Method and apparatus for establishing low-frequency impedance model based on multi-element information control

Abstract

The application provides a low-frequency impedance model establishment method and device based on multi-element information control. First, local preserving projection algorithm is used for seismic attribute dimension reduction to obtain characteristic attributes, and an LSTM network is used to calculate a low-frequency impedance model M1 under a deep model framework. Then, inverse proportional weighting is used for well interpolation extrapolation to obtain a well interpolation low-frequency impedance model M2. Finally, a whale optimization algorithm is used to obtain optimal weights of the two low-frequency impedance models, and linear combination is performed to obtain a final low-frequency impedance model M_final. The establishment of the low-frequency impedance model in the application is controlled by seismic, horizon and logging multi-element information. The application has high well coincidence with the low-frequency impedance model, and has higher resolution than a conventional well interpolation low-frequency impedance model, and has high popularization and application value.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of geophysical exploration, and more particularly, to a low-frequency impedance model establishing method and device based on multi-information control. BACKGROUND

[0002] An accurate low-frequency impedance model can effectively improve the precision of seismic inversion, and further improve the precision of reservoir prediction. However, low-frequency information is often missing in conventional seismic data, and how to obtain a relatively accurate low-frequency impedance model has become a difficulty in seismic inversion.

[0003] Currently, the low-frequency impedance model establishing methods mainly include the following methods:

[0004] (1) A low-frequency impedance model establishing method based on well interpolation, which has the defect that when the well spatial distribution is unreasonable, the method is prone to cause a bull's eye phenomenon:

[0005] (2) A low-frequency impedance model establishing method constrained by a seismic velocity body, which is limited to compensating for low-frequency information below 2Hz, and cannot compensate for all missing low-frequency information in seismic data:

[0006] (3) A phase-controlled low-frequency impedance model establishing method, which has the defect that accurate seismic facies are difficult to obtain, and seismic facies cannot fully reflect lithology and fluid information, so that the method is difficult to obtain an accurate low-frequency impedance model.

[0007] Since the above methods for establishing a low-frequency impedance model have certain limitations, there is an urgent need for a low-frequency impedance model establishing method with strong adaptability. SUMMARY

[0008] Therefore, the present application aims to provide a low-frequency impedance model establishing method with strong adaptability by comprehensively utilizing multi-information.

[0009] According to an aspect of the present application, a low-frequency impedance model establishing method based on multi-information control is provided, which comprises:

[0010] Step 1: Preprocessing the logging curve to obtain an optimized logging curve, wherein the optimized logging curve comprises a low-frequency impedance curve:

[0011] Step 2: Calibrating the well and seismic data by combining the optimized logging curve with the seismic data and horizon data:

[0012] Step 3: Extracting a plurality of seismic attributes from the seismic data:

[0013] Step 4: Reducing the dimensionality of the plurality of seismic attributes by using a local preserving projection algorithm to obtain an optimized feature attribute:

[0014] Step 5, obtaining a low-frequency impedance model M1 under a deep model framework by using the optimized feature attribute and the low-frequency impedance curve:

[0015] Step 6, obtaining a low-frequency impedance model M2 by using the layer data and the low-frequency impedance curve, and performing well interpolation extrapolation by using inverse proportional weighting:

[0016] Step 7, determining a weight coefficient of the low-frequency impedance model M1 and the low-frequency impedance model M2 by using a whale optimization algorithm, performing linear combination, and obtaining a final low-frequency impedance model M_final.

[0017] In some embodiments, in the step 1, the pre-processing of the logging curve includes correction and inter-well consistency processing of the logging curve.

[0018] In some embodiments, in the step 4, the plurality of seismic attributes are dimensionally reduced to obtain the optimized feature attribute Y={y1,y2,...,y m} based on the following formula:

[0019] y i =B T x i

[0020] Wherein, X={x1,x2,...,x n} is a data set of the plurality of seismic attributes, B is a projection matrix B, n is the number of the seismic attributes before dimension reduction, and m is the number of the feature attribute obtained after dimension reduction.

[0021] In some embodiments, in the step 4, the projection matrix B is determined by minimizing the following objective function:

[0022]

[0023] Wherein, W ij represents a weight matrix,

[0024]

[0025] In some embodiments, the optimized feature attribute includes a well trace feature attribute, and the step 5 specifically includes:

[0026] Extracting the well trace feature attribute from the optimized feature attribute, and combining the low-frequency impedance curve to make a sample set:

[0027] Learning a nonlinear mapping relationship between the well trace feature attribute and the low-frequency impedance curve based on a long short-term memory network:

[0028] The nonlinear mapping relationship is extended to the entire target area to obtain low-frequency impedance attributes of all positions in the target area, so as to generate the low-frequency impedance model M1.

[0029] In some embodiments, the step 7 specifically comprises:

[0030] The final low-frequency impedance model M_final is obtained based on the following formula:

[0031] M_final = a1*M1 + a2*M2

[0032] Wherein, a1 and a2 are weight coefficients determined based on a whale optimization algorithm.

[0033] According to another aspect of the present application, a low-frequency impedance model establishment device based on multi-element information control is provided, which comprises:

[0034] A well logging curve preprocessing unit is configured to preprocess well logging curves to obtain optimized well logging curves, wherein the optimized well logging curves comprise low-frequency impedance curves.

