An electromagnetic monitoring crack parameter identification method based on an improved marine predator algorithm

Abstract

This invention belongs to the field of oilfield development technology and discloses an electromagnetic monitoring fracture parameter identification method based on an improved marine predator algorithm. This invention optimizes and implements the IMPA algorithm, including improved population initialization, optimized optimization parameters, and adaptive boundary condition constraints. The improved population initialization uses a Sobol sequence to replace the traditional non-uniform random distribution population initialization. Parameter optimization proposes a staged optimization with differential partitioning to replace the traditional three-stage optimization of the MPA algorithm. Adaptive boundary condition constraints aim to discard local optimal solutions caused by parameter boundaries. A hydraulic fracturing single-fracture length model and an azimuth model are constructed. Electromagnetic monitoring involves supplying power to the fracturing wellbore and measuring the potential difference on the ground. The model forward modeling calculates the potential difference anomalies observed at all surface measuring points under the constructed single-fracture model, achieving precise identification of the single-fracture length and azimuth, thereby guiding the evaluation of hydraulic fracturing effectiveness.

Description

Technical Field

[0001] This invention belongs to the field of oilfield development technology, specifically relating to an electromagnetic monitoring method for identifying fracture parameters based on an improved marine predator algorithm. Background Technology

[0002] Among various oil and gas production enhancement and energy conversion measures, fracturing is both necessary and economically beneficial. Accurately inferring the morphology and dynamic propagation patterns of fracturing-induced fractures directly impacts the evaluation of fracturing effectiveness. Obtaining the true geometric parameters of fractures has long been a challenge for both academia and industry. Currently, microseismic and electromagnetic methods are the most common hydraulic pressure monitoring technologies. However, the microseismic signals induced by fracturing are extremely weak, resulting in a low signal-to-noise ratio and difficulty in source identification. Furthermore, the obtained reservoir stimulation volume (SRV) is much larger than the effective stimulation volume (ESRV) reached by the fracturing fluid, limiting the accuracy and precision of fracture identification. Electromagnetic methods are sensitive to the electrical changes caused by fluid flow and volume variations during fracturing, and the physical mechanism of electromagnetic monitoring technology is well-defined. Therefore, they hold great potential for dynamic monitoring of unconventional oil and gas reservoir stimulation.

[0003] However, current electromagnetic data processing still relies on traditional inversion techniques, heavily depending on accurate initial models. Furthermore, the interpretation accuracy is limited by the volume effect of electromagnetic methods, significantly impacting monitoring effectiveness. With the rapid development of computer hardware and algorithms, many scholars have introduced intelligent optimization algorithms such as Particle Swarm Optimization (PSO), Differential Evolution (DE), and Grey Wolf Optimization (GWO) into geophysical inversion, achieving good results. Compared to the aforementioned optimization algorithms, the Marine Predator algorithm offers faster optimization speed and higher accuracy, and has already been applied in scientific fields such as the power industry and medicine. However, there are no reports on its application in geophysics. The traditional MPA algorithm also suffers from poor stability in optimization performance. Therefore, applying an improved Marine Predator algorithm to electromagnetic monitoring for fracture parameter identification is of great significance for guiding the evaluation of hydraulic fracturing effectiveness. Summary of the Invention

[0004] The purpose of this invention is to provide an electromagnetic monitoring crack parameter identification method based on an improved marine predator algorithm, in order to solve the problem mentioned in the background art that the inversion effect of traditional hydraulic fracturing electromagnetic monitoring crack parameter is difficult to achieve in engineering applications.

[0005] To achieve the above objectives, the present invention adopts the following technical solution:

[0006] A method for identifying crack parameters using electromagnetic monitoring based on an improved marine predator algorithm includes the following steps:

[0007] S1: Optimization and Implementation of the IMPA Algorithm;

[0008] S2: Fracturing model construction;

[0009] S3: Forward modeling calculation;

[0010] S4: Inversion of single-slit endpoint coordinates using the IMPA algorithm;

[0011] S5: Crack parameter identification.

