Research Of Seismic Blind De-convolution Based On Maximal Entropy Theory

With the depth and development of oil and gas exploration, in order to be able to interpret formation, construction, reservoirs and other geological objectives much more refined, we need the higher resolution of seismic data. It is the key of seismic data processing to improve the resolution of seismic data. Traditional approach is to assume that the reflection coefficient are white noise sequences, using seismic records of autocorrelation, while the use of linear filtering (such as Wiener filtering, predictive de-convolution) or the best seismic de-convolution to estimate the wavelet character, while autocorrelation of seismic records has no wavelet phase information, so it is often assumed that the seismic wavelet is the minimum phase wavelet.Seismic deconvolution is basically a blind process. Usually seismic wavelet stimulated by the explosion of surface is unknown, and in the records seismic waves on adjacent seismic reflection are overlapping which can not be separated from the seismic records. Seismic wavelet and the reflection coefficient is unknown, therefore, it is very meaningful to do seismic deconvolution without any assumptions.This article references the work of previous studies, at first, we introduce the basic principles and methods of blind deconvolution, and classification. We focus on the two important methods of blind deconvolution - constrained sparse blind deconvolution based on the reflection coefficient and blind deconvolution Infomax method. By model trial caculation and actual seismic data processing, it shows that L1 model smallest bound sparse blind deconvolution can effectively improve the resolution of seismic data and have better fidelity. Infomax algorithm is the classical algorithm of blind source separation, this paper put forward in-depth study by K. Tokkola feedback network structure, Infomax algorithm will be extended to a wider range of circumstances, that is, with time delay of the source or aliasing aliasing signal deconvolution Blind Source Separation. Theoretical model and actual data show that the treatment method can adapt to non-Gaussian system, also get the reflection coefficient sequence of optimal estimation.In the general, buried hill is highly reflectivity for reflection on seismic profiles. However, with the cap rock lithology of the different hill, as well as the depth changes, continuity and amplitude of the top reflector is changed. In this paper, we combine the subject of buried hill identification to deal with the actual data, the actual results show that blind deconvolution can effectively identify the unconformity at the top of hill, also it has greater assistant to the hill identification and description of contours. At the same time it enhances the reflection information inside hill, provides more favorable conditions to prediction of buried hill reservoir.