Sedimentary cycle division method based on multiple logging curves

Abstract

The present invention discloses a method for dividing sedimentary cycles based on multiple logging curves. The specific division steps are as follows: Obtain logging curves, preprocess the logging curves to obtain a curve set A, extract the most characteristic curve from the curve set A, perform singular spectrum decomposition on the most characteristic curve to obtain a reconstructed curve set B, and use the signals of the reconstructed curves to divide sedimentary cycles. The sedimentary cycle division method of the present invention overcomes the problems of low efficiency and many interference factors in dividing sedimentary cycles by a single logging curve in the prior art.

Description

Technical Field

[0001] The present invention belongs to the technical field of physical exploration and relates to a method for dividing sedimentary cycles based on multiple logging curves. Background Art

[0002] In oil and gas exploration and development, sedimentary strata are the enrichment areas and research focuses of oil and gas resources. Affected by the periodicity of sedimentary tectonic movements, the petrophysical and chemical properties of sedimentary strata show regular alternating changes, indicating that the formation of strata has cyclicality. Studying the sedimentary cycles of strata helps to achieve accurate stratigraphic correlation, construct a stratigraphic framework, analyze the sedimentary characteristics of strata, and thus discover potential favorable reservoirs. Logging curves are the basic data for recording the changes in formation lithology and physical properties. Therefore, the division of sedimentary cycles of strata is particularly important.

[0003] Currently, for sedimentary cycle curves, the common methods are to divide the logging curves (such as natural gamma, spontaneous potential, resistivity, etc.) that are sensitive to sedimentary cyclicality, either directly divide the sedimentary cycles, or use transformation methods such as Fourier transform, wavelet transform, and empirical mode decomposition to transform them into the time-frequency domain to extract frequency domain features for dividing sedimentary cycles. These sedimentary cycle division methods often first target a single curve, then conduct comprehensive analysis, and finally determine the sedimentary cycles at all levels. In specific implementation, there are often problems such as huge workload, and being affected by factors such as personal experience, with many interference factors. Summary of the Invention

[0004] The purpose of the present invention is to provide a method for dividing sedimentary cycles based on multiple logging curves, which solves the problems of the sedimentary cycle division method in the prior art that requires artificial comprehensive determination based on the division results of a single curve, with a huge workload, being affected by factors such as personal experience, and having many interference factors.

[0005] The technical solution adopted by the present invention is a method for dividing sedimentary cycles based on multiple logging curves. The specific steps are as follows: obtain logging curves, perform preprocessing on the logging curves to obtain a curve set A, extract the most characteristic curve from the curve set A, perform singular spectrum decomposition on the most characteristic curve to obtain a reconstructed curve set B, and use the signals of the reconstructed curves to divide the sedimentary cycles.

[0006] The characteristics of the present invention also lie in that

[0007] The specific steps for obtaining logging curves are as follows: collect geological and logging data within the work area, determine the target interval for sedimentary cycle division, information such as formation lithology, and select M logging curves that are sensitive to sedimentary cyclicality.

[0008] The preprocessing steps of logging curves are as follows: After performing outlier correction and normalization on each logging curve respectively, the correlation coefficients between the curves are calculated using the method of linear fitting, and through mathematical transformation, a positive correlation relationship is ensured among all the curves, obtaining the basic curve set A of logging curves = {a1, a2, a3......, a M}, M≥2.

[0009] The specific steps for reconstructing the curve set are as follows: According to the contribution rates of the components of the singular spectrum, different combinations are divided, and singular spectrum reconstruction is performed to obtain a set B of N reconstructed curves = {b1, b2, b3......, b N}.

[0010] The principle for sedimentary cycle division is: medium - long - period cycles and short - period cycles.

[0011] The beneficial effects of the present invention are as follows: By fusing the characteristics of multiple logging curves, representative curves are created, the contribution rates of the components in the singular spectrum decomposition are analyzed, multiple reconstruction combinations are divided, and singular spectrum reconstruction is carried out, thereby highlighting the effective information, reducing the influence of noise, and using the different signals of the reconstructed curves to divide the sedimentary cycles into medium - long - period sedimentary cycles and short - period sedimentary cycles, reducing the workload of sedimentary cycle division and improving the accuracy of division. Brief Description of the Drawings

[0012] Figure 1 is the flow chart of the sedimentary cycle division method based on multiple logging curves of the present invention;

[0013] Figure 2 is the logging curve diagram selected according to the formation lithology and the new curve diagram extracted by dimensionality reduction in the sedimentary cycle division method based on multiple logging curves of the present invention;

[0014] Figure 3 is the contribution rate diagram of each component of the singular spectrum in the sedimentary cycle division method based on multiple logging curves of the present invention;

[0015] Figure 4 is the sedimentary cycle division diagram reconstructed based on the singular spectrum in the sedimentary cycle division method based on multiple logging curves of the present invention. Detailed Embodiment

[0016] The present invention will be described in detail below in conjunction with the drawings and specific embodiments.

[0017] A sedimentary cycle division method based on multiple logging curves of the present invention, as Figure 1 shown, the specific steps are as follows:

[0018] Step 1: Collect geological and logging data within the work area, determine the target intervals for sedimentary cycle division, information such as formation lithology, etc., and select M logging curves that are sensitive to sedimentary cyclicity.

[0019] Step 2: After performing outlier correction and normalization on each logging curve respectively, use the method of linear fitting to calculate the correlation coefficients between the curves. Through mathematical transformation, ensure that all curves maintain a positive correlation relationship, and obtain the basic curve set A = {a1, a2, a3......, a M}, M≥2.

[0020] Step 3: Adopt a dimensionality reduction algorithm to extract the most characteristic curve T from the basic curve set.

