| Spectral decomposition technology, which is a time-frequency analysis method,is widely used in reservoir interpretation in recent years. With the difficulty ofdiscovery and exploitation is growing, it is necessary to innovation a variety ofexisting methods including the traditional time-frequency spectral analysis methods,which can improve the accuracy of seismic data reservoir prediction. In this paper,based on the conventional time frequency analysis method, some new spectrumdecomposition technique methods are studied, which is used in a3D seismic data ofTriassic sand bodies in an area. The main contents and innovation points are asfollows.⑴. In this paper, the Fusion algorithm, which combines to EMD and ICAalgorithm in blind signal processing, is proposed. Compared with EMD algorithm, theJ-EMD algorithm has some advantages in the certain extent, which overcomes thedefects of EMD algorithm in the application. The application effects of EMDalgorithm and J-EMD algorithm are compared in the actual3D seismic data. EMDalgorithm decomposes a single into IMF components that can not effectively identifysignal spectrum of oil and gas and its inadequacy is confirmed. However, the J-EMDalgorithm is able to obtain the independent IMF component that can improve thereliability for oil and gas reservoir identification.⑵. This paper puts forward a method for spectral decomposition of seismic data,which combines the generalized S transformation and the JADE algorithm that is arobust ICA algorithm. The time-frequency spectrum decomposition of Seismic datacan obtain a lot of single frequency data that can give rise to the great inconveniencefor the seismic data analysis and interpretation work. The method can extract the oiland gas sensitive independent spectrum form the different single frequency data in thetime-frequency domain of seismic signals. Some of independent spectra are thecommon frequency information of some single-frequency data, which can reflectsome geological bodies’ information, respectively. So this method can effectivelyidentify and predict reservoir in the actual3D seismic data, and reduce ambiguity.⑶. In this paper, we define a new fractional time-frequency method, thefractional Fourier S transform (FRST), based on the idea of the S transform(ST) and the FRFT, and study its inverse transform and others properties. The fractional Fouriertransform (FRFT) can process the non-stationary signals, but using a global kernel,the spectra of FRFT don’t show the time information. However, FRST has theadvantages of FRFT and ST, which can process the seismic signals and othertime-varying, non-stationary signals. Then the flexibility of time-frequency signalanalysis is increased, and then it has broad application prospect.⑷. This paper study the application of FRST in seismic data reservoiridentification form three aspects as follows. First, the estimate of optimal fractionalparameter is studied in3D seismic data. Compared to ordinary ST, the method has agood effect in the FRST single-frequency slice of the3D seismic reservoir. Second,the optimal fractional parameter estimation method has some defects to apply in the3D seismic data. The combinative method of FRST and the complex cFastICAalgorithm is put forward, which can mine more effective spectra of geologicalinformation by using different fractional parameters, can extract the effectiveindependent spectrum of identifying geological characteristics information andimprove the efficiency of the interpretation of seismic data, while don’t need toestimate the optimal fractional parameter. Finally, the FRST is used to detect thelow-frequency shadows below oil and gas reservoir. At the same time, using theadvantages of the FRST fractional rotation parameters, this paper puts forward themethod that increases the reliability of reservoir identification and improve theefficiency of identification to exact the low-frequency shadows from multiplefractional parameters, low frequency single frequency data. |
PDF Full Text Request