AVO synchronous inversion method based on coupling simulated annealing algorithm

The AVO synchronous inversion method using coupled simulated annealing algorithm solves the problems of insufficient resolution and large computational load in pre-stack reservoir inversion, realizes high-resolution elastic parameter inversion, and improves the stability and efficiency of inversion.

CN119024431APending Publication Date: 2024-11-26CHINA PETROLEUM & CHEMICAL CORP +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202310592915.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2023-05-24
Publication Date
2024-11-26

AI Technical Summary

Technical Problem

Existing technologies for pre-stack reservoir inversion suffer from problems such as complex sandstone-thin reservoir structures, difficulty in combining well logging and seismic data, insufficient inversion resolution, and huge computational load, making it difficult to achieve fine characterization and quantitative description of a single thin layer in the reservoir.

Method used

The AVO synchronous inversion method based on coupled simulated annealing algorithm is adopted. An initial elastic parameter model is constructed by seismic tectonic interpretation data and well logging data. The initial model and constraint corridor for stochastic inversion are generated by combining the stacked gathers of near, middle and far incident angles. The coupled simulated annealing algorithm is used to perform nonlinear stochastic inversion to obtain high-resolution elastic parameter data.

Benefits of technology

It improves the vertical resolution and efficiency of inversion, enhances the stability of inversion and the lateral continuity of seismic data, and improves the robustness and search efficiency of traditional simulated annealing algorithms.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119024431A_ABST
    Figure CN119024431A_ABST
Patent Text Reader

Abstract

The invention discloses an AVO synchronous inversion method and device based on a coupling simulated annealing algorithm, electronic equipment and a medium. The method comprises the following steps: constructing an initial elastic parameter model through seismic structure interpretation data and logging data; generating an initial model and a constraint corridor of random inversion through the superposition gather of the near, middle and far incident angles and the initial elastic parameter model; introducing a model high-frequency component constraint term according to the initial model and the constraint corridor, performing L2 norm error measurement on the pre-stack seismic gather and the synthetic gather, and constructing an objective function of simulated annealing; and performing nonlinear random inversion on the target function through a coupling simulated annealing algorithm to obtain a high-resolution elastic parameter inversion body. Compared with conventional model inversion, the method has the advantages that the longitudinal resolution is higher, the convergence speed is higher, the inversion efficiency is higher, and the parallel operation capability is good.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of pre-stack AVO synchronous inversion in seismic data interpretation, and more specifically, to an AVO synchronous inversion method, device, electronic equipment and medium based on a coupled simulated annealing algorithm. Background Art

[0002] Prestack reservoir elastic parameter inversion is an important technology in the field of oil and gas exploration. It takes geophysics and rock physics as the theoretical background. According to the anti-color reflection and transmission theory of seismic waves, namely the Zeroppritz equation, the AVO (Amplitube versus Offset) or AVA (Amplitube Versus incidientAngle) law of seismic wave reflection amplitude changing with offset is used to analyze the changing characteristics and laws of prestack seismic data with offset distance, and combines seismic wavelets to invert elastic parameters such as P-wave velocity, S-wave velocity and density of the reservoir, and then establishes the fluid properties and rock physics relationship of the reservoir through rock physics intersection analysis, and then predicts the lithology and physical properties. The above steps are one of the core methods for reservoir and evaluation of concealed oil and gas reservoirs at this stage. With the development of oil and gas, the research focus of seismic exploration has shifted from exploration to development, and the accuracy requirements for seismic inversion are getting higher and higher. However, from the current research, although seismic inversion can obtain higher resolution imaging than the original seismic traces, the seismic data is band-limited, which will cause the inversion results to lack low-frequency and high-frequency components. Therefore, its resolution cannot meet the requirements of reservoir engineers, and it is often difficult to meet the actual needs. Although the application of regularization methods such as sparse constraints or total variation constraints has partially solved and improved the problem of insufficient resolution to a certain extent, it still cannot reach the bandwidth problem of logging data level. Random inversion integrates the prior model of reservoir logging, and this prior model provides low-frequency and high-frequency information of the reservoir. However, since the high-frequency changes in elastic properties do not affect the seismic amplitude, the estimation of the high-frequency part of the model parameters is highly non-unique.

[0003] In summary, the main problems in the current research on the AVO-based pre-stack synchronous inversion method for identifying thin layers are: 1. The structure of the inter-thin sandstone reservoir is complex, and the thickness of the mudstone and sandstone thin layers is less than the seismic resolution limit, which cannot achieve the fine characterization and quantitative description and prediction of a single thin layer in the reservoir; 2. There are measurement differences between logging data and seismic data. Seismic data has good lateral resolution and continuity but low vertical resolution, while logging data has high vertical resolution but can only reflect a smaller range around the well and cannot reflect changes in the entire area. How to combine the two types of data and construct a reasonable prior model has become the key to high-resolution inversion of reservoirs; 3. The computational complexity of random inversion is huge. How to improve the efficiency of random inversion is also an urgent problem to be solved.

[0004] Therefore, developing an AVO synchronous inversion method, device, electronic equipment and medium based on a coupled simulated annealing algorithm has high application value.