[0021] Step 4: Perform singular spectrum decomposition on curve T. According to the contribution rates of the components of the singular spectrum, divide different combinations and perform singular spectrum reconstruction to obtain a set B = {b1, b2, b3......, b N} containing N reconstructed curves.

[0022] Step 5: Use the reconstructed curve set to divide the sedimentary cycles into medium - long - period cycles and short - period cycles according to the different signals of the reconstructed curves.

[0023] The effectiveness of the method of the present invention is verified below in conjunction with specific embodiments:

[0024] Embodiment 1

[0025] Collect geological and logging data within the work area, determine the target intervals for sedimentary cycle division and the main lithology. The strata in the area are mainly sandstone and mudstone. Accordingly, select natural gamma (GR), spontaneous potential (SP), acoustic travel time (DT), and resistivity (RT) as the basic curves for sedimentary cycle division.

[0026] Taking Well - 1 as an example, the depth of the target interval is 710 - 860m. Extract 4 basic curves for outlier correction and normalization, and then calculate the correlation coefficients R between the curves respectively. Among them, R GR-SP = 0.66, R GR-DT = 0.67, R GR-RT = - 0.35, R SP-DT = 0.38, R SP-RT = - 0.23, R DT-RT = - 0.24. It can be seen from this that GR is positively correlated with SP and DT, and RT is negatively correlated with other curves. In this example, set RT′ = - RT, so that RT′ and other curves become positively correlated. Figure 2 The basic curve set A = {GR, SP, DT, RT'} that all show positive correlation.

[0027] In this example, principal component analysis (PCA) is adopted. Through linear transformation, multi-dimensional data is mapped into a low-dimensional subspace to achieve the goal of dimensionality reduction. Using the PCA dimensionality reduction algorithm, the maximum eigenvalue is obtained from the basic curve set A, and a most characteristic new curve T is extracted.

[0028] Taking the interval of 790 - 810m as an example Figure 2 As shown, the new curve T integrates the characteristics of multiple curves and retains the representative main information.

[0029] Perform singular spectrum decomposition on the new curve T. The analysis window length ranges from 2 - 5m. In this example, the minimum analysis window length is set to 2.5m, and the total number of singular spectrum components obtained is Statistical variances of each component {var1, var2, var3....var 60}, and define the contribution rate of each component as where var i is the variance of the singular spectrum component. Figure 3 Shows the contribution rate diagram of each singular spectrum component, on which multiple inflection points of the contribution rate change are marked. According to the definition of the singular spectrum and the contribution rate diagram of each component, 60 singular spectrum components are divided into 5 combinations of D = 1, D = 2 - 3, D = 4 - 13, D = 14 - 23, and D = 24 - 60, and signal reconstruction is performed respectively. Figure 4 The first curve in is curve T, and the 2nd - 6th are the reconstructed curve set B = {b1, b2, b3, b4, b5}. The ordinate of the curve is depth, with the unit of m, and the abscissa is the amplitude value, without unit. Among them, the reconstructed signal b1 contains trend information and can be used to divide medium - long - period cycles (the horizontal thick dashed line); the reconstructed signals of the remaining components are used to divide short - period cycles (the horizontal thin dashed line), especially the 3 curves of b1, b2, and b3, to complete the division of sedimentary cycles.

[0030] The above - mentioned embodiments further verify the effectiveness of the present invention.

[0031] The present invention is a method for dividing sedimentary cycles based on multiple logging curves. By analyzing geological and logging data, logging curves sensitive to sedimentary cyclicity are selected. After outlier correction and normalization processing, mathematical transformation is used to ensure positive correlation between each curve. A dimensionality reduction algorithm is used to obtain a characteristic new curve. Using singular spectrum decomposition and reconstruction algorithms, effective information is obtained, the influence of noise is reduced, and sedimentary cycles are divided. This method is a method for dividing sedimentary cycles based on multiple logging curves. The implementation process requires five steps, the operation process is simple, the division efficiency is high, and it further overcomes the problem of many interference factors in the existing division methods.

Claims

1. A method for dividing sedimentary cycles based on multiple logging curves, characterized in that, The specific division steps are as follows: Obtain logging curves, and after preprocessing the logging curves, obtain a set of basic curves A , extract the most characteristic curves from the set of basic curves A , perform singular spectrum decomposition on the most characteristic curves to obtain a set of reconstructed curves B , and use the signals of the reconstructed curves to complete the division of sedimentary cycles; The specific steps for obtaining well logging curves are as follows: collect geological and well logging data within the work area, determine the target interval for sedimentary cycle division, formation lithology information, and select M well logging curves sensitive to sedimentary cyclicity; The steps for extracting the most characteristic curve are as follows: The PCA dimensionality reduction algorithm is used. Through linear transformation, multi-dimensional data is mapped into a low-dimensional subspace to achieve the goal of dimensionality reduction. Specifically, the PCA dimensionality reduction algorithm is used to obtain the maximum eigenvalue from the basic curve set and extract a most characteristic curve. The preprocessing steps of the logging curves are as follows: After performing outlier correction and normalization on each logging curve respectively, the correlation coefficients between the curves are calculated by using the method of linear fitting, and through mathematical transformation, a positive correlation relationship is ensured among all the curves, thereby obtaining the basic curve set of the logging curves ; The specific steps for the set of reconstructed curves are as follows: According to the contribution rates of the components of the singular spectrum, different combinations are divided, and singular spectrum reconstruction is carried out to obtain a set of reconstructed curves containing N curves , where N is the number of reconstructed curves; The principle for sedimentary cycle division is: medium-long period cycles and short period cycles.

Images