Joint inversion method and device for electromagnetic wave logging data while drilling, and storage medium

Abstract

The application provides a method and device for joint inversion of electromagnetic wave logging data while drilling and a storage medium, comprising extracting a reservoir horizontal resistivity sequence from adjacent well data; extracting far-probing electromagnetic wave logging data while drilling; judging whether there is electromagnetic wave logging data while drilling and the category of the electromagnetic wave logging data while drilling; inverting current horizon information according to the category of the electromagnetic wave logging data while drilling by using a corresponding method; establishing an inversion model according to the far-probing electromagnetic wave logging data while drilling, and applying geological constraints to to-be-inverted parameters according to the extracted reservoir resistivity sequence; when there is electromagnetic wave logging data while drilling or conventional electromagnetic wave logging data while drilling, reducing the number of to-be-inverted parameters of the inversion model according to the extracted current horizon information; sampling and inverting the inversion model, and outputting an inversion result. The application can effectively utilize the edge-probing capability difference and complementary information of electromagnetic logging data of different scales, and improves the inversion speed and precision of the far-probing electromagnetic wave logging data while drilling.

Description

Technical Field

[0001] This invention relates to the technical field of petroleum exploration and development, specifically to a method and apparatus for joint inversion of electromagnetic logging data while drilling, and a storage medium. Background Technology

[0002] Long-range electromagnetic logging (LVM) can detect the formation structure distribution within tens of meters of the wellbore, with low dependence on information from nearby pilot wells and seismic data. Its application in deep-sea oil and gas reservoir geological steering has been increasing in recent years. Due to the complexity of the LVM response, the evaluation of this data must be achieved through inversion. The LVM detection range often encompasses multiple formations, requiring the use of multi-interface anisotropic formation models for inversion. Complex multi-layer inversion models not only lead to a sharp increase in inversion parameters but also increase the number of minimum cost functions. Furthermore, the lack of sufficient known information about the parameters to be inverted renders gradient optimization algorithms based on multiple initial values ​​unsuitable.

[0003] In existing technologies, azimuth electromagnetic logging while drilling (AWP) and traditional AWP offer high resolution, fast data inversion speed, strong stability, and high accuracy, making them widely used. However, their detection range is small, and their edge detection capability is limited, especially since traditional AWP lacks azimuth sensitivity. Effectively utilizing the differences in edge detection capability and complementary information between electromagnetic logging data at different scales would be significant for further broadening the geological steering view, improving the imaging accuracy of the wellbore formation structure, and increasing the inversion speed. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to overcome the shortcomings of the prior art and provide a method, device and storage medium for joint inversion of drilling electromagnetic wave logging data. By jointly inverting drilling azimuth electromagnetic wave logging data and traditional drilling electromagnetic wave logging data with remote detection drilling electromagnetic wave logging data, the number of parameters to be inverted in the multi-layer inversion model of remote detection instruments can be reduced, thereby improving the inversion speed and accuracy.

[0005] To solve the above-mentioned technical problems, the technical solution proposed by this invention is as follows:

[0006] A joint inversion method for drilling electromagnetic wave logging data includes the following steps: S01, extracting reservoir horizontal resistivity sequences from adjacent well data as geological constraints; S02, extracting remote sensing drilling electromagnetic wave logging data as basic inversion data; S03, determining whether drilling electromagnetic wave logging data is available and its type; S04, inverting the current stratigraphic information using the corresponding method based on the type of drilling electromagnetic wave logging data; S05, establishing a five-layer macroscopic anisotropy inversion model based on the remote sensing drilling electromagnetic wave logging data, and applying geological constraints to the inversion parameters based on the reservoir resistivity sequences extracted in step S01; S06, when drilling azimuth electromagnetic wave logging data or traditional drilling electromagnetic wave logging data is available, reducing the number of parameters to be inverted in the five-layer macroscopic anisotropy inversion model based on the current stratigraphic information extracted in step S04; S07, sampling and inverting the inversion model using a stochastic Bayesian inversion algorithm, and outputting the inversion results.