[0035] A well-seismic calibration unit is configured to calibrate wells and seismic data by combining the optimized well logging curves with seismic data and horizon data.

[0036] A seismic attribute extraction unit is configured to extract multiple seismic attributes from the seismic data.

[0037] A seismic attribute dimension reduction unit is configured to reduce the dimension of the multiple seismic attributes by using a local reserve projection algorithm to obtain optimized feature attributes.

[0038] A first low-frequency impedance model acquisition unit is configured to acquire a low-frequency impedance model M1 under a depth model framework by using the optimized feature attributes and the low-frequency impedance curves.

[0039] A second low-frequency impedance model acquisition unit is configured to acquire a low-frequency impedance model M2 by using the horizon data and the low-frequency impedance curves and performing well interpolation extrapolation by using inverse proportional weighting.

[0040] A final low-frequency impedance model acquisition unit is configured to determine weight coefficients of the low-frequency impedance model M1 and the low-frequency impedance model M2 by using a whale optimization algorithm, perform linear combination, and acquire a final low-frequency impedance model M_final.

[0041] In some embodiments, in the well logging curve preprocessing unit, preprocessing the well logging curves comprises correcting the well logging curves and performing inter-well consistency processing.

[0042] In some implementations, the seismic attribute dimensionality reduction unit is specifically used to reduce the dimensionality of the plurality of seismic attributes based on the following formula to obtain optimized feature attributes Y = {y1, y2, ..., y...} m}:

[0043] y i =B T x i

[0044] Where X = {x1, x2, ..., x} n} represents the dataset of the multiple seismic attributes, B is the projection matrix B, n is the number of seismic attributes before dimensionality reduction, and m is the number of feature attributes obtained after dimensionality reduction.

[0045] In some implementations, the projection matrix B is determined in the seismic attribute dimensionality reduction unit by minimizing the following objective function:

[0046]

[0047] Among them, W ij Represents the weight matrix,

[0048]

[0049] In some implementations, the optimized feature attributes include wellbore access feature attributes, and the first low-frequency impedance model acquisition unit is specifically used for:

[0050] Extract the wellbore access feature attributes from the optimized feature attributes, and combine them with the low-frequency impedance curve to create a sample set:

[0051] The nonlinear mapping relationship between the wellbore characteristics and the low-frequency impedance curve is learned based on a long short-term memory network:

[0052] The nonlinear mapping relationship is extended to the entire target region to obtain the low-frequency impedance properties at all locations within the target region, thereby generating the low-frequency impedance model M1.

[0053] In some implementations, the final low-frequency impedance model acquisition unit is specifically used for:

[0054] The final low-frequency impedance model M_final is obtained based on the following formula:

[0055] M_final = α1*M1 + α2*M2

[0056] Where α1 and α2 are weight coefficients determined based on the whale optimization algorithm.

[0057] According to another aspect of the present invention, an electronic device is also provided, the electronic device comprising:

[0058] a memory storing executable instructions:

[0059] a processor running the executable instructions in the memory to implement the low-frequency impedance model establishing method based on multi-element information control described above.

[0060] According to another aspect of the present application, a computer readable storage medium is also provided, which stores a computer program that is executed by a processor to implement the low-frequency impedance model establishing method based on multi-element information control described above.

[0061] The beneficial effects of the present application at least include:

[0062] 1. The present application integrates seismic, well logging and horizon multi-element data, and combines data processing, deep learning and optimization algorithms to propose a low-frequency impedance model establishing scheme with strong adaptability:

[0063] 2. The present application uses local preserving projection algorithm (LPP) for seismic attribute dimension reduction, which can effectively eliminate redundant information in the attribute and extract feature attributes sensitive to low-frequency impedance, providing a good data basis for subsequent modeling:

[0064] 3. The present application obtains a low-frequency impedance model under a deep model framework based on long short-term memory network (LSTM), which can learn the nonlinear mapping relationship between feature attributes and low-frequency impedance and establish a data-driven low-frequency impedance model:

[0065] 4. The present application uses well interpolation method to obtain another low-frequency impedance model, providing a contrast model with physical constraints:

[0066] 5. The present application uses whale optimization algorithm to find the optimal combination of the two low-frequency impedance models, fully utilizes the advantages of data-driven models and physically constrained models to obtain a more accurate comprehensive low-frequency impedance model:

[0067] 6. The final low-frequency impedance model obtained according to the present application has the characteristics of strong adaptability, high quality, good precision and high resolution, which can effectively support subsequent seismic inversion and resource evaluation work.

[0068] In summary, the technical scheme according to the present application combines multi-element information, combines data driving and physical constraints, and obtains a final low-frequency impedance model with strong adaptability, high consistency with well data and good resolution, achieving very beneficial technical effects.

[0069] The method and apparatus of the present application have other features and advantages which will be apparent from or which will be elucidated with regard to the drawings that show an exemplary embodiment of the application and the following description of the application. These drawings show, in: BRIEF DESCRIPTION OF DRAWINGS

[0070] These and other objects, features and advantages of the present application will become apparent after review of the following disclosure, figures and drawings, in which like reference numerals refer to similar or identical elements throughout, and in which:

[0071] Figure 1 A flow chart of a method for establishing a low frequency impedance model based on multi-information control according to an embodiment of the present application is shown.

