Modeling optimization method and device of geologic model, equipment, storage medium and product

Abstract

The invention provides a modeling optimization method and device of a geological model, equipment, a storage medium and a product, and relates to the technical field of geological sciences, and the method comprises the steps: constructing an initial geological voxel model based on the geological information of a to-be-researched region; performing level set function modeling on each geological interface to obtain a plurality of initial voxel level set functions; based on each initial voxel level set function, performing gravity anomaly forward modeling to obtain theoretical gravity anomaly information; based on a genetic algorithm, optimizing each initial voxel level set function to obtain a coarse resolution geologic voxel model; and based on the actual gravity anomaly information and the theoretical gravity anomaly information, performing fine resolution optimization on the coarse resolution geologic voxel model to obtain a target geologic model. Through the above mode, the resolution of the model can be effectively improved, the accuracy and reliability of the target geologic model are ensured, and the target geologic model can accurately reflect the real geologic structure of the to-be-studied area.

Description

Technical Field

[0001] The present invention relates to the technical field of geological science, and in particular, to a method, device, equipment, storage medium and product for optimizing the modeling of a geological model. Background Art

[0002] As an important tool for geological research, three-dimensional geological modeling can intuitively present the spatial form and characteristics of geological structures in the research area. The complexity of deep geological structures with multiple faults makes its modeling and optimization face many challenges. Accurately modeling and optimizing deep geological structures with multiple faults is of great significance for fields such as mineral resource exploration and geological disaster prevention.

[0003] In some related technologies, three-dimensional geological modeling can be carried out according to the geological information of the research area to construct a three-dimensional geological voxel model corresponding to the area. Then, activities such as geological science research, mineral resource exploration, and geological disaster prevention can be carried out based on the three-dimensional geological voxel model.

[0004] However, the accuracy and reliability of the geological voxel model constructed by the existing modeling methods are low, and it is difficult to accurately reflect the true geological structure of the research area. Summary of the Invention

[0005] The present invention provides a method, device, equipment, storage medium and product for optimizing the modeling of a geological model to solve the defect that the accuracy and reliability of the geological voxel model constructed by the existing modeling methods are low and it is difficult to accurately reflect the true geological structure of the research area.

[0006] The present invention provides a method for optimizing the modeling of a geological model, including: constructing an initial geological voxel model based on the geological information of the area to be studied; the initial geological voxel model includes geological interfaces of multiple geological bodies; respectively performing level set function modeling on each geological interface to obtain multiple initial voxel level set functions; based on each initial voxel level set function, performing forward gravity anomaly calculation to obtain theoretical gravity anomaly information; based on the genetic algorithm, optimizing each initial voxel level set function to obtain a coarse-resolution geological voxel model; based on the actual gravity anomaly information and the theoretical gravity anomaly information, performing fine-resolution optimization on the coarse-resolution geological voxel model to obtain a target geological model.

[0007] An optimization method for building a geological model provided by the present invention, based on a genetic algorithm, optimizes each initial voxel level set function to obtain a coarse-resolution geological voxel model, including: generating an initial population of the genetic algorithm; the initial population includes multiple individuals, and one individual is an initial voxel level set function; based on the elitist retention strategy and the tournament selection strategy, select multiple individuals from the initial population as parental individuals; perform population crossover operations on the multiple parental individuals to obtain a parental population; perform Cauchy mutation and Gaussian mutation on the parental population to obtain multiple offspring individuals; perform population update based on the multiple offspring individuals to obtain an updated population; the updated population includes multiple updated individuals, and one updated individual is an optimized voxel level set function; based on the updated population, determine whether a preset termination condition is satisfied; if the preset termination condition is not satisfied, return to the step of selecting multiple individuals from the initial population as parental individuals based on the elitist retention strategy and the tournament selection strategy until the preset termination condition is satisfied, and generate a coarse-resolution geological voxel model based on the updated population.