[0007] The joint inversion method for drilling electromagnetic wave logging data according to the present invention utilizes drilling electromagnetic wave logging information from multiple depths. Leveraging the advantages of shallow-depth logging data (small detection range, high resolution), it enables the detection of resistivity and upper / lower interfaces within the instrument layer. Utilizing the detection capability of far-depth logging data for multi-layer interfaces, it achieves imaging of the formation structure at distant well depths. Compared to formation structure extraction or imaging methods based on single far-depth drilling electromagnetic wave logging data, this joint inversion method provides richer information and effectively reduces the uncertainties of traditional inversion methods, significantly improving the speed and accuracy of the inversion. This enables rapid inversion and imaging of wellbore interfaces, formation structure, and resistivity distribution.

[0008] The above technical solution can be further improved as described below.

[0009] The method for joint inversion of drilling electromagnetic wave logging data according to the present invention is characterized in that step S03 includes: determining whether there is drilling azimuth electromagnetic wave logging data and conventional drilling electromagnetic wave logging data; if there is drilling azimuth electromagnetic wave logging data, then step S041 is executed; if only conventional drilling electromagnetic wave logging data is available, then step S042 is executed.

[0010] Specifically, in a preferred embodiment, step S04 includes the following sub-steps: S041, based on the azimuth electromagnetic logging data while drilling, establish a three-layer anisotropic formation model and invert the resistivity and interface information of the current layer; S042, based on the traditional electromagnetic logging data while drilling, use the adaptive multiplier regularization Guass-Newton algorithm to invert the resistivity information of the current layer.

[0011] Further, in a preferred embodiment, step S041 includes the following sub-steps: S0411, based on the drilling azimuth electromagnetic wave logging data, establish a three-layer anisotropic formation inversion model, and determine the initial values ​​of each parameter to be inverted according to the formation information given in step S01; S0412, perform adaptive multiplier regularized Gauss-Newton inversion iteration on the model to obtain the inversion results; S0413, according to the detection characteristics of the drilling azimuth electromagnetic wave logging instrument, classify the inversion results into three categories. The first type is where the positions h1 and h2 of the upper and lower interfaces are both greater than the preset height; the second type is where the position h1 or h2 of the upper or lower interface is less than the preset height; the third type is where the positions h1 and h2 of the upper and lower interfaces are both less than the preset height. S0414. For the first type of inversion result, output resistivity as the current layer information; for the second type of inversion result, output resistivity and the position of the upper or lower interface as the current layer information; for the third type of inversion result, output resistivity and the positions of the upper and lower interfaces as the current layer information.

[0012] Specifically, in a preferred embodiment, in steps S0412 and S042, the adaptive multiplier regularization Gauss-Newton inversion iterative algorithm is as follows:

[0013] Based on the gradient inversion method, the cost function C(x) can be expressed as follows in the k-th iteration:

[0014] C k (m)=F(x k )·R k (m) (7)

[0015] In the formula, the first term of the cost function represents the L2 norm of the measured data d and the forward response S(m).

[0016] m is the parameter vector to be inverted, R k (m) represents the adaptive regularization term:

[0017]

[0018] In the formula, δ is a constant value, which can be determined through numerical experiments. p As a reference model vector, m can be set to the model state from the previous iteration during the inversion process. p =m k-1 ;

[0019] The Gauss-Newton optimization method is used to solve (1), and its derivative with respect to x is zero. The k-th iteration... have to

[0020]

[0021] In the formula, J represents the Jacobian matrix and T represents the matrix transpose.

[0022] Specifically, in a preferred embodiment, the preset height is 1.7–1.9 m. Specifically, in a preferred embodiment, step S06 specifically includes the following sub-steps:

[0023] S061. If there is azimuth logging data while drilling, proceed to steps S062-S064. If only conventional azimuth logging data while drilling is available, proceed to step S065. S062. If step S0414 only outputs the current layer resistivity information, the number of parameters to be inverted in the five-layer macroscopic anisotropic inversion model is reduced to 10. S063. If step S0414 outputs the current layer resistivity information and the position of the upper or lower interface, the number of parameters to be inverted in the five-layer macroscopic anisotropic inversion model is reduced to 9. S064. If step S0414 outputs the current layer resistivity information and the positions of the upper and lower interfaces, the number of parameters to be inverted in the five-layer macroscopic anisotropic inversion model is reduced to 8. S064. Based on the current layer resistivity information output in step S042, the number of parameters to be inverted in the five-layer macroscopic anisotropic inversion model is reduced to 10.