[0072] Figure 2 A typical well calibration map for a target zone according to an exemplary embodiment of the present application is shown.

[0073] Figure 3 A well interpolation low frequency impedance model profile according to an exemplary embodiment of the present application is shown.

[0074] Figure 4 A final low frequency impedance model profile according to an exemplary embodiment of the present application is shown. DETAILED DESCRIPTION

[0075] Preferred embodiments of the present application will be described herein below with reference to the accompanying drawings. While the preferred embodiments of the present application are shown in the drawings, it is understood that the present application can be embodied in various forms without being limited by the embodiments set forth herein. Rather, these embodiments are provided so that the present application will be thorough and complete, and will fully convey the scope of the present application to those skilled in the art.

[0076] Example 1

[0077] Figure 1 A flow chart of a method for establishing a low frequency impedance model based on multi-information control according to an embodiment of the present application is shown. As shown, the example includes steps 1-7.

[0078] Step 1, pre-process the well logging curves to obtain optimized well logging curves, the optimized well logging curves including low frequency impedance curves.

[0079] In some embodiments, pre-processing the well logging curves includes correcting the well logging curves and performing inter-well consistency processing.

[0080] According to the pre-processing of the well logging curve in the embodiment, the purposes of improving data quality, increasing signal-to-noise ratio and improving consistency of inter-well data can be achieved.

[0081] Step 2, the optimized well logging curve is combined with seismic data and horizon data to perform well-seismic calibration.

[0082] The well-seismic calibration refers to a process of associating well logging data and seismic data, and the well-seismic calibration is mainly used to integrate information of different data sources in a unified depth coordinate system.

[0083] Specifically, the workflow of the well-seismic calibration can include the following steps.

[0084] (1) Well logging data pre-processing: performing correction of well logging data, eliminating horizon effects, and obtaining a real impedance curve of a well section, which can be completed in step 1 above:

[0085] (2) Seismic data pre-processing: performing data correction of a seismic profile, increasing signal-to-noise ratio, and enhancing contrast of stratum reflection coefficients:

[0086] (3) Merging well site information: associating well logging curves with horizons of strata under a well to establish a corresponding relationship between well depth and time-depth:

[0087] (4) Extracting seismic amplitudes: extracting seismic profile amplitudes near a well site to obtain seismic reflection coefficients:

[0088] (5) Calibration: according to a requirement, a specific algorithm, such as linear regression, statistical regression or the like, is used to calibrate well logging impedance and seismic reflection coefficients to establish a mathematical corresponding relationship therebetween;

[0089] (6) Inspection: according to the established corresponding relationship, a theoretical seismic profile is calculated, and a comparison is performed between the theoretical seismic profile and an actual seismic profile to judge calibration effect;

[0090] (7) Iterative optimization: according to a comparison result, a parameter is adjusted, and the above steps are repeated until the calibration effect reaches an expectation.

[0091] According to the present application, fine well-seismic calibration can be performed by using seismic, horizon and well logging data, and information obtained through different measurement methods is converted to a same domain, which lays a good foundation for subsequent establishment of a low-frequency impedance model.

[0092] Step 3, a plurality of seismic attributes are extracted from the seismic data.

[0093] Amplitude, frequency, phase and various seismic attributes can be extracted by using seismic data.

[0094] Step 4: Use the local preservative projection algorithm to reduce the dimensionality of the multiple seismic attributes to obtain optimized feature attributes.

[0095] For the various seismic attributes extracted in step 3, the seismic attributes can be dimensionality reduced based on the Local Preserving Projection (LPP) algorithm to obtain feature attributes, remove redundant information from various seismic attributes, and obtain feature attributes that are sensitive to low-frequency impedance, such as well-side tunnel feature attributes.

[0096] In some implementations, the multiple seismic attributes can be dimensionality reduced specifically based on the following formula to obtain optimized feature attributes Y = {y1, y2, ..., y...} m}:

[0097] y i =B T x i (1)

[0098] Where X = {x1, x2, ..., x} n} represents the dataset of the multiple seismic attributes, B is the projection matrix B, n is the number of seismic attributes before dimensionality reduction, and m is the number of feature attributes obtained after dimensionality reduction.

[0099] In some examples, the projection matrix B can be determined by minimizing the following objective function:

[0100]

[0101] Among them, W ij Represents the weight matrix,

[0102]

[0103] By introducing constraints and combining them with equation (1), the problem can be transformed into solving the following generalized eigenvalue problem:

[0104] XLX T B = λXDX T B (4)

[0105] Where λ is the generalized eigenvalue, L is the generalized eigenvector, D is the degree matrix, and X is the high-dimensional sample matrix, i.e., X = {x1, x2, ..., x...} n}, where B is the projection matrix.

[0106] When solving the problem, the generalized eigenvalue λ in formula (4) can be calculated first. The calculated λ can be substituted into the original problem to obtain the eigenvector. Finally, the eigenvectors can be combined to obtain the projection matrix B.

[0107] Step 5: Using the optimized feature attributes and the low-frequency impedance curve, obtain the low-frequency impedance model M1 under the deep model framework.

