An Innovative Approach to Time-Frequency Analysis and Its Application in Seismic Signal Processing

Abstract

Summary Traditional time-frequency analysis, such as short-term Fourier transform frequency domain, has a lower resolution, and the center frequency of S transform shifts to a higher frequency direction. Because the wavelet transform is limited by the uncertainty principle, it is difficult to achieve the optimal time resolution and frequency resolution at the same time. In this study, a time-frequency analysis method that supports time-frequency super-resolution – adaptive Superlet transform is proposed by combining short wavelets with high temporal resolution and long wavelets with high frequency resolution. First, a series of wavelet atomic families are defined, and then wavelet atoms are combined by minimum mean cross-entropy technology, and finally the seismic signal and the combined wavelet atoms are convolved to obtain a new time spectrum. The method was applied to the detection of fluid “low-frequency shadow” in the oilfield of southern China, and the detection results of 6 oil and gas wells were good. The time-frequency center of the ultra-wavelet transformation method is closer to the main frequency of the signal, and the low-frequency band has a high-resolution time-frequency structure, and the numerical calculation is more stable.