新的谱估计方法与Welch法相比有如下优点:(1)可以对数据作整段处理,即无需分段计算;(2)不需加数据窗,即只加矩形窗;(3)只用较少的数据便可得到较好的估计质量;(4)具有计算简单,无混迭误差,分辨率高,泄漏误差小,估计质量较好等特 In this paper, we use Fourier transform (FFT) approximation algorithm (WFTFT) [1] to do the time-series power spectral density estimation obtained by sampling and form a new classical spectral estimation method. The article describes the principle and characteristics of this spectral estimation method, and compares the new spectral estimation method with the better Welch method in classical spectral estimation by example calculation. The calculation shows that, in general, the new spectral estimation method has the following advantages compared with the Welch method: (1) The data can be processed in the whole process, that is, no segmentation calculation is needed; (2) Rectangular window; (3) better estimation quality can be obtained with less data; (4) with simple calculation, no aliasing error, high resolution, small leakage error and good estimation quality