[0012] Furthermore, the IMPA algorithm is implemented, specifically including the following steps:

[0013] S1.1: Population initialization, wherein the parameters for population initialization include the prey matrix. Maximum number of iterations Current iteration number Minimum fit difference Top Predator Matrix The prey matrix The matrix uses a uniformly random, low-discretion Sobol sequence instead of a traditional non-uniform random distribution for initialization; this is the apex predator matrix. Depend on Initially calculate the fit difference and obtain the value corresponding to the minimum fit difference. ;

[0014] S1.2: Iteration termination condition judgment: if If the condition is met, proceed to step S1.3; otherwise, proceed to step S1.5.

[0015] S1.3: Prey update;

[0016] S1.4: Fitness updated, Elite updated;

[0017] S1.5: Output the top predator matrix The This represents the globally optimal solution.

[0018] Further updates to Prey are as follows:

[0019] S1.3.1: If Update the prey matrix using formula (1) Otherwise, proceed to step S1.3.2.

[0020] (1);

[0021] In the formula, . Population size; It is a random vector of Brownian motion, exhibiting normal distribution characteristics; for The maximum value of the vector elements. Let be a constant, taken as 3.2. For the top predator matrix, For the prey matrix, For the movement step size, ; Generate random numbers in the range [0, 1]. This is the term-by-term multiplication operator;

[0022] S1.3.2: If The first half is updated using equation (2). The latter half is updated using equation (3). Otherwise, proceed to step S1.3.3;

[0023] (2)

[0024] (3);

[0025] In the formula, Let be a random vector of the Lévy motion distribution. for The maximum value of the vector elements. An adaptive parameter for the predator's step size. , , and are all constants, where =3.2, =2, =0.9, =0.4;

[0026] S1.3.3: If Update using equation (4) Otherwise, proceed to step S1.4;

[0027] (4).

[0028] Further updates to Fitness and Elite are as follows:

[0029] Calculate all in the population The corresponding fit difference, if the minimum fit difference is obtained, satisfies the condition that the corresponding parameter is not on the boundary and its value is less than 1. Assign it to and the corresponding Assign to Update using formula (5) , Conversely, update directly. and ;

[0030] (5);

[0031] In the formula, , indicating that The probability of the effect. The vector array is randomly generated. If the generated number is less than 0.2, it is replaced with 0; otherwise, it is replaced with 1. r is a random number between [0,1]. and Represents a random index in the prey matrix.

[0032] Furthermore, in the construction of the fracturing model, the fracturing model is a single-fracturing model, and the depth of the single-fracturing model is within the range of 8000m, including a fracturing length model and an orientation model.

[0033] Furthermore, the forward modeling calculation is performed as follows:

[0034] Electromagnetic monitoring adopts the observation method of power supply from the fractured well and ground measurement. The forward modeling calculation is the calculation of the potential difference of all measuring points on the surface under the constructed single fracture model. The potential difference is calculated according to formula (6)~(8).

[0035] (6);

[0036] (7);

[0037] (8);

[0038] In the formula, Indicates in The potential of electrode M is measured at all times. for The potential difference at observation point M at any given time is the difference in potential between the measuring electrode M and the reference electrode N. It is a dipole moment. This represents the radius vector, i.e., the distance between the dipole center and M. Let k represent the imaginary number, and k be the wave number. Unit vector, For supply current, The electrical conductivity of the surrounding rock.

[0039] Furthermore, the inversion of single-slit endpoint coordinates using the IMPA algorithm is detailed as follows:

[0040] The coordinates include the coordinates of the two endpoints AB of the single slit. The inversion process can be described as the fitting difference between the forward model data generated in each iteration and the measured data. The fitting difference The minimum fit difference is calculated according to formula (9). That is, the fit difference Minimum value;

[0041] (9);

[0042] In the formula, , The length of the observed data, Represents observation data, This represents the forward modeling data of the inversion model.

[0043] Furthermore, the crack parameter identification includes single crack length identification and orientation identification. The specific identification process uses AB coordinate values ​​to further calculate the crack length and orientation.