[0108] In some embodiments, the optimized feature attributes include wellbore trace feature attributes, and the step 5 specifically includes:

[0109] The wellbore trace feature attributes are extracted from the optimized feature attributes, and a sample set is made in combination with the low-frequency impedance curves:

[0110] A nonlinear mapping relationship between the wellbore trace feature attributes and the low-frequency impedance curves is learned based on a long short-term memory network (LSTM):

[0111] The nonlinear mapping relationship is extended to the entire target area to obtain low-frequency impedance attributes of all positions in the target area, so as to generate the low-frequency impedance model M1.

[0112] In the present embodiment, the wellbore trace feature attributes are low-frequency impedance-sensitive feature attributes obtained after dimensionality reduction by a local reserve projection algorithm.

[0113] The long short-term memory network (LSTM) used in the present embodiment is a variant of a recurrent neural network (RNN) and can effectively solve the problem that RNNs miss long-term memory.

[0114] In the present embodiment, the nonlinear mapping relationship between the feature attributes and the low-frequency impedance is obtained based on the LSTM, and then a low-frequency impedance model M1 under a deep model framework is acquired.

[0115] In the present embodiment, after the LSTM model learns the nonlinear mapping relationship between the wellbore trace feature attributes and the low-frequency impedance, the mapping relationship is extended and applied to the entire target area, that is, the trained LSTM model is used to predict the entire area, so as to obtain low-frequency impedance attributes of all positions in the target area, thereby generating a low-frequency impedance model M1 of the entire area.

[0116] Step 6: Using the stratigraphic data and the low-frequency impedance curves, well interpolation extrapolation is performed by inverse proportional weighting to obtain a low-frequency impedance model M2.

[0117] In some embodiments, inverse proportional weighting can be performed by the following steps:

[0118] (1) The distance between each target interpolation point and the adjacent sampling points is calculated:

[0119] (2) Each adjacent sampling point is assigned a weight, and the weight is inversely proportional to the distance:

[0120] (3) The low-frequency impedance value of each sampling point is multiplied by the corresponding weight to obtain the contribution of the sampling point to the target difference point:

[0121] (4) The contributions of all adjacent logging sampling points are summed up to obtain the final low-frequency impedance value of the target interpolation point.

[0122] The application adopts inverse proportional weighted interpolation, near-well point data has greater contribution, conforms to the principle of distance attenuation, and is beneficial to obtaining a high-quality low-frequency impedance model.

[0123] According to the application, local interpolation can be performed in the area with wells first, and then the interpolation result is extrapolated to the entire well-free area, so as to realize low-frequency impedance prediction of the entire target area.

[0124] Step 7, the weight coefficients of the low-frequency impedance model M1 and the low-frequency impedance model M2 are obtained by using the whale optimization algorithm, linear combination is performed, and the final low-frequency impedance model M_final is obtained.

[0125] In some embodiments,

[0126] The final low-frequency impedance model M_final is obtained based on the following formula:

[0127] M_final = α1*M1 + α2*M2 (5)

[0128] Wherein, α1 and α2 are weight coefficients determined based on the whale optimization algorithm.

[0129] The whale optimization algorithm is an optimization algorithm simulating the hunting process of a whale. One whale individual corresponds to one solution, and the position of the whale individual is constantly updated to search for the solution of the problem until the termination condition is met.

[0130] The whale optimization algorithm (Whale Optimization Algorithm) is a bionic optimization algorithm, which simulates the behavior of a whale searching for food. The algorithm corresponds one whale individual to one solution, and the position of the whale individual is constantly updated to search for the solution of the problem until the termination condition is met. The algorithm is modeled based on the following three whale movement modes:

[0131] (1) Surround the prey

[0132] Simulate the spiral movement of a whale around food. In the algorithm, the position of the whale individual is updated to simulate this spiral movement to approach the optimal solution:

[0133] (2) Search for food

[0134] Simulate the feeding of a whale through bubble net. In the algorithm, the position of the whale is randomly mutated, which is equivalent to searching for a new solution by bubble net:

[0135] (3) Search

[0136] Whale random search for food. In the algorithm, a random disturbance is added to the individual to increase population diversity and avoid falling into local optimum:

[0137] The application determines the weight coefficients of the low-frequency impedance model M1 and the low-frequency impedance model M2 by adopting the whale optimization algorithm, so that the solving process has the advantages of strong global search ability and fast convergence speed, and the global optimal solution can be quickly found, and the quality of the finally obtained low-frequency impedance model M_final is further improved.

[0138] The embodiment has at least the following innovations compared with the existing technology:

[0139] (1) The embodiment is based on the local reserved projection algorithm, which can effectively eliminate the redundant information in the seismic attribute, and provides a solid data foundation for subsequent low-frequency impedance model establishment;

[0140] (2) The embodiment obtains the nonlinear mapping relationship between the characteristic attribute and the low-frequency impedance curve based on the LSTM network according to the time sequence characteristics of the seismic signal;

[0141] (3) The embodiment can effectively obtain the optimal weight coefficient based on the whale optimization algorithm, and then optimally combine the low-frequency impedance model under the deep model framework and the well interpolation low-frequency impedance model to obtain the final low-frequency impedance model;

[0142] (4) The embodiment establishes a low-frequency impedance model establishment method process controlled by multiple information, and fully integrates seismic, horizon and logging multiple information in the low-frequency impedance model establishment process.