[0005] The information disclosed in the background technology section of the present invention is only intended to deepen the understanding of the general background technology of the present invention, and should not be regarded as acknowledging or suggesting in any form that the information constitutes the prior art already known to those skilled in the art. Summary of the invention

[0006] The present invention proposes an AVO synchronous inversion method, device, electronic device and medium based on a coupled simulated annealing algorithm, which can perform preliminary inversion to form a constraint corridor required for simulated annealing inversion through AVO Bayesian linear inversion and multivariate Gaussian hypothesis of reservoir parameter model, use multivariate Gaussian random model for random sampling to form an initial inversion model, and then use coupled simulated annealing algorithm for iteration to invert and obtain high-precision reservoir elastic parameter data. The elastic parameter inversion result of the present invention has higher vertical resolution than conventional model inversion. Due to the use of CSA simulated annealing algorithm, it has faster convergence speed, higher inversion efficiency, and good parallel computing capability.

[0007] In a first aspect, the disclosed embodiment of the present invention provides an AVO synchronous inversion method based on a coupled simulated annealing algorithm, comprising:

[0008] Construct an initial elastic parameter model through seismic structural interpretation data and well logging data;

[0009] Generate an initial model and a constraint corridor for random inversion by stacking gathers of near, medium and far incident angles and the initial elastic parameter model;

[0010] A high-frequency component constraint term of the model is introduced for the initial model and the constraint corridor, the L2 norm error of the prestack seismic gather and the synthetic gather is measured, and an objective function of simulated annealing is constructed;

[0011] A nonlinear random inversion is performed on the objective function by using a coupled simulated annealing algorithm to obtain a high-resolution elastic parameter inversion volume.

[0012] In a second aspect, the present disclosure also provides an AVO synchronous inversion device based on a coupled simulated annealing algorithm, comprising:

[0013] The initial elastic parameter model building module builds the initial elastic parameter model through seismic structural interpretation data and logging data;

[0014] The initial model and constraint corridor construction module generates the initial model and constraint corridor of random inversion through the superposition gathers of near, medium and far incident angles and the initial elastic parameter model;

[0015] An objective function construction module introduces a model high-frequency component constraint term for the initial model and the constraint corridor, measures the L2 norm error between the prestack seismic gather and the synthetic gather, and constructs an objective function of simulated annealing;

[0016] The nonlinear random inversion module performs nonlinear random inversion on the objective function by coupling simulated annealing algorithm to obtain a high-resolution elastic parameter inversion volume.

[0017] In a third aspect, an embodiment of the present disclosure further provides an electronic device, the electronic device comprising:

[0018] A memory storing executable instructions;

[0019] A processor runs the executable instructions in the memory to implement the AVO synchronous inversion method based on the coupled simulated annealing algorithm.

[0020] In a fourth aspect, an embodiment of the present disclosure further provides a computer-readable storage medium, which stores a computer program, and when the computer program is executed by a processor, the AVO synchronous inversion method based on the coupled simulated annealing algorithm is implemented.

[0021] Its beneficial effects are:

[0022] 1. Using the inversion results of the Bayesian linear inversion algorithm as the constraint term of the random inversion effectively reduces the randomness of the model's random disturbance, improves the convergence efficiency of the random inversion algorithm, and effectively maintains the lateral continuity of the seismic data;

[0023] 2. The use of coupled simulated annealing algorithm greatly improves the optimization search efficiency and performance, avoids the algorithm being limited to the local optimal solution too early, and effectively improves the shortcomings of the traditional simulated annealing algorithm in terms of poor robustness to initial parameter selection, thereby enhancing the stability of the inversion.

[0024] The methods and apparatus of the present invention have other features and advantages that will be apparent from, or will be described in detail in, the accompanying drawings and subsequent detailed descriptions incorporated herein, which together serve to explain the specific principles of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] The above and other objects, features and advantages of the present invention will become more apparent through a more detailed description of exemplary embodiments of the present invention in conjunction with the accompanying drawings, wherein like reference numerals generally represent like components throughout the exemplary embodiments of the present invention.

[0026] Figure 1A flowchart of AVO synchronous three-parameter inversion based on CSA simulated annealing according to an embodiment of the present invention is shown.

[0027] Figure 2 A schematic diagram of a well-connected section of an elastic parameter model using Kriging interpolation according to an embodiment of the present invention is shown.

[0028] Figure 3a , Figure 3b , Figure 3c Schematic diagrams of superimposed cross-sections of near (θ=6°), medium (θ=20°), and far (θ=31°) incident angles according to an embodiment of the present invention are respectively shown.

[0029] Figure 4 A schematic diagram of multi-parameter multi-process simulated annealing according to an embodiment of the present invention is shown.

[0030] Figure 5 A schematic diagram of a CSA-M algorithm flow according to an embodiment of the present invention is shown.

[0031] Figure 6 A flowchart showing the steps of an AVO synchronous inversion method based on a coupled simulated annealing algorithm according to an embodiment of the present invention is shown.

[0032] Figure 7 A schematic diagram showing a deterministic inversion result according to an embodiment of the present invention is shown.

[0033] Figure 8 A schematic diagram showing a CSA inversion result according to an embodiment of the present invention is shown.

[0034] Figure 9a , Figure 9b Schematic diagrams showing comparisons between deterministic inversion results and CSA random inversion results of the HB107 well according to an embodiment of the present invention are shown respectively.

[0035] Fig.10 A block diagram of an AVO synchronous inversion device based on a coupled simulated annealing algorithm according to an embodiment of the present invention is shown.