[0044] Technical effects and advantages of the present invention: The electromagnetic monitoring crack parameter identification method based on the improved marine predator algorithm proposed in this invention has the following advantages compared with the prior art:

[0045] This invention optimizes and implements the IMPA algorithm, including improved population initialization, optimized search parameters, and adaptive boundary condition constraints. The improved population initialization uses a Sobol sequence instead of the traditional non-uniform random distribution. Parameter optimization proposes a phased optimization based on difference partitioning instead of the traditional three-stage optimization of the MPA algorithm. Adaptive boundary condition constraints aim to discard local optima caused by parameter boundaries. A hydraulic fracturing single-fracture length model and an azimuth model are constructed. Electromagnetic monitoring, through powering the fracturing wellbore, measures the potential difference on the ground. The model forward modeling calculates the potential difference anomalies observed at all surface measuring points under the constructed single-fracture model, achieving precise identification of the single-fracture length and azimuth, thereby guiding the evaluation of hydraulic fracturing effectiveness. Attached Figure Description

[0046] Figure 1 This is a flowchart of an electromagnetic monitoring crack parameter identification method based on an improved marine predator algorithm according to an embodiment of the present invention;

[0047] Figure 2 This is a flowchart illustrating the implementation of the IMPA algorithm in an embodiment of the present invention;

[0048] Figure 3 This is a schematic diagram of a hydraulic fracturing electromagnetic monitoring fracture model according to an embodiment of the present invention;

[0049] Figure 4 This is a schematic diagram illustrating the inversion of seam length model coordinate values ​​under different noise levels in an embodiment of the present invention;

[0050] Figure 5 This is a schematic diagram illustrating the relative error of seam length recognition under different noise levels in an embodiment of the present invention;

[0051] Figure 6 This is a schematic diagram illustrating the inversion of the location model coordinate values ​​under different noise levels in an embodiment of the present invention;

[0052] Figure 7 This is a schematic diagram of the absolute error of position recognition under different noise levels in an embodiment of the present invention;

[0053] Figure 8 for Figure 5 (a) Noise-free magnified view;

[0054] Figure 9 for Figure 5 (b) Magnified view of 5% random noise;

[0055] Figure 10 for Figure 5 (c) Magnified view of 10% random noise;

[0056] Figure 11 for Figure 5 (d) Magnified view of 15% random noise;

[0057] Figure 12 for Figure 7 (a) Noise-free magnified view;

[0058] Figure 13 for Figure 7 (b) Magnified view of 5% random noise;

[0059] Figure 14 for Figure 7 (c) Magnified view of 10% random noise;

[0060] Figure 15 for Figure 7 (d) Magnified view of 15% random noise. Detailed Implementation

[0061] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The specific embodiments described herein are merely used to explain the present invention and are not intended to limit the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0062] This invention provides a method for identifying electromagnetic monitoring crack parameters based on an improved marine predator algorithm, exemplarily, such as... Figure 1-7 The process, as shown, includes the following steps:

[0063] S1: Optimization and implementation of the IMPA algorithm, specifically including the following steps:

[0064] S1.1: Population initialization, wherein the parameters for population initialization include the prey matrix. Maximum number of iterations Current iteration number Minimum fit difference Top Predator Matrix The prey matrix The matrix is ​​initialized using a uniformly random, low-discretion Sobol sequence instead of a traditional non-uniform random distribution, and the maximum number of iterations... The current iteration number =0, the minimum fitting difference The initial value is 1000, and the top predator matrix is... Depend on Initially calculate the fit difference and obtain the value corresponding to the minimum fit difference. ;

[0065] S1.2: Iteration termination condition judgment: if If the condition is met, proceed to step S1.3; otherwise, proceed to step S1.5.

[0066] S1.3: Prey updated, details are as follows:

[0067] S1.3.1: If Update the prey matrix using formula (1) Otherwise, proceed to step S1.3.2;

[0068] (1);

[0069] In the formula, . Population size; It is a random vector of Brownian motion, exhibiting normal distribution characteristics; for The maximum value of the vector elements. Let be a constant, taken as 3.2. For the top predator matrix, For the prey matrix, For the movement step size, ; Generate random numbers in the range [0, 1]. This is the term-by-term multiplication operator;

[0070] S1.3.2: If The first half is updated using equation (2). The latter half is updated using equation (3). Otherwise, proceed to step S1.3.3;

[0071] (2)

[0072] (3);