[0143] Example 2

[0144] According to an embodiment of the application, a low-frequency impedance model establishment device based on multiple information control is provided.

[0145] The device comprises:

[0146] A logging curve preprocessing unit is configured to preprocess the logging curve to obtain an optimized logging curve, wherein the optimized logging curve comprises a low-frequency impedance curve;

[0147] A well-seismic calibration unit is configured to combine the optimized logging curve with seismic data and horizon data to perform well-seismic calibration;

[0148] A seismic attribute extraction unit is configured to extract multiple seismic attributes from the seismic data;

[0149] A seismic attribute dimension reduction unit is configured to perform dimension reduction on the multiple seismic attributes by using a local reserved projection algorithm to obtain optimized characteristic attributes;

[0150] The first low-frequency impedance model acquisition unit is used to acquire the low-frequency impedance model M1 under the deep model framework by utilizing the optimized feature attributes and the low-frequency impedance curve.

[0151] The second low-frequency impedance model acquisition unit is used to obtain the low-frequency impedance model M2 by using the stratigraphic data and the low-frequency impedance curve, and performing well interpolation extrapolation with inverse weighting.

[0152] The final low-frequency impedance model acquisition unit is used to determine the weight coefficients of the low-frequency impedance model M1 and the low-frequency impedance model M2 using the whale optimization algorithm, and then perform a linear combination to obtain the final low-frequency impedance model M_final.

[0153] In some embodiments, the well logging curve preprocessing unit performs well logging curve preprocessing, which includes correcting the well logging curves and performing inter-well consistency processing.

[0154] In some implementations, the seismic attribute dimensionality reduction unit is specifically used to reduce the dimensionality of the plurality of seismic attributes based on the following formula to obtain optimized feature attributes Y = {y1, y2, ..., y...} m}:

[0155] y i =B T x i (1)

[0156] Where X = {x1, x2, ..., x} n} represents the dataset of the multiple seismic attributes, B is the projection matrix B, n is the number of seismic attributes before dimensionality reduction, and m is the number of feature attributes obtained after dimensionality reduction.

[0157] In some implementations, the projection matrix B is determined in the seismic attribute dimensionality reduction unit by minimizing the following objective function:

[0158]

[0159] Among them, W ij Represents the weight matrix,

[0160]

[0161] By introducing constraints and combining them with equation (1), the problem can be transformed into solving the following generalized eigenvalue problem:

[0162] XLX T B = λXDX T B (4)

[0163] Wherein, λ is a generalized eigenvalue, L is a generalized eigenvector, D is a degree matrix, X is a high-dimensional sample matrix, that is, X={x1, x2,..., x n}, B is a projection matrix.

[0164] In solving, the generalized eigenvalue λ in formula (4) is solved first, the solved λ is substituted into the original problem, the eigenvector is solved, and finally the projection matrix B is obtained by combining the eigenvector.

[0165] In some embodiments, the optimized feature attribute includes a wellbore feature attribute, and the first low-frequency impedance model acquisition unit is specifically configured to:

[0166] The wellbore feature attribute is extracted from the optimized feature attribute, and a sample set is made in combination with the low-frequency impedance curve:

[0167] A nonlinear mapping relationship between the wellbore feature attribute and the low-frequency impedance curve is learned based on a long short-term memory network:

[0168] The nonlinear mapping relationship is extended to the entire target area to obtain low-frequency impedance attributes of all positions in the target area, so as to generate the low-frequency impedance model M1.

[0169] In the embodiment, the wellbore feature attribute is a feature attribute sensitive to low-frequency impedance obtained after dimensionality reduction by a local reserve projection algorithm.

[0170] The long short-term memory network (LSTM) used in the embodiment is a variant of a recurrent neural network (RNN) and can effectively solve the problem that RNN can miss long-term memory.

[0171] In the embodiment, the nonlinear mapping relationship between the feature attribute and the low-frequency impedance is obtained based on the LSTM, and then the low-frequency impedance model M1 under the deep model framework is obtained.

[0172] In the embodiment, after the LSTM model learns the nonlinear mapping relationship between the feature attribute of the wellbore and the low-frequency impedance, the mapping relationship is extended and applied to the entire target area, that is, the trained LSTM model is used to predict the entire area, so as to obtain low-frequency impedance attributes of all positions in the target area, thereby generating a low-frequency impedance model M1 in the entire area.

[0173] In some embodiments, in the second low-frequency impedance model acquisition unit, the inverse proportional weighting can be performed through the following steps:

[0174] (1) Calculate the distance between each target interpolation point and the adjacent sampling point:

[0175] (2) Assign a weight to each adjacent sampling point, which is inversely proportional to the distance:

[0176] (3) Multiply the low-frequency impedance value of each sampling point by the corresponding weight to obtain the contribution of the sampling point to the target difference point:

[0177] (4) Sum the contributions of all adjacent logging sampling points to obtain the final low-frequency impedance value of the target interpolation point.

[0178] The present application adopts inverse proportional weighted interpolation, the near well point data contribution is greater, conforms to the principle of distance attenuation, and is beneficial to obtain high-quality low-frequency impedance model.