[0008] An optimization method for building a geological model provided by the present invention, an optimized voxel level set function includes multiple parameters; generating a coarse-resolution geological voxel model based on the updated population, including: determining the population fitness of the updated population; based on the contribution degree of each optimized voxel level set function to the population fitness, freeze some parameters of each optimized voxel level set function respectively to obtain multiple reduced-dimensional voxel level set functions; based on the Markov chain Monte Carlo method, perform uncertainty quantification on the multiple reduced-dimensional voxel level set functions to generate a coarse-resolution geological voxel model.

[0009] An optimization method for building a geological model provided by the present invention, the population fitness is determined based on the fitness value, the individual age, and a preset coefficient for controlling the attenuation of the individual age; wherein, the fitness value is determined based on the model complexity penalty value, the prior knowledge fusion value, and the root mean square error of the actual gravity anomaly information and the theoretical gravity anomaly information.

[0010] An optimization method for building a geological model provided by the present invention, based on the actual gravity anomaly information and the theoretical gravity anomaly information, perform fine-resolution optimization on the coarse-resolution geological voxel model to obtain a target geological model, including: determining the gravity anomaly error based on the actual gravity anomaly information and the theoretical gravity anomaly information; determining the voxel change error based on each initial voxel level set function and each optimized voxel level set function; perform weighted aggregation operation on the gravity anomaly error and the voxel change error to obtain the total error; respectively determine the error gradient of the total error with respect to the variables of each optimized voxel level set function; perform fine-resolution optimization on the coarse-resolution geological voxel model based on each error gradient to obtain a target geological model.

[0011] According to a method for optimizing the modeling of a geological model provided by the present invention, based on the geological information of the area to be studied, an initial geological voxel model is constructed, including: based on the geological information of the area to be studied, surface modeling is performed on the surface of each geological body to obtain a surface model; voxelization modeling is performed on the surface model to obtain an initial geological voxel model.

[0012] The present invention also provides a device for optimizing the modeling of a geological model, including: a first construction module for constructing an initial geological voxel model based on the geological information of the area to be studied; the initial geological voxel model includes geological interfaces of a plurality of geological bodies; a second construction module for performing level set function modeling on each geological interface respectively to obtain a plurality of initial voxel level set functions; a gravity anomaly forward modeling module for performing gravity anomaly forward modeling based on each initial voxel level set function to obtain theoretical gravity anomaly information; a first optimization module for optimizing each initial voxel level set function based on a genetic algorithm to obtain a coarse-resolution geological voxel model; a second optimization module for performing fine-resolution optimization on the coarse-resolution geological voxel model based on actual gravity anomaly information and theoretical gravity anomaly information to obtain a target geological model.

[0013] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor, and when the processor executes the computer program, it implements the method for optimizing the modeling of a geological model as described in any one of the above.

[0014] The present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, it implements the method for optimizing the modeling of a geological model as described in any one of the above.

[0015] The present invention also provides a computer program product, including a computer program, and when the computer program is executed by a processor, it implements the method for optimizing the modeling of a geological model as described in any one of the above.

[0016] The modeling optimization method, device, equipment, storage medium and product of the geological model provided by the present invention, after constructing an initial geological voxel model according to the geological information of the area to be studied, perform level set function modeling on the initial geological voxel model to obtain a plurality of initial voxel level set functions, and then use the genetic algorithm to perform coarse-resolution optimization on each initial voxel level set function to obtain a coarse-resolution geological voxel model. Finally, use the actual gravity anomaly information and the theoretical gravity anomaly information obtained by gravity anomaly forward modeling to perform fine-resolution optimization on the coarse-resolution geological voxel model to obtain the target geological model. In the above manner, after obtaining the initial geological voxel model, first perform coarse-resolution optimization through level set function modeling and genetic algorithm, and then perform fine-resolution optimization through the theoretical gravity anomaly information obtained by gravity anomaly forward modeling, which can effectively improve the resolution of the model, ensure the accuracy and reliability of the target geological model, and enable the target geological model to accurately reflect the true geological structure of the area to be studied. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following will briefly introduce the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0018] Figure 1 It is a schematic flowchart of the modeling optimization method of the geological model provided by the present invention.

