透過您的圖書館登入
IP:18.224.44.108
  • 期刊

有關波浪分析中的FFT本質相位研究

A Study on Essential FFT Phases in Water Wave Analysis

摘要


本文探討利用快速傅立葉轉換(FFT)進行頻譜分析所可能發生的相位流失問題,在進行一系列數值模擬實驗後,找到週期性連續波列的本質相位,讓頻率域和時間域訊號分析有更密切的連結。實驗中藉由指定主頻、單位振幅及初始相位產生單一餘弦函數波,進行FFT分析並比較了主頻波的初始相位及分析相位。發現FFT分析在運算中會自動賦予每個成分波一個介於0到π之間與頻譜分頻呈線性變化關係的本質相位,而真實相位可由分析相位扣除對應的本質相位而得。如用真實相位、頻譜分頻及頻譜振幅可在時間域以餘弦函數線性疊加再生波形,不需在頻率域進行傅立葉逆轉換(iFFT),顯示一般所謂經過頻譜分析可能造成相位流失的問題並不存在。研判應是FFT的演算法及相關的複雜因次分解計算所致。部分成果與討論已呈現於林(2017)。為了作進一步證實,本研究也擴大進行了多種成分波組合測試,並比較原始波形與再生波形的差異性,包含線性波及非線性波、週期性/非週期性波動等,發現除了孤立波以外,週期性的線性或非線性波形部分,如果主頻等於頻譜分頻之一則有明確的本質相位關係;否則其本質相位關係尚待探討,但仍可以充分地再生波形。藉由本質相位的修正,使得波動可以在時間域及頻率域間作密切的結合與應用。包括各個成分波在時間或空間的掌握,數位濾波的發展,或者主頻波的研究與其他應用等,仍待深入研究。

並列摘要


In this study, an essential phase shift of the fast Fourier transform (FFT) was found through a series of numerical experiments on designed wave signals. The tested signal was produced from single sinusoidal waveforms with various principal frequency, amplitude and initial phase. After extracting the assigned initial and analyzed phases from FFT, from the relation of analyzed phases vs. principal frequencies with different initial phases, it was found that FFT automatically adds-in an essential phase shift which linearly distributes from 0 to π along the whole spectral component frequencies. Although FFT has already been treated to have phase lost problem that the signal cannot be recovered in time domain with spectral frequencies, amplitudes and analyzed phases, except the inverse FFT was employed. After subtracting the essential phase shifts from the analyzed phases of each components to gain the true phases. The signal can be fully recovered in time domain from spectral component frequencies, amplitudes and the true phases with only acceptable differences. Some of the results had been discussed by Lin (2017).

延伸閱讀