[0036] Description of reference numerals:

[0037] 201. Initial elastic parameter model building module; 202. Initial model and constraint corridor building module; 203. Objective function building module; 204. Nonlinear random inversion module. DETAILED DESCRIPTION

[0038] The preferred embodiments of the present invention will be described in more detail below. Although the preferred embodiments of the present invention are described below, it should be understood that the present invention can be implemented in various forms and should not be limited to the embodiments set forth herein.

[0039] The present invention provides an AVO synchronous inversion method based on a coupled simulated annealing algorithm, comprising:

[0040] Construct an initial elastic parameter model through seismic structural interpretation data and well logging data;

[0041] Generate the initial model and constraint corridor of random inversion by stacking gathers of near, medium and far incident angles and initial elastic parameter model;

[0042] The high-frequency component constraint term of the model is introduced for the initial model and the constraint corridor, the L2 norm error of the pre-stack seismic gather and the synthetic gather is measured, and the objective function of simulated annealing is constructed;

[0043] The nonlinear random inversion is carried out according to the objective function by coupling simulated annealing algorithm to obtain the high-resolution elastic parameter inversion volume.

[0044] In one example, the initial model and constraint corridors for random inversion are generated by stacking gathers at near, medium, and far incidence angles and the initial elastic parameter model, including:

[0045] Establish a linear forward model for AVO;

[0046] Generate forward synthetic records of pre-stack and post-stack seismic data, and then generate the initial model for stochastic inversion;

[0047] The posterior probability expression is established by observing the linear relationship between the pre-stack seismic gather vector and the elastic parameters to be inverted.

[0048] Calculate the expectation and covariance matrix of the posterior distribution, and then obtain the MAP solution of the reservoir elastic parameters;

[0049] The statistical confidence interval is used to obtain an analytical solution, and this interval is used as the nonlinear inversion constraint corridor.

[0050] In one example, the linear forward model for AVO is:

[0051]

[0052] Where, ΔV p =V p2 -V p1 , ΔV s =V s2 -V s1 ,

[0053] In one example, the posterior probability expression is:

[0054]

[0055] Among them, C m represents the prior covariance matrix of the elastic parameters, m p represents the prior expectation of the elasticity parameter, C D represents the covariance matrix of the observed data likelihood function, G is the prestack AVO forward modeling matrix, d obs is the pre-stack seismic gather vector of the observed data.

[0056] In an example, the expectation and covariance matrices of the posterior distribution are:

[0057] μ m|d =μ m +∑ m G T (G∑ m G T +∑ e ) -1 (d obs -Gμ m )

[0058] ∑ m|d =∑ m -∑ m G T (G∑ m G T +∑ e ) -1 GΣ m

[0059] Among them, μ m|d is the expectation of the posterior distribution, ∑ m|d is the covariance matrix.

[0060] In one example, performing nonlinear stochastic inversion via a coupled simulated annealing algorithm includes:

[0061] Step 1: Define the transition probability;

[0062] Step 2: Set N optimization processes, loop variable k = 0, and solve Perform random initialization and calculate the function value Set the initial temperature in Related to de-perturbation and transition probability;

[0063] Step 3: Start the inner loop, generate the perturbation to obtain the trial solution y, and calculate the energy value of the objective function {E(y1), E(y2), ..., E(y k ), ..., E(yN )};

[0064] Step 4: Accept the trial solution by using the Merropolis criterion. If E(y k )<E(m k ), then accept the trial solution y k , otherwise accept the trial solution y with the transition probability shown k ;

[0065] Step 5: Calculate the variance σ of the transition probability described in step 4 2 ,like Increase Otherwise reduce

[0066] Step 6: If the inner loop termination condition is reached, stop execution and update the temperature

[0067] Step 7: If the algorithm termination condition is reached, stop executing and output the optimization result as

[0068]

[0069] In one example, the transition probability is defined as:

[0070]

[0071] in,

[0072] Figure 1 A flowchart of AVO synchronous three-parameter inversion based on CSA simulated annealing according to an embodiment of the present invention is shown.

[0073] Specifically, Figure 1 As shown in the figure, using the seismic structural interpretation of the stratigraphic frame and sequence stratigraphic framework and logging data, the Kriging interpolation algorithm or conditional simulation algorithm is used to construct an initial elastic parameter layered model based on the sedimentary pattern, including a P-wave velocity model, a S-wave velocity model and a density model, and a low-frequency filtering model is performed on the initial elastic parameter model as the initial input model for deterministic inversion. The well-connected section of the elastic parameter model of Kriging interpolation is shown in the figure. Figure 2 shown.

[0074] Figure 3a , Figure 3b , Figure 3c Schematic diagrams of superimposed cross-sections of near (θ=6°), medium (θ=20°), and far (θ=31°) incident angles according to an embodiment of the present invention are respectively shown.