[0179] According to the present application, local interpolation can be performed in the area with wells first, and then the interpolation results are extrapolated to the entire well-free area to realize low-frequency impedance prediction of the entire target area.

[0180] In some embodiments, the final low-frequency impedance model acquisition unit is specifically configured to:

[0181] The final low-frequency impedance model M_final is obtained based on the following formula:

[0182] M_final=α1*M1+α2*M2

[0183] Wherein, alpha1 and alpha2 are weight coefficients determined based on the whale optimization algorithm.

[0184] The whale optimization algorithm is an optimization algorithm that simulates the hunting process of a whale. A whale individual corresponds to a solution, and the position of the whale individual is constantly updated to search for the solution of the problem until the termination condition is met.

[0185] The whale optimization algorithm (Whale Optimization Algorithm) is a bionic optimization algorithm that simulates the behavior of a whale searching for food. The algorithm corresponds a whale individual to a solution, and the position of the whale individual is constantly updated to search for the solution of the problem until the termination condition is met. The algorithm is modeled based on the following three whale movement modes:

[0186] (1) Surround the prey

[0187] Simulate the spiral movement of a whale around food. In the algorithm, the position of the whale individual is updated to simulate this spiral motion to approach the optimal solution:

[0188] (2) Search for food

[0189] Simulate the feeding of a whale through bubble net. In the algorithm, the position of the whale is randomly mutated, which is equivalent to searching for a new solution by bubble net:

[0190] (3) Search

[0191] The whale simulates random search for food. In the algorithm, a random disturbance is added to the individual to increase the population diversity and avoid falling into local optimum:

[0192] The application determines the weight coefficients of the low-frequency impedance model M1 and the low-frequency impedance model M2 by adopting the whale optimization algorithm, so that the solving process has the advantages of strong global search ability and fast convergence speed, can quickly find the global optimal solution, and further improves the quality of the finally obtained low-frequency impedance model M_final.

[0193] Other detailed descriptions and advantages related to the present embodiment can be referred to the corresponding descriptions in the foregoing embodiments, which will not be repeated here.

[0194] Example 3

[0195] According to another aspect of the application, an electronic device is also provided. The electronic device comprises:

[0196] a memory storing executable instructions:

[0197] a processor running the executable instructions in the memory to implement the low-frequency impedance model establishment method based on multi-element information control according to the application.

[0198] The method comprises the following steps:

[0199] Step 1, pre-processing the logging curve to obtain an optimized logging curve, wherein the optimized logging curve comprises a low-frequency impedance curve:

[0200] Step 2, combining the optimized logging curve with seismic data and horizon data to perform well-seismic calibration:

[0201] Step 3, extracting a plurality of seismic attributes from the seismic data:

[0202] Step 4, using a locally reserved projection algorithm to reduce the dimensionality of the plurality of seismic attributes to obtain an optimized feature attribute:

[0203] Step 5, using the optimized feature attribute and the low-frequency impedance curve to obtain a low-frequency impedance model M1 under a depth model framework:

[0204] Step 6, using the horizon data and the low-frequency impedance curve to perform well interpolation extrapolation by inverse proportional weighting to obtain a low-frequency impedance model M2:

[0205] Step 7, using a whale optimization algorithm to determine the weight coefficients of the low-frequency impedance model M1 and the low-frequency impedance model M2, performing linear combination to obtain a final low-frequency impedance model M_final.

[0206] In some embodiments, step 1, preprocessing the logging curves includes correcting the logging curves and performing inter-well consistency processing.

[0207] In some implementations, step 4 specifically involves dimensionality reduction of the plurality of seismic attributes based on the following formula to obtain optimized feature attributes Y = {y1, y2, ..., y...} m}:

[0208] y i =B T x i

[0209] Where X = {x1, x2, ..., x} n} represents the dataset of the multiple seismic attributes, B is the projection matrix B, n is the number of seismic attributes before dimensionality reduction, and m is the number of feature attributes obtained after dimensionality reduction.

[0210] In some implementations, in step 4, the projection matrix B is determined by minimizing the following objective function:

[0211]

[0212] Among them, W ij Represents the weight matrix,

[0213]

[0214] In some implementations, the optimized feature attributes include wellbore access feature attributes, and step 5 specifically includes:

[0215] Extract the wellbore access feature attributes from the optimized feature attributes, and combine them with the low-frequency impedance curve to create a sample set:

[0216] The nonlinear mapping relationship between the wellbore characteristics and the low-frequency impedance curve is learned based on a long short-term memory network:

[0217] The nonlinear mapping relationship is extended to the entire target region to obtain the low-frequency impedance properties at all locations within the target region, thereby generating the low-frequency impedance model M1.

[0218] In some implementations, step 7 specifically includes:

[0219] The final low-frequency impedance model M_final is obtained based on the following formula:

[0220] M_final = α1*M1 + α2*M2

[0221] Where α1 and α2 are weight coefficients determined based on the whale optimization algorithm.

[0222] In particular, the memory can include one or more computer program products that can include various forms of computer-readable storage media, such as volatile and / or non-volatile memory. The volatile memory, for example, can include random access memory (RAM), and / or a cache, and / or the like. The non-volatile memory, for example, can include read-only memory (ROM), hard disk, flash memory, and / or the like.