[0073] In the formula, Let be a random vector of the Lévy motion distribution. for The maximum value of the vector elements. An adaptive parameter for the predator's step size. , , and are all constants, where =3.2, =2, =0.9, =0.4;

[0074] S1.3.3: If Update using equation (4) Otherwise, proceed to step S1.4;

[0075] (4);

[0076] S1.4: Fitness updated, Elite updated;

[0077] Calculate all in the population The corresponding fit difference, if the minimum fit difference is obtained, satisfies the condition that the corresponding parameter is not on the boundary and its value is less than 1. Assign it to and the corresponding Assign to Update using formula (5) , Conversely, update directly. and ;

[0078] (5);

[0079] In the formula, , indicating that The probability of the effect. The vector array is randomly generated. If the generated number is less than 0.2, it is replaced with 0; otherwise, it is replaced with 1. r is a random number between [0,1]. and Represents a random index in the prey matrix;

[0080] S1.5: Output the top predator matrix The This represents the globally optimal solution.

[0081] S2: Fracturing model construction, as detailed below:

[0082] The fracturing model is a horizontal single-fracture model with a burial depth of 4000m. It includes a fracture length model and an azimuth model. The fracture length model maintains an azimuth of 0° and has fracture lengths of 100m, 200m, 300m, and 400m. The azimuth model maintains a fracture length of 200m and has azimuths of 0°, 10°, 20°, and 30°. Figure 3 A schematic diagram of a hydraulic fracturing electromagnetic monitoring fracture model. Figure 3 In the diagram, 'a' is a three-dimensional view of the model, and 'b' is a top-view view of the model.

[0083] S3: Forward Modeling Calculation

[0084] The electromagnetic monitoring adopts the observation method of power supply from the fractured well and surface area measurement. The area measurement area is 2000*2000m2, the distance between measuring points is 50m, the line distance is 50m, and the number of measuring points is 1681. The forward modeling calculation is the calculation of the potential difference of all measuring points on the surface under the constructed single fracture model. The potential difference is calculated according to formula (6)~(8).

[0085] (6)

[0086] (7)

[0087] (8);

[0088] In the formula, Indicates in The potential of electrode M is measured at all times. for The potential difference at observation point M at any given time is the difference in potential between the measuring electrode M and the reference electrode N. It is a dipole moment. This represents the radius vector, i.e., the distance between the dipole center and M. Let k represent the imaginary number, and k be the wave number. Unit vector, For supply current, 1A. The electrical conductivity of the surrounding rock is 0.01 S / m, and the power supply frequency is 1 Hz.

[0089] S4: Inversion of single-slit endpoint coordinates using the IMPA algorithm

[0090] The coordinates include the coordinates of the two endpoints AB of the single slit. The Z components Az and Bz of AB are fixed at 4000m. The coordinate values ​​of Ax, Ay, Bx, and By are inverted. The inversion process can be described as the fitting difference between the forward model data and the measured data generated in each iteration. The fitting difference The minimum fit difference is calculated according to formula (9). That is, the fit difference Minimum value;

[0091] (9);

[0092] In the formula, , The length of the observed data, Represents observation data, This represents the forward modeling data of the inversion model.

[0093] S5: Crack parameter identification: The crack parameter identification includes single crack length identification and orientation identification. The specific identification process uses AB coordinate values ​​to further calculate the crack length and orientation.

[0094] The results show that the proposed method has good feasibility. It is evident that the proposed method can accurately identify hydraulic fracturing fractures. Through optimization and implementation of the IMPA algorithm, including improved population initialization, optimized search parameters, and adaptive boundary condition constraints, the improved population initialization uses a Sobol sequence instead of the traditional non-uniform random distribution population initialization. Parameter optimization proposes a staged optimization with differential partitioning instead of the traditional three-stage optimization of the MPA algorithm. Adaptive boundary condition constraints aim to discard local optima caused by parameter boundaries. A hydraulic fracturing single-fracture length model and an azimuth model are constructed. Electromagnetic monitoring, through powering the fracturing wellbore and measuring potential differences on the ground, calculates the potential difference anomalies observed at all surface measuring points under the constructed single-fracture model, achieving accurate identification of the single-fracture length and azimuth, thereby guiding the evaluation of hydraulic fracturing effectiveness.