[0075] Through Figure 3a , Figure 3b , Figure 3c The stacked gathers and initial elastic parameter models of near, medium and far incident angles shown in the figure are inverted by linear Bayesian prestack synchronous three-parameter inversion based on the multivariate Gaussian distribution assumption to obtain the initial model of random inversion. In this embodiment, the reservoir elastic parameter model inverted is P-wave velocity, S-wave velocity and density. The specific method principle is as follows:

[0076] Step 1: First, establish a linear forward model for AVO: There are many linear approximate AVO & AVA formulas based on the Zeoppritz equation under different assumptions. Here, the most classic Aki-Richard's AVO approximate formula is introduced as an example (Aki and Richad, 1980). Aki-Richard's PP wave AVA approximate formula is expressed as P wave velocity reflectivity S wave velocity reflectivity and density reflectivity The linear combination of is as follows:

[0077]

[0078] Where, ΔV p =V p2 -V p1 , ΔV s =V s2 -V s1 ,

[0079] In 1985, Slot and Weglein extended the AVA formula of the PP wave reflection coefficient of the two-layer model to a continuous layered medium, and the reflection coefficient expression can be written as:

[0080]

[0081] The coefficient before the elastic parameter term c p (θ), c s (θ), c ρ (θ) is a function that is only related to the incident angle θ:

[0082]

[0083] In inversion, the comparative analysis of forward synthetic records and actual observation data is an important basic work for reservoir prediction inversion analysis and reservoir dynamic and static simulation analysis. Therefore, the generation of forward synthetic records of pre-stack and post-stack seismic data is introduced below. In reflection wave exploration, a one-dimensional convolution model is generally used for forward modeling:

[0084] d(t,θ)=w(t,θ)*r pp (t,θ)=∫w(u,θ)rpp (tu,θ)du (4)

[0085] Where w(t, θ) represents the seismic wavelet function corresponding to different incident angles θ; r pp (t, θ) represents the reflection coefficient function corresponding to different incident angles θ, which can be calculated by the reflection coefficient expression.

[0086] Since seismic data are finite discrete data in practical applications, the continuous convolution forward modeling formula should be discretized and written as a linear equation system for subsequent inversion. First, the continuous function of the reservoir elastic parameters is discretized and written as a finite length column vector m:

[0087] m=[ln(V p ), ln(V s ), ln(ρ)] T (5)

[0088] Among them, V p =[V p1 , V p2 , ..., V pn ] T , V s =[V s1 , V s2 , ..., V sn ] T ,ρ=[ρ1,ρ2,...,ρ n ] T , the number of sampling points of the reservoir discrete model is n, and the dimension of the reservoir elastic parameter model vector m is 3n. Therefore, combining equation (4) and equation (2), the linear equation expression of AVO prestack forward modeling can be obtained:

[0089] d=WADm+ε (6)

[0090] Where d is the pre-stack angle gather data. Assuming that there are n pre-stack angle gathers θ The seismic traces of incident angles (reflection angles) are arranged side by side in a vector, which can be recorded as a one-dimensional column vector (dimension is n-1):

[0091]

[0092] In formula (6), W is the pre-stack wavelet matrix, which is a block diagonal matrix with dimension n θ (n-1)×n θ (n-1), which is of the following form:

[0093]

[0094] Where W i (θi ) is a (n-1)×(n-1) band wavelet matrix, which is in the following form:

[0095]

[0096] Among them, w i The incident angle is θ i The corresponding length is k discrete wavelet sequence, i=1…k.

[0097] In formula (6), A is the AVO coefficient matrix, whose dimension is (n-1)n θ ×3(n-1), where the coefficient c of the AVO formula related to the PP wave incident angle θ p (θ),c s (θ),c ρ (θ), which is expressed as follows:

[0098]

[0099] Among them, A p (θ i ),A s (θ i ),A ρ (θ i ) are diagonal matrices of the coefficients of longitudinal wave velocity reflectivity, shear wave velocity reflectivity, and density reflectivity, respectively. Their dimensions are all (n-1)×(n-1) main diagonal matrices, which can be written as:

[0100]

[0101] In formula (6), D is a first-order differential matrix with a dimension of 3n×3(n-1), which can be expressed as follows:

[0102]

[0103] In order to facilitate the description of the subsequent inversion process, the forward operator matrix G = WAD, and equation (6) can be simplified to d = Gm.

[0104] In order to analyze the uncertainty of the inversion results and provide a constraint corridor for the subsequent random inversion, the Bayesian inversion method is used to use the prestack seismic gather vector d obs The linear relationship between the elastic parameter m to be inverted is that the prior model of the elastic parameter satisfies the multivariate Gaussian hypothesis in the inversion. According to the Bayesian theory, the posterior probability expression is established as follows:

[0105]

[0106] In formula (13), C mrepresents the prior covariance matrix of the elastic parameters, m p It represents the prior expectation of the elastic parameters. The initial elastic parameter model can be used as the prior model. D represents the covariance matrix of the observed data likelihood function, G is the prestack AVO forward modeling matrix, d obs is the pre-stack seismic gather vector of the observed data. For the convenience of subsequent calculation expression, let μ m =m p .

[0107] According to the construction method of the prior function, assuming that ∑ m The correlation between the parameters in the model parameter vector m is reflected. Taking the pre-stack model data as an example, its covariance matrix should reflect the correlation between the P-wave velocity, S-wave velocity, and density, as well as the correlation of the data in space, giving a ∑ m A simple construction form:

[0108]

[0109] where Σ0 represents the covariance matrix of the correlations between elastic parameters, where is the Kronecker product operator, ∑ t is the time-dependent matrix. According to the conditional theorem of multivariate Gaussian joint distribution, the expectation μ of the posterior distribution is obtained. m|d and the covariance matrix ∑ m|d :

[0110] μ m|d =μ m +Σ m G T (GΣ m G T +Σ e ) -1 (d obs -Gμ m ) (15)

[0111] Σ m|d =Σ m -Σ m G T (GΣ m G T +Σ e ) -1 GΣ m (16)

[0112] Since m is the natural logarithm of the reservoir elastic parameters, it can be seen that the reservoir elastic parameters satisfy the Log-Gaussian distribution, and the MAP solution of the inverted reservoir elastic parameters can be written as follows:

[0113]

[0114] The confidence interval of this statistic can also be solved analytically; for example, given a 95% confidence interval I 0.95 :

[0115] I 0.95 =[exp(μ m∣d -1.96σ m∣d ),exp(μ m∣d +1.96σ m∣d )] (18)

[0116] σ in formula (17) and (18) m|d is the covariance matrix of the posterior function ∑ m|d The square root of the main diagonal element vector of . This confidence interval serves as the constraint corridor for the nonlinear inversion.

