全相位FFT密集谱识别与校正

Discrimination and Correction for Dense All-phase FFT Spectrums

为估计密集谱成分信号的频率和相位值,本文提出基于全相位FFT(all-phase FFT,apFFT)的密集谱识别与校正算法.利用全相位FFT良好的抑制谱泄漏性质和其特有的"相位不变性",本文方法可在传统振幅谱峰识别法失效情况下,根据相位谱是否满足平坦分布来识别密集谱与单频成分谱;结合全相位时移相位差校正法与相邻apFFT谱线间的相位关系,推导出了一种新的密集谱频率与相位的估计方法,并给出了辨别两种谱分布的参数指标.仿真实验表明,在不同信噪比情况下,本算法精度比原有FFT密集谱校正法有所提高,当SNR>20dB时,本文方法有较高的实用价值.

To estimate the frequency and phase information of the signals containing dense spectral components,a novel discrimination and correction method for dense allphase FFT（apFFT）spectrums is proposed.Due to apFFT＇s excellent performance in suppressing spectral leakage and the property of ＇phase invariant＇,this paper can distinguish the dense spectrum from single-tone spectrum by judging whether the phase spectral lines behave as flat distribution,when the conventional discrimination method by identifying peak amplitude lines malfunctions.Combining with all-phase time-shift correcting spectrum method and the phase relation between adjacent spectral lines,this paper deduces a novel method to estimate frequencies and phases for dense spectrums and gives the discrimination index for these two cases.Simulation shows that,for the different cases of SNR（Signal to Noise Ratio）,the accuracy of the proposed method is higher than that of the conventional correcting dense spectrum method,guaranteeing that the proposed method is of high applied value when SNR〉20dB.