[0223] The processor can be a central processing unit (CPU) or other form of processing unit that has data processing and / or instruction execution capabilities, and can control other components in the electronic device to perform desired functions. In one embodiment of the present application, the processor is used to run the computer-readable instructions stored in the memory.

[0224] The technical solution according to the present embodiment fuses multiple information, and obtains a final low-frequency impedance model with strong adaptability, high coincidence degree with uphole data and good resolution by combining data driving and physical constraints, thereby achieving very beneficial technical effects.

[0225] Detailed descriptions related to the present embodiment can refer to the corresponding descriptions in the foregoing embodiments, which will not be repeated here.

[0226] Example 4

[0227] According to another aspect of the present application, there is also provided a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the low-frequency impedance model establishing method based on multi-information control according to the present application.

[0228] The method comprises the following steps:

[0229] Step 1: pre-processing the logging curves to obtain optimized logging curves, wherein the optimized logging curves include a low-frequency impedance curve:

[0230] Step 2: combining the optimized logging curves with seismic data and horizon data to perform well-seismic calibration:

[0231] Step 3: extracting multiple seismic attributes from the seismic data:

[0232] Step 4: using a locally reserved projection algorithm to reduce the dimensionality of the multiple seismic attributes to obtain optimized feature attributes:

[0233] Step 5: using the optimized feature attributes and the low-frequency impedance curve to obtain a low-frequency impedance model M1 under a depth model framework:

[0234] Step 6: Using the stratigraphic data and the low-frequency impedance curve, perform well interpolation extrapolation using inverse proportional weighting to obtain the low-frequency impedance model M2.

[0235] Step 7: Use the whale optimization algorithm to determine the weight coefficients of the low-frequency impedance model M1 and the low-frequency impedance model M2, and perform a linear combination to obtain the final low-frequency impedance model M_final.

[0236] In some embodiments, step 1, preprocessing the logging curves includes correcting the logging curves and performing inter-well consistency processing.

[0237] In some implementations, step 4 specifically involves dimensionality reduction of the plurality of seismic attributes based on the following formula to obtain optimized feature attributes Y = {y1, y2, ..., y...} m}:

[0238] y i =B T x i

[0239] Where X = {x1, x2, ..., x} n} represents the dataset of the multiple seismic attributes, B is the projection matrix B, n is the number of seismic attributes before dimensionality reduction, and m is the number of feature attributes obtained after dimensionality reduction.

[0240] In some implementations, in step 4, the projection matrix B is determined by minimizing the following objective function:

[0241]

[0242] Among them, W ij Represents the weight matrix,

[0243]

[0244] In some implementations, the optimized feature attributes include wellbore access feature attributes, and step 5 specifically includes:

[0245] Extract the wellbore access feature attributes from the optimized feature attributes, and combine them with the low-frequency impedance curve to create a sample set:

[0246] The nonlinear mapping relationship between the wellbore characteristics and the low-frequency impedance curve is learned based on a long short-term memory network:

[0247] The nonlinear mapping relationship is extended to the entire target region to obtain the low-frequency impedance properties at all locations within the target region, thereby generating the low-frequency impedance model M1.

[0248] In some embodiments, the step 7 specifically comprises:

[0249] The final low-frequency impedance model M_final is obtained based on the following formula:

[0250] M_final = a1*M1 + a2*M2

[0251] Wherein, a1 and a2 are weight coefficients determined based on the whale optimization algorithm.

[0252] The computer readable storage medium according to the embodiment of the present application has non-transitory computer readable instructions stored thereon. When the non-transitory computer readable instructions are run by a processor, all or part of the steps of the method of each embodiment of the present application described above are executed.

[0253] The computer readable storage medium described above includes, but is not limited to, optical storage media (such as CD-ROM and DVD), magneto-optical storage media (such as MO), magnetic storage media (such as magnetic tape or mobile hard disk), media with built-in rewritable non-volatile memory (such as memory card), and media with built-in ROM (such as ROM cartridge).

[0254] Those skilled in the art should understand that, in order to solve the technical problem of how to obtain a good user experience effect, the embodiment can also include well-known structures such as a communication bus, an interface, etc., which should also be included in the protection scope of the present application.

[0255] The technical solution according to the embodiment fuses multiple information, obtains the final low-frequency impedance model by combining data driving and physical constraints, and has strong adaptability, high consistency with uphole data, and good resolution, thereby achieving very beneficial technical effects.

[0256] Detailed descriptions of the embodiment can be referred to the corresponding descriptions in the foregoing embodiments, which will not be repeated here.

[0257] Example 5

[0258] Figure 2 、 Figure 3 and Figure 4 An application example according to one exemplary embodiment of the present application is shown.

[0259] Figure 2 A typical well calibration map of a target area is shown.

[0260] Figure 3 A well interpolation low-frequency impedance model profile schematic diagram according to one exemplary embodiment of the present application is shown. As can be seen from Figure 3 , in the blind well area, the low-frequency impedance model M2 trend has poor consistency with the uphole low-frequency impedance trend.

[0261] Figure 4 Fig. 6 shows a final low frequency impedance model profile according to an example embodiment of the present application. From Fig. 6, it can be seen that the low frequency impedance model trend in the blind well area is consistent with the low frequency impedance trend in the well, and the final low frequency impedance model M_final has a significant improvement in resolution compared with the low frequency impedance model M2 obtained by conventional well interpolation. Figure 4

[0262] Further details about the example embodiment can be found in the corresponding description of the previous embodiments, which are not repeated here.