[0117] In the deterministic inversion, the reservoir elastic parameters (P-wave velocity, S-wave velocity, density) with seismic resolution are obtained as the initial model and constraint corridor, and the high-frequency component constraint term α||L(m)-L(m p )||2, where α is the high-frequency model term control parameter, L is the high-pass filter of the elastic parameter, m is the elastic parameter to be inverted, and m p The elastic parameter model established for geostatistics uses the Coupled Simulated Anneling (CSA) algorithm for high-frequency perturbation to further improve the resolution of the inversion results.

[0118] Summarizing the characteristics of the classic simulated annealing algorithm, it can be found that the algorithm's solution is closely related to the temperature parameter. If the temperature is too high, the jump of the particle state will become violent and the convergence of the algorithm will slow down; if the temperature is too low, the particles will be easily trapped in the suboptimal area. In actual work, if the optimization algorithm is run multiple times under different parameter configurations, and then the optimal solution under the optimal parameter configuration is found, it is a formally feasible but essentially inefficient approach. In addition, the random characteristics of the simulated annealing algorithm determine that even if the algorithm parameter configuration is fixed, the optimization result is not uniquely determined, so it is still necessary to execute multiple programs to obtain the optimal solution. The core source of the above problems is obviously determined by the structure of the classic simulated annealing algorithm itself, which makes it necessary to improve the classic simulated annealing algorithm.

[0119] Figure 4A schematic diagram of multi-parameter multi-process simulated annealing according to an embodiment of the present invention is shown, wherein k represents a process index, i represents the number of iterations, and N represents the total number of processes.

[0120] In the simulated annealing algorithm, it is difficult to obtain satisfactory results in a single optimization process. The problem is how to robustly obtain satisfactory results through multiple optimization processes. Samuel (2010) started with the transition probability of the annealing algorithm and transformed its structure, so that multiple originally independent simulated annealing optimization processes can be naturally coupled together. The schematic diagram of multi-parameter multi-process simulated annealing is shown in the figure. Figure 4 As shown in the figure, this type of algorithm is collectively referred to as coupled simulated annealing algorithm (Coupled Simulated Annealing, CSA).

[0121] The core difference between the CSA algorithm and the classic SA algorithm lies in the transition probability. The transition probability is no longer only related to the current solution in this process. It is not related to the current solution set of multiple processes. The breakthrough of the CSA algorithm is to couple the information of multiple optimization processes together. Its optimization process has excellent global search capabilities and can eventually make the algorithm converge to a low energy value.

[0122] Figure 5 A schematic diagram of a CSA-M algorithm flow according to an embodiment of the present invention is shown.

[0123] like Figure 5 As shown, the CSA-M algorithm process:

[0124] Step 1: Define the transition probability as:

[0125]

[0126] in

[0127] Step 2: Set N optimization processes, loop variable k = 0, and solve Perform random initialization and calculate the function value Set the initial temperature in It is related to the de-disturbance and transition probability; the target energy function in this inversion is defined as follows:

[0128] E(m)=||d obs -Gm||2+α||L(m)-L(m p )||2 (20)

[0129] Where G is the AVO forward modeling matrix, d obsis the observed data, α is the high-frequency model term control parameter, L is the high-pass filter of the elasticity parameter, m p Elastic parameter models for geostatistics.

[0130] Step 3: Start the inner loop, generate the perturbation to obtain the trial solution y, and calculate the energy value of the objective function {E(y1), E(y2), ...E(y k ), ..., E(y N )}, which is calculated as follows:

[0131] u k ∈U[0,1]

[0132]

[0133]

[0134]

[0135] Step 4: Use the Merropolis criterion to determine the accepted trial solution. If E(y k )<E(m k ), then accept the trial solution y k , otherwise accept the trial solution y with the transition probability k The Metropolis determination criteria are as follows:

[0136] Calculate the transition probability of Metropolis:

[0137]

[0138]

[0139] If ΔE≤0, accept the trial solution:

[0140]

[0141]

[0142] If ΔE>0, take a random number r=U[0,1], when P>r, update the model to accept the trial solution:

[0143]

[0144]

[0145] Step 5: Calculate the variance σ of the transition probability in step 3 2 ,like Increase Otherwise reduce

[0146] Step 6: If the inner loop termination condition is reached, stop execution and update the temperature

[0147]

[0148] Otherwise, enter the loop and go to step 3.

[0149] Step 7: If the algorithm termination condition is reached, stop executing and output the optimization result as Otherwise, proceed to step 3.