[0024] Specifically, in a preferred embodiment, in step S07, the Bayesian inversion algorithm is as follows:

[0025] Based on Bayesian theory, the posterior distribution p(m|d) of the inversion solution obs This can be represented as:

[0026]

[0027] In the formula, m is the stratigraphic model, and d obs For measured data, p(d) obs |m) is the likelihood function, C d The data covariance matrix is ​​used; the Markov chain-Monte Carlo algorithm is employed to sample the likelihood function distribution.

[0028] p(x|d obs )q(x,y)=p(y|d obs )q(y,x) (11)

[0029] In the formula, x and y are arbitrary models, and p(x|d obs Let be the limiting distribution of the Markov chain, and q(x,y) be the transition kernel function. Using the Metropolis-Hsting (MH) algorithm, an acceptor function α(y,x) is introduced, such that:

[0030] p(x|d obs)q(x,y)α(y,x)=p(y|d obs )q(y,x)α(y,x) (12)

[0031] In the formula:

[0032]

[0033] Specifically, in a preferred embodiment, the Markov chain-Monte Carlo algorithm includes: generating the next successor model y based on the current iterative model x and the transition function q(x,y); then calculating the acceptance probability α(x,y) and comparing it with a random number β; if β≥α(x,y), then returning model y, otherwise returning model x; repeating the above steps until the termination condition is met.

[0034] The drilling electromagnetic logging data joint inversion device of the second aspect of the present invention is capable of implementing the method described above.

[0035] The storage medium of the third aspect of the present invention stores a computer program, which, when run by a processor, executes the joint inversion method for electromagnetic logging data as described above.

[0036] Compared with existing technologies, the advantages of this invention are: it can effectively utilize the differences in edge detection capabilities and complementary information of electromagnetic logging data at different scales, thereby improving the inversion speed and accuracy of remote detection electromagnetic logging data. Attached Figure Description

[0037] The invention will now be described in more detail with reference to embodiments and the accompanying drawings.

[0038] Figure 1 The flowchart illustrates the joint inversion method for electromagnetic logging data during drilling according to an embodiment of the present invention.

[0039] Figure 2 The diagram illustrates a three-layer stratigraphic inversion model in an embodiment of the present invention;

[0040] Figure 3 The five-layer stratigraphic inversion model in an embodiment of the present invention is illustrated schematically;

[0041] Figure 4 The diagram schematically illustrates the two-dimensional curtain diagram inverted using the Bayesian algorithm from a five-layer stratigraphic model in an embodiment of the present invention.

[0042] Figure 5 a to Figure 5 d schematically illustrates the joint inversion of remote detection logging-while-drilling electromagnetic wave data and traditional logging-while-drilling electromagnetic wave data in an embodiment of the present invention;

[0043] Figure 6 a to Figure 6 d schematically illustrates the joint inversion of remote detection electromagnetic logging while drilling and azimuth electromagnetic logging data in an embodiment of the present invention.

[0044] In the accompanying drawings, the same parts use the same reference numerals. The drawings are not drawn to scale. Detailed Implementation

[0045] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments, but this does not limit the scope of protection of the present invention.

[0046] Figure 1 The schematic diagram illustrates the process of the joint inversion method for electromagnetic logging data during drilling according to an embodiment of the present invention. Figure 2 The schematic diagram illustrates a three-layer stratigraphic inversion model in an embodiment of the present invention. Figure 3 The diagram illustrates a five-layer stratigraphic inversion model in an embodiment of the present invention. Figure 4 The illustration schematically shows the two-dimensional curtain diagram inverted using the Bayesian algorithm from a five-layer stratigraphic model in an embodiment of the present invention. Figure 5 a to Figure 5 d schematically illustrates the joint inversion of remote detection electromagnetic logging while drilling data and traditional electromagnetic logging while drilling data in an embodiment of the present invention. Figure 6 a to Figure 6 d schematically illustrates the joint inversion of remote detection electromagnetic logging while drilling and azimuth electromagnetic logging data in an embodiment of the present invention.