[0263] The above has described the embodiments of the present application, the above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and changes are obvious to those skilled in the art without departing from the scope and spirit of the described embodiments. The choice of terms used herein is intended to best explain the principles of the embodiments, practical application, or technical improvement to the technology in the market, or to enable other ordinary skilled in the art to understand the embodiments disclosed herein.​

Claims

1. A method for establishing a low-frequency impedance model based on multi-element information control, characterized in that, The method comprises: Step 1, pre-processing the well logging curve to obtain an optimized well logging curve, wherein the optimized well logging curve comprises a low-frequency impedance curve; Step 2, calibrating the well logging curve with seismic data and horizon data; Step 3, extracting a plurality of seismic attributes from the seismic data; Step 4, reducing the dimensionality of the plurality of seismic attributes by using a locally preserving projection algorithm to obtain optimized feature attributes; Step 5, obtaining a low-frequency impedance model M1 under a depth model framework by using the optimized feature attributes and the low-frequency impedance curve; Step 6, obtaining a low-frequency impedance model M2 by using the horizon data and the low-frequency impedance curve and performing well interpolation extrapolation by using inverse proportional weighting; Step 7, determining a weight coefficient of the low-frequency impedance model M1 and the low-frequency impedance model M2 by using a whale optimization algorithm, performing linear combination, and obtaining a final low-frequency impedance model M_final.

2. The method of claim 1, characterized in that, In the step 1, the pre-processing of the well logging curve comprises correcting the well logging curve and performing inter-well consistency processing.

3. The method of claim 1, wherein, In the step 4, the multiple seismic attributes are reduced dimensionally based on the following formula to obtain a data set of optimized feature attributes , representing data serial number: wherein, is the data set of the plurality of seismic attributes, B is a projection matrix, the number of the seismic attributes before dimensionality reduction is n , and the number of the feature attributes obtained after dimensionality reduction is m .

4. The method of claim 3, wherein, In the step 4, the projection matrix B is determined by minimizing the following objective function: wherein denotes a weight matrix, 。 5. The method of claim 1, wherein, The optimized feature attributes comprise well trace feature attributes, and the step 5 specifically comprises: extracting the well trace feature attributes from the optimized feature attributes and combining the low-frequency impedance curve to make a sample set; learning a nonlinear mapping relationship between the well trace feature attributes and the low-frequency impedance curve based on a long short-term memory network; extending the nonlinear mapping relationship to all positions in a target area to obtain a low-frequency impedance attribute of all positions in the target area, so as to generate the low-frequency impedance model M1.

6. The method of claim 1, wherein, The step 7 specifically comprises: obtaining the final low-frequency impedance model M_final based on the following formula: wherein, with are weight coefficients determined based on the whale optimization algorithm.

7. A low-frequency impedance model establishing device based on multi-element information control, characterized in that, The device comprises: a well logging curve pre-processing unit configured to pre-process a well logging curve to obtain an optimized well logging curve, wherein the optimized well logging curve comprises a low-frequency impedance curve; a well seismic calibration unit configured to calibrate the optimized well logging curve with seismic data and horizon data; a seismic attribute extraction unit configured to extract a plurality of seismic attributes from the seismic data; a seismic attribute dimensionality reduction unit configured to reduce the dimensionality of the plurality of seismic attributes by using a locally preserving projection algorithm to obtain optimized feature attributes; a first low-frequency impedance model acquisition unit configured to obtain a low-frequency impedance model M1 under a depth model framework by using the optimized feature attributes and the low-frequency impedance curve; a second low-frequency impedance model acquisition unit configured to obtain a low-frequency impedance model M2 by using the horizon data and the low-frequency impedance curve and performing well interpolation extrapolation by using inverse proportional weighting; a final low-frequency impedance model acquisition unit configured to determine a weight coefficient of the low-frequency impedance model M1 and the low-frequency impedance model M2 by using a whale optimization algorithm, perform linear combination, and obtain a final low-frequency impedance model M_final.

8. The apparatus of claim 7, wherein, The seismic attribute dimensionality reduction unit is specifically configured to: The plurality of seismic attributes are reduced in dimensionality based on the following equation to obtain optimized feature attributes , represent data serial number: wherein, is the data set of the plurality of seismic attributes, B is a projection matrix, the number of the seismic attributes before dimensionality reduction is n , and the number of the feature attributes obtained after dimensionality reduction is m .

9. The apparatus of claim 7, wherein, The optimized feature attributes comprise well trace feature attributes, and the first low-frequency impedance model acquisition unit is specifically configured to: extract the well trace feature attributes from the optimized feature attributes, and combine the low-frequency impedance curves to make a sample set: learn a nonlinear mapping relationship between the well trace feature attributes and the low-frequency impedance curves based on a long short-term memory network: extend the nonlinear mapping relationship to the entire target area to obtain low-frequency impedance attributes of all positions in the target area, to generate the low-frequency impedance model M1.

10. An electronic device, comprising: The electronic device comprises: a memory storing executable instructions: a processor running the executable instructions in the memory to implement the method of any one of claims 1-7.

Images