Research Of Reservoir Prediction Method Based On Synchrosqueezing Wavelet Transform

In the process of seismic data processing and interpretation,time-frequency analysis technology has always been the focus of scholars.It is a method for characterizing the relationship between time and instantaneous frequency of signal,and it can extract many hidden information from seismic signals.In practical application,the common time-frequency analysis methods have Short-Time Fourier Transform,Continuous Wavelet Transform,Three Parameter Wavelet Transform,S-transform,Hilbert-Huang Transform,Wigner-Ville Distribution and so on,although these methods could get good results in real data,but they would have a series of shortcomings,such as the low time-frequency resolution,irreversibilityrely,relying on wavelet function and so on.Also with the increasingly complex of seismic data and the accuracy of the increasingly high demand in data processing and interpretation,the conventional time-frequency methods have been unable to satisfy the requirements of seisimic data processing and interpretation,so there is a high time-frequency resolution,high energy focusing time-frequency and strong robustness signal analysis method-synchrosqueezing wavelet transform-CSWT.Starting from the mathematical principle of synchrosqueezing wavelet transform,the mathematical derivation and common mathematical properties of the algorithm are introduced in this thesis.Then through the theoretical model to analyze the wavelet function types and its parameters,the results show that the algorithm is not sensitive to wavelet function types,is more sensitive to the center frequency and the time-frequency resolution is far higher than that of conventional time-frequency analysis method;it has good characteristic of robustness,reversibility and anti noise.These advantages are that the traditional time-frequency analysis methods do not have.Therefore,known from the analysis,we can see that the selection of appropriate parameters can make the CSWT algorithm get better effects than conventional time-frequency methods when dealing with seismic data.Secondly,based on the shortcomings of synchrosqueezing wavelet transform denoising algorithm,this paper proposes an improved denoising algorithm combined with CEEMD and synchrosqueezing wavelet transform,the analog data and actual data analysis shows that the denoising algorithm can remove the random noise effectively,and can preserve the amplitude of effective signals.Finally,this paper introduces the principle and analyze the feasibility of detecting oil and gas in the thin layers by using spectral decomposition method and absorption attenuation gradient method base on CSWT.Then,we use the actual original seismic data to verify it.First,removing the noise from the original seismic profile by using the CEEMD-CSWT denoising algorithm..Next,using CSWT to decompose the spectrum to and using the "low frequency anomaly" phenomenon of instantaneous spectrum to detect the oil and gas.And comparing the result of CSWT with that of CWT and ST,showing the effectiveness and superiority of the CSWT algorithm.On the other hand,the left and right slope method is used to extract the low and high frequency absorption attenuation gradient of the actual seismic section,and it is compared with the attribute map extracted by CWT,and the validity and superiority of the method in reservoir prediction is verified again.