[0019] Figure 2 It is a schematic flowchart of the optimization process of the genetic algorithm provided by the present invention.

[0020] Figure 3 It is a schematic flowchart of the fine-resolution optimization provided by the present invention.

[0021] Figure 4 It is a schematic structural diagram of the modeling optimization device of the geological model provided by the present invention.

[0022] Figure 5 It is a schematic structural diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0023] To make the objectives, technical solutions, and advantages of the present invention clearer, the following will clearly and completely describe the technical solutions in the present invention with reference to the accompanying drawings in the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art without creative efforts based on the embodiments of the present invention belong to the scope of protection of the present invention.

[0024] Please refer to Figure 1 , Figure 1 which is a schematic flow chart of the method for optimizing the modeling of the geological model provided by the present invention. In this embodiment, the method for optimizing the modeling of the geological model includes steps S110 to S150, and the specific steps are as follows: S110: Based on the geological information of the area to be studied, construct an initial geological voxel model.

[0025] The initial geological voxel model includes the geological interfaces of multiple geological bodies.

[0026] Optionally, obtain the geological information of the area to be studied through field geological surveys and geophysical explorations.

[0027] Field geological surveys refer to obtaining geological information such as faults and rocks in the area to be studied through means such as on-site observation and measurement, providing basic data for subsequent modeling. In field geological surveys, the primary goal is to clarify the spatial distribution of different types of geological bodies (such as rocks, strata, ore bodies, etc.). Through on-site observation, determine the outcrop positions, ranges of various geological bodies, and their contact relationships. For example, when conducting surveys in mountainous areas, it is necessary to determine the contact boundary between granite bodies and surrounding sedimentary rock strata, which is crucial for understanding the geological structure evolution and subsequent modeling work. Identifying geological structures such as faults and folds is a key content of field geological surveys. For faults, it is necessary to determine parameters such as their positions, strikes, dips, and throw distances, and faults can be identified by observing signs such as the displacement of strata, fracture zones of rocks, and fault gouges. For folds, it is necessary to determine their types (such as anticlines, synclines), the strike and plunge directions of the axial planes, and the attitudes of the strata on both wings.

[0028] Geophysical exploration is a method of detecting underground geological structures and the distribution of geological bodies by studying and observing the changes in various geophysical fields, which makes full use of the differences in physical properties (such as density, magnetism, electricity, elasticity, etc.) of different geological bodies. The differences in physical properties of different geological bodies will cause corresponding changes in geophysical fields (such as gravity fields, magnetic fields, electric fields, seismic wave fields, etc.), and the actual situation of underground geology can be inferred by measuring and analyzing these changes. Geophysical exploration includes seismic exploration, gravity exploration, magnetic exploration, and electrical exploration, etc.

[0029] Seismic exploration utilizes the propagation characteristics of artificially generated seismic waves in subsurface media to detect geological structures. When an artificial seismic source (such as an explosive explosion or a vibroseis) generates seismic waves, the seismic waves will undergo reflection, refraction, and transmission at the interfaces of different subsurface media. By deploying geophones on the ground or in wells, these reflected or refracted seismic wave signals can be received. Since the elastic properties (such as P-wave velocity, S-wave velocity, density, etc.) of different rocks and formations are different, the propagation velocity, reflection, and refraction of seismic waves in them are also different. For example, the seismic wave velocity of sedimentary rocks is lower than that of igneous rocks. By analyzing parameters such as the travel time, amplitude, and frequency of seismic waves, the thickness, burial depth, lithological changes of subsurface formations, and the location of geological structures such as faults can be inferred. The seismic profile is the main result of seismic exploration. Technicians can infer the distribution of subsurface formations and geological structures by identifying characteristics such as the shape, continuity, and amplitude variation of seismic reflection interfaces. For example, a continuous strong reflection interface may represent a stable formation interface, while the interruption or dislocation of the reflected wave may indicate the presence of a fault. By combining with some known geological information (such as borehole data), the formation and fault structures in the 3D geological model can be constructed.