[0150] The present invention also provides an AVO synchronous inversion device based on a coupled simulated annealing algorithm, comprising:

[0151] The initial elastic parameter model building module builds the initial elastic parameter model through seismic structural interpretation data and logging data;

[0152] The initial model and constraint corridor construction module generates the initial model and constraint corridor of random inversion through the superposition of gathers at near, medium and far incidence angles and the initial elastic parameter model;

[0153] The objective function construction module introduces the high-frequency component constraint term of the model for the initial model and the constraint corridor, measures the L2 norm error of the pre-stack seismic gather and the synthetic gather, and constructs the objective function of simulated annealing;

[0154] The nonlinear random inversion module performs nonlinear random inversion on the objective function through the coupled simulated annealing algorithm to obtain a high-resolution elastic parameter inversion volume.

[0155] In one example, the initial model and constraint corridors for random inversion are generated by stacking gathers at near, medium, and far incidence angles and the initial elastic parameter model, including:

[0156] Establish a linear forward model for AVO;

[0157] Generate forward synthetic records of pre-stack and post-stack seismic data, and then generate the initial model for stochastic inversion;

[0158] The posterior probability expression is established by observing the linear relationship between the pre-stack seismic gather vector and the elastic parameters to be inverted.

[0159] Calculate the expectation and covariance matrix of the posterior distribution, and then obtain the MAP solution of the reservoir elastic parameters;

[0160] The statistical confidence interval is used to obtain an analytical solution, and this interval is used as the nonlinear inversion constraint corridor.

[0161] In one example, the linear forward model for AVO is:

[0162]

[0163] Where, ΔV p =V p2 -V p1 , ΔV s =V s2 -V s1 ,

[0164] In one example, the posterior probability expression is:

[0165]

[0166] Among them, C m represents the prior covariance matrix of the elastic parameters, m p represents the prior expectation of the elasticity parameter, C D represents the covariance matrix of the observed data likelihood function, G is the prestack AVO forward modeling matrix, d obs is the pre-stack seismic gather vector of the observed data.

[0167] In an example, the expectation and covariance matrices of the posterior distribution are:

[0168] μ m|d =μ m +∑ m G T (G∑ m G T +∑ e ) -1 (d obs -Gμ m )

[0169] ∑ m|d =∑ m -∑ m G T (G∑ m G T +∑ e ) -1 GΣ m

[0170] Among them, μ m|d is the expectation of the posterior distribution, ∑ m|d is the covariance matrix.

[0171] In one example, performing nonlinear stochastic inversion via a coupled simulated annealing algorithm includes:

[0172] Step 1: Define the transition probability;

[0173] Step 2: Set N optimization processes, loop variable k = 0, and solve Perform random initialization and calculate the function value Set the initial temperature in Related to de-perturbation and transition probability;

[0174] Step 3: Start the inner loop, generate the perturbation to obtain the trial solution y, and calculate the energy value of the objective function {E(y1), E(y2), …E(y k ),…,E(y N )};

[0175] Step 4: Accept the trial solution by using the Merropolis criterion. If E(y k )<E(m k ), then accept the trial solution y k , otherwise accept the trial solution y with the transition probability shown k ;

[0176] Step 5: Calculate the variance σ of the transition probability in step 4 2 ,like Increase Otherwise reduce

[0177] Step 6: If the inner loop termination condition is reached, stop execution and update the temperature

[0178] Step 7: If the algorithm termination condition is reached, stop executing and output the optimization result as

[0179]

[0180] In one example, the transition probability is defined as:

[0181]

[0182] in,

[0183] The present invention also provides an electronic device, which includes: a memory storing executable instructions; and a processor, which runs the executable instructions in the memory to implement the above-mentioned AVO synchronous inversion method based on the coupled simulated annealing algorithm.

[0184] The present invention also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the AVO synchronous inversion method based on the coupled simulated annealing algorithm is implemented.

[0185] To facilitate understanding of the solutions and effects of the embodiments of the present invention, four specific application examples are given below. Those skilled in the art should understand that the examples are only for facilitating understanding of the present invention, and any specific details thereof are not intended to limit the present invention in any way.

[0186] Example 1

[0187] Figure 5 A flowchart showing the steps of an AVO synchronous inversion method based on a coupled simulated annealing algorithm according to an embodiment of the present invention is shown.

[0188] like Figure 5 As shown, the AVO synchronous inversion method based on the coupled simulated annealing algorithm includes: step 101, constructing an initial elastic parameter model through seismic structural interpretation data and logging data; step 102, generating an initial model and constraint corridor for random inversion through stacked gathers of near, medium and far incident angles and the initial elastic parameter model; step 103, introducing a model high-frequency component constraint term for the initial model and the constraint corridor, measuring the L2 norm error of the pre-stack seismic gathers and the synthetic gathers, and constructing a simulated annealing objective function; step 104, performing nonlinear random inversion on the objective function through the coupled simulated annealing algorithm to obtain a high-resolution elastic parameter inversion volume.

[0189] Figure 6 A schematic diagram showing a deterministic inversion result according to an embodiment of the present invention is shown.

[0190] Figure 7 A schematic diagram showing a CSA inversion result according to an embodiment of the present invention is shown.

[0191] like Figure 6 , Figure 7 As shown, it can be seen that among the deterministic inversion results and the CSA inversion results, the CSA random inversion results have good lateral continuity.

[0192] Figure 9a , Figure 9b A schematic diagram comparing the deterministic inversion result of the HB107 well and the CSA random inversion result according to an embodiment of the present invention is shown, wherein the red line is the inversion result and the black line is the elastic parameter of the HB107 well.