[0047] Example 1

[0048] like Figure 1 As shown, a multi-scale electromagnetic logging-while-drilling data joint inversion method includes the following steps:

[0049] S01. Extract the reservoir horizontal resistivity sequence from adjacent well data as a geological constraint condition;

[0050] S02. Extract remote detection electromagnetic wave logging data while drilling as basic inversion data;

[0051] S03. Determine whether there is drilling electromagnetic logging data and the type of drilling electromagnetic logging data; specifically, this includes: determining whether there is drilling azimuth electromagnetic logging data and traditional drilling electromagnetic logging data. If there is drilling azimuth electromagnetic logging data, proceed to step S041. If only traditional drilling electromagnetic logging data is available, proceed to step S042. Based on the type of drilling electromagnetic logging data, use the corresponding method to invert the current formation information.

[0052] S04. Based on the type of electromagnetic logging data, use the corresponding method to invert the current stratigraphic information; specifically including the following sub-steps:

[0053] S041. Based on the azimuth electromagnetic logging data while drilling, establish a three-layer anisotropic formation model and invert the resistivity and interface information of the current layer.

[0054] S042. Based on traditional electromagnetic logging data, the adaptive multiplier regularization Guass-Newton algorithm is used to invert the resistivity information of the current formation.

[0055] Specifically, step S041 includes the following sub-steps:

[0056] S0411. Based on the azimuth electromagnetic logging data while drilling, establish a three-layer anisotropic formation inversion model. According to the formation information given in step S01, determine the initial values ​​of each parameter to be inverted.

[0057] S0412. Perform adaptive multiplier regularization Gauss-Newton inversion iteration on the model to obtain the inversion results;

[0058] The adaptive multiplier regularized Gauss-Newton inversion iterative method can be specifically described as follows:

[0059] Based on the gradient inversion method, the cost function C(x) can be expressed as follows in the k-th iteration:

[0060] C k (m)=F(x k )·R k (m) (1)

[0061] In the formula, the first term of the cost function represents the L2 norm of the measured data d and the forward response S(m).

[0062] m is the parameter vector to be inverted, R k (m) represents the adaptive regularization term:

[0063]

[0064] In the formula, δ is a constant value, which can be determined through numerical experiments. p As a reference model vector, m can be set to the model state from the previous iteration during the inversion process. p =m k-1 .

[0065] Solve equation (8) using the Gauss-Newton optimization method, setting its derivative with respect to x to zero, and perform the k-th iteration. have to

[0066]

[0067] In the formula, J represents the Jacobian matrix, and T represents the matrix transpose;

[0068] S0413. Based on the detection characteristics of the azimuth logging instrument while drilling, the inversion results are divided into three categories. The first category is where the positions h1 and h2 of the upper and lower interfaces are both greater than 2m, preferably 1.8m. The second category is where the positions h1 or h2 of the upper or lower interface are less than 2m, preferably 1.8m. The third category is where the positions h1 and h2 of the upper and lower interfaces are both less than 2m, preferably 1.8m. Since the accuracy of the inversion decreases and the uncertainty increases the farther the azimuth logging instrument is from the formation interface, the inversion results are generally accurate and reliable when the interface is within 2m of the instrument.

[0069] S0414. For the first type of inversion result, output resistivity as the current stratigraphic information; for the second type of inversion result, output resistivity and the position of the upper or lower interface as the current stratigraphic information; for the third type of inversion result, output resistivity and the positions of the upper and lower interfaces as the current stratigraphic information.

[0070] S05. Based on the remote detection and drilling electromagnetic wave logging data, establish a five-layer macroscopic anisotropic inversion model, and apply geological constraints to the inversion parameters based on the reservoir resistivity sequence extracted in step S01.

[0071] S06. When drilling azimuth electromagnetic logging data or traditional drilling electromagnetic logging data are available, based on the current stratigraphic information extracted in step S041 or S042, reduce the number of parameters to be inverted in the five-layer macroscopic anisotropy inversion model; specifically including the following sub-steps:

[0072] S061. If there is azimuth logging data while drilling, proceed to steps S062 to S064. If only conventional azimuth logging data while drilling is available, proceed to step S065.

[0073] S062. If step S0414 only outputs the resistivity information of the current layer, then the number of parameters to be inverted in the five-layer macroscopic anisotropy inversion model is reduced to 10.

[0074] S063. If step S0414 outputs the current layer resistivity information and the upper (lower) interface position, then the number of parameters to be inverted in the five-layer macroscopic anisotropy inversion model is reduced to 9.