[0095] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for identifying electromagnetic monitoring crack parameters based on an improved marine predator algorithm, characterized in that, Includes the following steps: S1: Optimization and implementation of the IMPA algorithm, specifically including the following steps: S1.1: Population initialization, wherein the parameters for population initialization include the prey matrix. Maximum number of iterations Current iteration number Minimum fit difference Top Predator Matrix The prey matrix The matrix uses a uniformly random, low-discretion Sobol sequence instead of a traditional non-uniform random distribution for initialization; this is the apex predator matrix. Depend on Initially calculate the fit difference and obtain the value corresponding to the minimum fit difference. ; S1.2: Iteration termination condition judgment: if If the condition is met, proceed to step S1.3; otherwise, proceed to step S1.

5. S1.3: Prey update; S1.4: Fitness updated, Elite updated; S1.5: Output the top predator matrix The Indicates the globally optimal solution; S2: Construction of fracturing model, wherein the fracturing model is a single-fracturing model, the depth of which is within 8000m, including fracturing length model and orientation model; S3: Forward model calculation, as detailed below: Electromagnetic monitoring adopts the observation method of power supply from the fractured well and ground measurement. The forward modeling calculation is the calculation of the potential difference of all measuring points on the surface under the constructed single fracture model. The potential difference is calculated according to formula (6)~(8). (6) (7) (8)； In the formula, Indicates in The potential of electrode M is measured at all times. for The potential difference at observation point M at any given time is the difference in potential between the measuring electrode M and the reference electrode N. It is a dipole moment. This represents the radius vector, i.e., the distance between the dipole center and M. Let k represent the imaginary number, and k be the wave number. Unit vector, For supply current, The electrical conductivity of the surrounding rock; S4: Inversion of single-slit endpoint coordinates using the IMPA algorithm, as detailed below: The coordinates include the coordinates of the two endpoints AB of the single slit. The inversion process can be described as the fitting difference between the forward model data generated in each iteration and the measured data. The fitting difference The minimum fit difference is calculated according to formula (9). That is, the fit difference Minimum value; (9)； In the formula, , The length of the observed data, Represents observation data, Represents the forward modeling data of the inversion model; S5: Crack parameter identification.

2. The electromagnetic monitoring crack parameter identification method based on an improved marine predator algorithm according to claim 1, characterized in that: Prey update, the specific steps are as follows: S1.3.1: If Update the prey matrix using formula (1) Otherwise, proceed to step S1.3.

2. （1）； In the formula, . Population size; It is a random vector of Brownian motion, exhibiting normal distribution characteristics; for The maximum value of the vector elements. Let be a constant, taken as 3.

2. For the top predator matrix, For the prey matrix, For the movement step size, ; Generate random numbers in the range [0, 1]. This is the term-by-term multiplication operator; S1.3.2: If The first half is updated using equation (2). The latter half is updated using equation (3). Otherwise, proceed to step S1.3.3; （2） （3）； In the formula, Let be a random vector of the Lévy motion distribution. for The maximum value of the vector elements. An adaptive parameter for the predator's step size. , , and are all constants, where =3.2, =2, =0.9, =0.4; S1.3.3: If Update using equation (4) Otherwise, proceed to step S1.4; （4）。 3. The electromagnetic monitoring crack parameter identification method based on an improved marine predator algorithm according to claim 2, characterized in that: Fitness and Elite updates are detailed below: Calculate all in the population The corresponding fit difference, if the minimum fit difference is obtained, satisfies the condition that the corresponding parameter is not on the boundary and its value is less than 1. Assign it to and the corresponding Assign to Update using formula (5) , Conversely, update directly. and ; （5）； In the formula, , indicating that The probability of the effect. The vector array is randomly generated. If the generated number is less than 0.2, it is replaced with 0; otherwise, it is replaced with 1. r is a random number between [0,1]. and Represents a random index in the prey matrix.

4. The electromagnetic monitoring crack parameter identification method based on an improved marine predator algorithm according to claim 3, characterized in that: The crack parameter identification includes single crack length identification and orientation identification. The specific identification process uses AB coordinate values ​​to further calculate the crack length and orientation.

Images