[0193] like Figure 9a , Figure 9b As shown in the figure, the well side channel inversion results of the extraction test well HB107 show that the CSA inversion results are more consistent with the logging curve than the deterministic inversion results, and the vertical resolution is higher.

[0194] Example 2

[0195] FIG. 9 shows a block diagram of an AVO synchronous inversion device based on a coupled simulated annealing algorithm according to an embodiment of the present invention.

[0196] As shown in FIG9 , the AVO synchronous inversion device based on the coupled simulated annealing algorithm includes:

[0197] An initial elastic parameter model building module 201 builds an initial elastic parameter model through seismic structure interpretation data and well logging data;

[0198] The initial model and constraint corridor construction module 202 generates the initial model and constraint corridor of random inversion through the superposition gathers of near, medium and far incident angles and the initial elastic parameter model;

[0199] The objective function construction module 203 introduces the model high-frequency component constraint term for the initial model and the constraint corridor, measures the L2 norm error between the pre-stack seismic gather and the synthetic gather, and constructs the objective function of simulated annealing;

[0200] The nonlinear random inversion module 204 performs nonlinear random inversion on the target function by means of a coupled simulated annealing algorithm to obtain a high-resolution elastic parameter inversion volume.

[0201] In one example, the initial model and constraint corridors for random inversion are generated by stacking gathers at near, medium, and far incidence angles and the initial elastic parameter model, including:

[0202] Establish a linear forward model for AVO;

[0203] Generate forward synthetic records of pre-stack and post-stack seismic data, and then generate the initial model for stochastic inversion;

[0204] The posterior probability expression is established by observing the linear relationship between the pre-stack seismic gather vector and the elastic parameters to be inverted.

[0205] Calculate the expectation and covariance matrix of the posterior distribution, and then obtain the MAP solution of the reservoir elastic parameters;

[0206] The statistical confidence interval is used to obtain an analytical solution, and this interval is used as the nonlinear inversion constraint corridor.

[0207] In one example, the linear forward model for AVO is:

[0208]

[0209] In one example, the posterior probability expression is:

[0210]

[0211] Among them, C mrepresents the prior covariance matrix of the elastic parameters, m p represents the prior expectation of the elasticity parameter, C D represents the covariance matrix of the observed data likelihood function, G is the prestack AVO forward modeling matrix, d obs is the pre-stack seismic gather vector of the observed data.

[0212] In an example, the expectation and covariance matrices of the posterior distribution are:

[0213] μ m|d =μ m +∑ m G T (G∑ m G T +∑ e ) -1 (d obs -Gμ m )

[0214] ∑ m|d =∑ m -∑ m G T (G∑ m G T +∑ e ) -1 G∑ m

[0215] Among them, μ m|d is the expectation of the posterior distribution, ∑ m|d is the covariance matrix.

[0216] In one example, performing nonlinear stochastic inversion via a coupled simulated annealing algorithm includes:

[0217] Step 1: Define the transition probability;

[0218] Step 2: Set N optimization processes, loop variable k = 0, and solve Perform random initialization and calculate the function value Set the initial temperature in Related to de-perturbation and transition probability;

[0219] Step 3: Start the inner loop, generate the perturbation to obtain the trial solution y, and calculate the energy value of the objective function {E(y1), E(y2), ...E(y k ), ..., E(y N )};

[0220] Step 4: Accept the trial solution by using the Merropolis criterion. If E(y k )<E(m k), then accept the trial solution y k , otherwise accept the trial solution y with the transition probability shown k ;

[0221] Step 5: Calculate the variance σ of the transition probability in step 4 2 ,like Increase Otherwise reduce

[0222] Step 6: If the inner loop termination condition is reached, stop execution and update the temperature

[0223] Step 7: If the algorithm termination condition is reached, stop executing and output the optimization result as

[0224]

[0225] In one example, the transition probability is defined as:

[0226]

[0227] in,

[0228] Example 3

[0229] The present disclosure provides an electronic device, which includes: a memory storing executable instructions; a processor running the executable instructions in the memory to implement the above-mentioned AVO synchronous inversion method based on the coupled simulated annealing algorithm.

[0230] An electronic device according to an embodiment of the present disclosure includes a memory and a processor.

[0231] The memory is used to store non-temporary computer-readable instructions. Specifically, the memory may include one or more computer program products, which may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may, for example, include random access memory (RAM) and / or cache memory (cache), etc. The non-volatile memory may, for example, include read-only memory (ROM), hard disk, flash memory, etc.

[0232] The processor may be a central processing unit (CPU) or other forms of processing units having data processing capabilities and / or instruction execution capabilities, and may control other components in the electronic device to perform desired functions. In one embodiment of the present disclosure, the processor is used to run the computer-readable instructions stored in the memory.

[0233] Those skilled in the art should be able to understand that in order to solve the technical problem of how to obtain a good user experience, the present embodiment may also include well-known structures such as a communication bus and an interface, and these well-known structures should also be included in the protection scope of the present disclosure.

[0234] For detailed description of this embodiment, reference may be made to the corresponding descriptions in the aforementioned embodiments, which will not be repeated here.

[0235] Example 4

[0236] An embodiment of the present disclosure provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, the AVO synchronous inversion method based on the coupled simulated annealing algorithm is implemented.

[0237] According to the computer-readable storage medium of the embodiment of the present disclosure, non-transitory computer-readable instructions are stored thereon. When the non-transitory computer-readable instructions are executed by a processor, all or part of the steps of the above-mentioned methods of each embodiment of the present disclosure are executed.