[0075] S064. If step S0414 outputs the current layer resistivity information and the positions of the upper and lower interfaces, then the number of parameters to be inverted in the five-layer macroscopic anisotropy inversion model is reduced to 8.

[0076] S065. Based on the current layer resistivity information output in step S042, the number of parameters to be inverted in the five-layer macroscopic anisotropy inversion model is reduced to 10.

[0077] S07. The inversion model is sampled and inverted using the stochastic Bayesian inversion algorithm, and the inversion results are output. Based on Bayesian theory, the posterior distribution p(m|d) of the inversion solution is... obs This can be represented as:

[0078]

[0079] In the formula, m is the stratigraphic model, and d obs For measured data, p(d) obs |m) is the likelihood function, C d Let be the data covariance matrix. The Markov chain-Monte Carlo (MCMC) algorithm is used to sample the likelihood function distribution.

[0080] p(x|d obs )q(x,y)=p(y|d obs )q(y,x) (14)

[0081] In the formula, x and y are arbitrary models, and p(x|d obs Let be the limiting distribution of the Markov chain, and q(x,y) be the transition kernel function. Using the Metropolis-Hsting (MH) algorithm, an acceptor function α(y,x) is introduced, such that:

[0082] p(x|d obs )q(x,y)α(y,x)=p(y|d obs )q(y,x)α(y,x) (15)

[0083] In the formula:

[0084]

[0085] The specific implementation steps of MCMC are as follows:

[0086] 1) Generate the next candidate model y based on the current iterative model x and the transfer function q(x,y);

[0087] 2) Then calculate the acceptance probability α(x,y) and compare it with the random number β. If β≥α(x,y), return model y; otherwise, return model x.

[0088] 3) Repeat the above steps until the termination condition is met.

[0089] like Figures 2 to 6 As shown, a specific inversion example is given below, with the instrument traversing the strata from top to bottom:

[0090] Figure 2The diagram schematically illustrates a three-layer stratigraphic inversion model in an embodiment of the present invention; wherein, Figure 2 In the diagram, the meanings of each parameter are as follows: R t H represents the resistivity of the formation in which the instrument is located. up and H down θ represents the distance of the instrument from the upper and lower interfaces; θ represents the well inclination angle.

[0091] Figure 3 This diagram shows a five-layer formation model in an inversion example, with the black solid line representing the wellbore trajectory.

[0092] Figure 4 This represents the Bayesian inversion two-dimensional curtain diagram of the five-layer formation model in the inversion example, with the black solid line representing the wellbore trajectory.

[0093] Figure 5 This is a diagram showing the joint inversion results of traditional and remote-detection electromagnetic logging (EMR) data from three formations, using a five-layer formation inversion model. The black solid line represents the wellbore trajectory. Figure 5 (a) is the joint inversion result of the Bayesian algorithm with 4000 single-chain samplings; Figure 5 (b) is the joint inversion result of the Bayesian algorithm with 8000 single-chain samplings; Figure 5 (c) is the joint inversion result of the Bayesian algorithm with 20,000 single-chain samplings; Figure 5 (d) is the inversion result of reservoir imaging logging while drilling after 20,000 samplings using the Bayesian algorithm.

[0094] Figure 6 This is a diagram showing the joint inversion results of drilling azimuth electromagnetic logging and long-range drilling electromagnetic logging data from three formations, using a five-layer formation inversion model in an inversion example. The black solid line represents the wellbore trajectory. Figure 6 (a) is the inversion result of remote detection electromagnetic logging data during drilling; Figure 6 (b) is the result of joint inversion of remote detection electromagnetic logging while drilling and traditional electromagnetic logging while drilling; Figure 6 (c) is the inversion result based on 8000 single-chain sampling times of remote detection electromagnetic wave logging and drilling azimuth electromagnetic wave logging; Figure 6 (d) is the inversion result based on 20,000 single-chain sampling times of remote detection electromagnetic wave logging and drilling azimuth electromagnetic wave logging.

[0095] The inversion results obtained from the above inversion examples are highly consistent with the stratigraphic model, verifying the accuracy and reliability of the inversion algorithm in this embodiment of the invention.

