Research On Multi-scale Analysis And Its Application To Well Logging

Using a single log curve to evaluate strata leads to multi-solution and limitation. In order to overcome those aspects, multi-scale characters and the fusion of well logging data are studied systematically, based on the theory of multi-scale analysis, to extract the essential features of strata and to establish the corresponding relation between logging data and the strata, and therefore improving the reliability of log data applied to the evaluation of strata.From the theory of multi-scale analysis, the multi-scale analysis of logging curve are studied systematically on the basis of the optimum wavelet base selected by means of correlation coefficient. With the scale variation, the algorithm of multi-scale edge detection and filtering of logging curve are constructed. The multi-scale data fusion are studied for well logging multi-scale system. Multi-scale analysis in well logging and its algorithm are developed through the multi-scale analysis of a single log curve and the fusion of multiple log curves. This method leads to the theory based on which sequence stratigraphy and neutron time spectrometry are studied.For different frequency domains in multi-scale analysis, the fusion rules of logging data are discussed repectively, where high frequency coefficients choose maximal absolute values and low frequency coefficients are selectecd by virtue of average method, weighted method and edge method. Then combining the rules and multi-scale decomposition and reconstruction to study the fusion of one or more types of logging data. Finally based on the algorithm of reconstruction from modulus maxima of the wavelet transform and multi-scale edge detection, another fusion algorithm of logging data is built, where the maximal absolute values of high frequency and the weighted method of low frequency are selected. The results obtained by considering standard error and entropy to be evaluated criterion of the fusion data show the fusing curves enhance the common information of log curves and increase the information of strata.By multi-scale analysis the logging data are transformed from one-dimension depth into two-dimension depth-scale domain, which makes the collection and distribution of its inner energy clearly revealed. Time-frequency chromatogram and energy information displayed by wavelet coefficients on different scales is used to determine the optimum scale corresponding to boundaries of stratigraphic sequence. When Morlet wavelet, Gauss wavelet and orthogonal wavelets are applied to a single log curve and 2-spline wavelet is used to the curve fused by multiple log curves, the corresponding relation between wavelet coeffients curves and the stratigraphic sequence is founded by detecting periodic components of logging data in multi-scale and observing the oscillating characteristic of wavelet coefficients, and therefore realizing the clarification of various rank stratigraphic sequence automatically. High and low frequency responses of wavelet coefficients revealing the change in sedimentary gyre are suitable to recognize sedimentary characteristics of the inner sequence stratum. All these researches provide a completely new method for the quantitative demarcating of sequence and its inner depositional features.Based on multi-scale filtering algorithm, the chromatogram of thermal neutron counting efficiency is drawn from PNN logging instruments, which show the rule of neutron decay. Following the steps above, oil and water zone can be identified qualitatively. Time-frequency chromatogram and oscillating characteristics of wavelet coefficients are obtained from applications of Morlet wavelet to PNN logging curve and the fusion curve. Combining them to lead to clarify oil and water zone.According to multi-exponential inversion algorithm, the inversion of thermal neutron life spectrum(τspectrum) is developed. Theτspectrum is studied by simulation data to analyze the effect of formation water salinity, porosity and oil saturation on the eigen value of thermal neutron life. Simulations results show that the double-peak ofτspectrum reveal the ability of the wellbore and the formation to capture thermal neutron; and the appearance of single-peak or double-peak forτspectrum has bearing on the time chose. The research shows that the application ofτspectrum to actual logging data may identify oil and water zone. Therefore this method may ameliorate log interpretation of neutron life and improve market value.