[0238] The above-mentioned computer-readable storage media include, but are not limited to: optical storage media (e.g., CD-ROM and DVD), magneto-optical storage media (e.g., MO), magnetic storage media (e.g., magnetic tape or mobile hard disk), media with built-in rewritable non-volatile memory (e.g., memory card) and media with built-in ROM (e.g., ROM box).

[0239] Those skilled in the art should understand that the purpose of the above description of the embodiments of the present invention is only to exemplarily illustrate the beneficial effects of the embodiments of the present invention, and is not intended to limit the embodiments of the present invention to any given examples.

[0240] The embodiments of the present invention have been described above, and the above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and changes will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments.

Claims

1. An AVO synchronous inversion method based on coupled simulated annealing algorithm, characterized in that: include: Construct an initial elastic parameter model through seismic structural interpretation data and well logging data; Generate an initial model and a constraint corridor for random inversion by stacking gathers of near, medium and far incident angles and the initial elastic parameter model; A high-frequency component constraint term of the model is introduced for the initial model and the constraint corridor, and an objective function of simulated annealing is constructed by combining the L2 norm error measurement of the prestack seismic gather and the synthetic gather; A nonlinear random inversion is performed on the objective function by using a coupled simulated annealing algorithm to obtain a high-resolution elastic parameter inversion volume.

2. The AVO synchronous inversion method based on coupled simulated annealing algorithm according to claim 1, wherein: By using the stacked gathers of near, medium and far incidence angles and the initial elastic parameter model, the initial model and constraint corridor for random inversion are generated, including: Establish a linear forward model for AVO; Generate forward synthetic records of pre-stack and post-stack seismic data, and then generate the initial model for stochastic inversion; The posterior probability expression is established by observing the linear relationship between the pre-stack seismic gather vector and the elastic parameters to be inverted. Calculate the expectation and covariance matrix of the posterior distribution, and then obtain the MAP solution of the reservoir elastic parameters; An analytical solution is obtained from the confidence interval statistically obtained from the MAP solution, and the confidence interval is used as a nonlinear inversion constraint corridor.

3. The AVO synchronous inversion method based on coupled simulated annealing algorithm according to claim 2, wherein: The linear forward model of AVO is: where, ΔV p = V p2 - V p1 , ΔV s = V s2 - V s1 , 4. The AVO synchronous inversion method based on coupled simulated annealing algorithm according to claim 2, wherein: The posterior probability expression is: Among them, C m represents the prior covariance matrix of the elastic parameters, m p represents the prior expectation of the elasticity parameter, C D represents the covariance matrix of the observed data likelihood function, G is the prestack AVO forward modeling matrix, d obs is the pre-stack seismic gather vector of the observed data.

5. The AVO synchronous inversion method based on coupled simulated annealing algorithm according to claim 2, wherein: The expectation and covariance matrices of the posterior distribution are: m m|d =μ m +S m G T (GΣ m G T +S e ) -1 (d obs -Gμ m ) S m|d =S m -S m G T (GΣ m G T +S e ) -1 GS m Among them, μ m|d is the expectation of the posterior distribution, ∑ m|d is the covariance matrix.

6. The AVO synchronous inversion method based on coupled simulated annealing algorithm according to claim 1, wherein: Nonlinear stochastic inversion via coupled simulated annealing algorithm includes: Step 1: Define the transition probability; Step 2: Set N optimization processes, loop variable k = 0, and solve Perform random initialization and calculate the function value Set the initial temperature in Related to de-perturbation and transition probability; Step 3: Start the inner loop, generate the perturbation to obtain the trial solution y, and calculate the energy value of the objective function {E(y1), E(y2), …, E(y k ),…,E(y N )}; Step 4: Accept the trial solution by using the Merropolis criterion. If E(y k )<E(m k ), then accept the trial solution y k , otherwise accept the trial solution y with the transition probability shown k ; Step 5: Calculate the variance σ of the transition probability described in step 4 2 ,like Increase Otherwise reduce Step 6: If the inner loop termination condition is reached, stop execution and update the temperature Step 7: If the algorithm termination condition is reached, stop executing and output the optimization result as 7. The AVO synchronous inversion method based on coupled simulated annealing algorithm according to claim 6, wherein: The transition probability is defined as: in, 8. An AVO synchronous inversion device based on coupled simulated annealing algorithm, characterized in that: include: The initial elastic parameter model building module builds the initial elastic parameter model through seismic structural interpretation data and logging data; The initial model and constraint corridor construction module generates the initial model and constraint corridor of random inversion through the superposition gathers of near, medium and far incident angles and the initial elastic parameter model; An objective function construction module introduces a model high-frequency component constraint term for the initial model and the constraint corridor, measures the L2 norm error between the prestack seismic gather and the synthetic gather, and constructs an objective function of simulated annealing; The nonlinear random inversion module performs nonlinear random inversion on the objective function by coupling simulated annealing algorithm to obtain a high-resolution elastic parameter inversion volume.

9. An electronic device, characterized in that: The electronic device comprises: A memory for storing executable instructions; A processor, wherein the processor runs the executable instructions in the memory to implement the AVO synchronous inversion method based on the coupled simulated annealing algorithm according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the AVO synchronous inversion method based on a coupled simulated annealing algorithm described in any one of claims 1 to 7 is implemented.