[0096] This invention combines the deep-penetration advantages of remote-detection electromagnetic logging while drilling with the high resolution, fast inversion speed, and high inversion accuracy of azimuth electromagnetic logging while drilling and traditional electromagnetic logging while drilling. By inverting data from azimuth electromagnetic logging while drilling and traditional electromagnetic logging while drilling, formation resistivity and interface location information are obtained and used as known parameters to achieve rapid and accurate inversion of remote-detection electromagnetic logging while drilling data. Numerical experiments show that when only traditional electromagnetic logging-while-drilling (EMWL) data is available, the resistivity information of the current formation can be obtained. The number of parameters to be inverted from remote-detection EMWL data is reduced from 11 to 10, and the inversion speed is increased by approximately 3 times. When azimuth EMWL data is available, three scenarios can be considered: 1. The instrument distance from the upper and lower interfaces exceeds the detection range of the azimuth EMWL instrument. In this case, the resistivity information of the current formation can be obtained through inversion. The number of parameters to be inverted from remote-detection EMWL data is reduced to 10, and the inversion speed is increased by 3 times; 2. Instrument... When the distance from one of the interfaces is less than the instrument's detection range, the resistivity of the current formation and the location information of the interface can be obtained by inverting the azimuth electromagnetic wave logging data while drilling. The number of parameters to be inverted by the far-detection azimuth electromagnetic wave logging data is reduced to 9, and the inversion speed is increased by about 20 times. 3. When the distance from the instrument to both the upper and lower interfaces is less than the instrument's detection range, the resistivity of the current formation and the location information of the upper and lower interfaces can be obtained by inverting the azimuth electromagnetic wave logging data while drilling. The number of parameters to be inverted by the far-detection azimuth electromagnetic wave logging data is reduced to 8, and the inversion speed is increased by more than two orders of magnitude.

[0097] Example 2

[0098] The drilling electromagnetic logging data joint inversion device of this invention can implement the methods of the above embodiments.

[0099] Example 3

[0100] The storage medium of this invention stores a computer program, which, when run by a processor, executes the joint inversion method for electromagnetic logging data during drilling as described in the above embodiments.

[0101] As can be seen from the above embodiments, the joint inversion method, device and storage medium for electromagnetic logging data during drilling involved in this invention can effectively utilize the differences in edge detection capabilities and complementary information of electromagnetic logging data at different scales, thereby improving the inversion speed and accuracy of remote detection electromagnetic logging data during drilling.

[0102] Although the invention has been described with reference to preferred embodiments, various modifications can be made and components can be replaced with equivalents without departing from the scope of the invention. In particular, the technical features mentioned in the various embodiments can be combined in any manner as long as there is no structural conflict. The invention is not limited to the specific embodiments disclosed herein, but includes all technical solutions falling within the scope of the claims.

Claims

1. A method for joint inversion of electromagnetic logging-while-drilling data, characterized in that, Includes the following steps: S01. Extract the reservoir horizontal resistivity sequence from adjacent well data as a geological constraint condition; S02. Extract remote detection electromagnetic wave logging data while drilling as basic inversion data; S03. Determine whether there is electromagnetic logging-while-drilling data and the type of electromagnetic logging-while-drilling data. S04. Based on the type of electromagnetic logging data during drilling, use the corresponding method to invert the current stratigraphic information; S05. Based on the remote detection and drilling electromagnetic wave logging data, establish a five-layer macroscopic anisotropic inversion model, and apply geological constraints to the inversion parameters based on the reservoir resistivity sequence extracted in step S01. S06. When there is azimuth logging data or traditional logging data, reduce the number of parameters to be inverted in the five-layer macroscopic anisotropy inversion model based on the current layer information extracted in step S04. S07. Use the stochastic Bayesian inversion algorithm to perform sampling inversion on the inversion model and output the inversion results; Step S03 includes: determining whether there is azimuth logging data and conventional azimuth logging data. If there is azimuth logging data, proceed to step S041; if only conventional azimuth logging data is available, proceed to step S042. Step S04 specifically includes the following sub-steps: S041. Based on the azimuth electromagnetic logging data while drilling, establish a three-layer anisotropic formation model and invert the resistivity and interface information of the current layer. S042. Based on traditional electromagnetic logging data, the adaptive multiplier regularization Guass-Newton algorithm is used to invert the resistivity information of the current formation.

