Abstract:When a signal is in minimum phase, its amplitude spectrum and phase spectrum will satisfy Hilbert transform condition.Having known the amplitude spectrum of a non-minimum phase signal,we can conduct Hilbert transform to extract a minimum phase signal of which the amplitude spectrum is identical with that of the non-minimum phase signal.This paper recommends a method far directly converting logarithmic amplitude spectrum into minimum phase spectrum by Hilbert transform in time domain.Parzen's window function is used to reduce the truncation effect of Hilbert factor and speed up convergence because parzen window frequency spectrum has neither side lobes nor"missing" prablem,This method is characterized by combined operation in frequency domain and time domain.Taking RDS-500 computer as an instance, the combined operation time is half of the time taken when the opera-Lion is only conducted in frequency domain.Knocks far making minimum phase spectrum in different cases are illustrated in this paper, The only difference between the amplitude spectrum of seismic record and that of source wavelet is a constant coefficient when the reflection coefficient series is white noise series.In this case, the amplitude the amplitude spectrum of seismic record can be used as substitute for the amplitude spectrum of source wavelet, then the minimum phase spectrum of source wavelet can be obtained using the method recommended in the paper.A mixed phase signal input also can be transformed into minimum phase signal in this way.
收稿日期: 1985-07-29
引用本文:
王卫华. 用时间域希尔伯特变换求取信号的最小相位谱[J]. 石油地球物理勘探, 1986, 21(3): 268-275.
Wang Weihua. Extracting the minimum phase spectrum of signal by Hilbert transform in time domain. OGP, 1986, 21(3): 268-275.