2. The method for joint inversion of electromagnetic logging data while drilling according to claim 1, characterized in that, Step S041 includes the following sub-steps: S0411. Based on the azimuth electromagnetic logging data while drilling, establish a three-layer anisotropic formation inversion model. According to the formation information given in step S01, determine the initial values ​​of each parameter to be inverted. S0412. Perform adaptive multiplier regularized Gauss-Newton inversion iteration on the model to obtain the inversion results; S0413. Based on the detection characteristics of the drilling azimuth electromagnetic wave logging instrument, the inversion results are divided into three categories: the first category is that the positions h1 and h2 of the upper and lower interfaces are both greater than the preset height; the second category is that the positions h1 or h2 of the upper or lower interface are less than the preset height; and the third category is that the positions h1 and h2 of the upper and lower interfaces are both less than the preset height. S0414. For the first type of inversion result, output resistivity as the current stratigraphic information; for the second type of inversion result, output resistivity and the position of the upper or lower interface as the current stratigraphic information; for the third type of inversion result, output resistivity and the positions of the upper and lower interfaces as the current stratigraphic information.

3. The method for joint inversion of electromagnetic logging data during drilling according to claim 2, characterized in that, In steps S0412 and S042, the adaptive multiplier regularization Gauss-Newton inversion iterative algorithm is specifically as follows: Based on the gradient inversion method, a cost function is used. In the k-th iteration, it is represented as: (1) In the formula, the first term of the cost function represents the L2 norm of the measured data d and the forward response S(m). ; m is the parameter vector to be inverted, Represents the adaptive regularization term: (2) In the formula, It is a fixed value, determined through numerical experiments; As a reference model vector, it is set to the model state from the previous iteration during the inversion process. ; The Gauss-Newton optimization method is used to solve (1), and its derivative with respect to x is zero. The k-th iteration... ,have to (3) In the formula, J represents the Jacobian matrix and T represents the matrix transpose.

4. The method for joint inversion of electromagnetic logging data during drilling according to claim 2 or 3, characterized in that, The preset height is 1.7~1.9m.

5. The method for joint inversion of electromagnetic logging data during drilling according to any one of claims 1 to 3, characterized in that, Step S06 specifically includes the following sub-steps: S061. If there is azimuth logging data while drilling, proceed to steps S062 to S064. If only conventional azimuth logging data while drilling is available, proceed to step S065. S062. If step S0414 only outputs the resistivity information of the current layer, then the number of parameters to be inverted in the five-layer macroscopic anisotropy inversion model is reduced to 10. S063. If step S0414 outputs the current layer resistivity information and the position of the upper or lower interface, then the number of parameters to be inverted in the five-layer macroscopic anisotropy inversion model is reduced to 9. S064. If step S0414 outputs the current layer resistivity information and the positions of the upper and lower interfaces, then the number of parameters to be inverted in the five-layer macroscopic anisotropy inversion model is reduced to 8. S065. Based on the current layer resistivity information output in step S042, the number of parameters to be inverted in the five-layer macroscopic anisotropy inversion model is reduced to 10.

6. The method for joint inversion of electromagnetic logging data during drilling according to any one of claims 1 to 3, characterized in that, In step S07, the Bayesian inversion algorithm is specifically as follows: Based on Bayesian theory, the posterior distribution of the inversion solution Represented as: (4) In the formula, For stratigraphic models, These are actual measured data. Let be the likelihood function. The data covariance matrix is ​​used; the Markov chain-Monte Carlo algorithm is employed to sample the likelihood function distribution. (5) In the formula, and Each is an arbitrary model. For the limiting distribution of Markov chains, To transfer the kernel function, the "Metropolis-Hsting" algorithm is adopted, and an acceptor function is introduced. , so that: (6) In the formula: 。 7. The method for joint inversion of electromagnetic logging data while drilling according to claim 6, characterized in that, Markov chain Monte Carlo algorithm includes: based on the current iteration model and transfer function Generate the next successor model Then calculate the acceptance probability. and with random numbers In comparison, if Then return the model. Otherwise, return the model. Repeat the above steps until the termination condition is met.

8. A joint inversion device for electromagnetic logging-while-drilling data, characterized in that, It is capable of implementing the method according to any one of claims 1 to 7.

9. A storage medium, characterized in that, It stores a computer program, which, when run by a processor, executes the method for joint inversion of electromagnetic logging data as described in any one of claims 1 to 7.

Images