[0030] Gravity exploration is based on the gravity field changes caused by the density differences of different geological bodies to detect subsurface geological structures. According to Newton's law of universal gravitation, a geological body with a larger mass (higher density) generates a larger gravitational acceleration. By measuring the gravitational acceleration values at various points on the ground and comparing them with the theoretical gravity values (normal gravity values), the gravity anomaly values can be obtained. Positive gravity anomaly values may indicate the presence of high-density geological bodies (such as metal ore bodies, basic rock intrusions, etc.) underground, and negative gravity anomaly values may indicate the presence of low-density geological bodies (such as salt domes, cavities, etc.) underground. For example, when searching for salt mines, due to the low density of salt, an obvious negative gravity anomaly value will appear above the salt mine. By analyzing the shape, size, and distribution of gravity anomaly values, combined with known geological information and the density parameters of geological bodies, the shape, size, burial depth, and distribution of subsurface geological bodies can be inferred. For example, for an isolated gravity high anomaly value, it can be speculated that there is a small high-density rock mass underground. By establishing a physical model for forward gravity anomaly simulation (calculating the theoretical gravity anomaly according to the assumed geological body model) and inverse gravity anomaly simulation (inferring the geological body model based on the actual gravity anomaly values), the model can be continuously optimized, and the results of gravity exploration can be used to describe the distribution of geological bodies in 3D geological modeling.

[0031] Magnetic prospecting refers to the use of the magnetic differences of different geological bodies to detect the underground geological structure. The Earth itself is a large magnet, and underground rocks and geological bodies will acquire different degrees of magnetism during their formation and evolution. Magnetic geological bodies generate magnetic fields in the surrounding space, causing changes in the magnetic field intensity on the ground and forming magnetic anomalies. For example, geological bodies with more ferromagnetic minerals (such as magnetite) have stronger magnetism and will cause positive magnetic anomalies; while some non-magnetic or weakly magnetic geological bodies (such as limestone) may produce weak magnetic anomalies or negative magnetic anomalies. Based on the characteristics of magnetic anomalies (such as intensity, shape, strike, etc.), combined with the magnetic parameters of geological bodies and known geological information, the type, shape, size, and burial depth of underground geological bodies can be inferred. For example, a linear strong magnetic anomaly zone may indicate the existence of a magnetic dike or fault zone underground. The results of magnetic prospecting can provide information on the magnetic distribution of geological bodies for 3D geological modeling, assisting in determining geological structures and lithological boundaries.

[0032] Electrical prospecting refers to the detection of geological structures and the distribution of geological bodies based on the electrical differences of underground geological bodies (such as resistivity, dielectric constant, etc.). The resistivity differences between different rocks and strata are relatively large. For example, the resistivity of water-bearing loose sediments is relatively low, while that of dense rocks is relatively high. When an electric field is applied underground or on the ground, the distribution of current in the underground medium will be affected by the electrical properties of geological bodies. By measuring the changes in electrical parameters such as potential and electric field intensity on the ground or in wells, the underground geological conditions can be inferred. Based on the characteristics of resistivity images or other electrical prospecting parameter images (such as the distribution and shape of high-resistivity areas and low-resistivity areas), combined with known geological information, the lithology, water content, location of geological structures such as faults of underground geological bodies can be inferred. For example, a low-resistivity area may indicate the existence of a water-bearing formation or a metal sulfide ore body underground. The results of electrical prospecting can provide information on the electrical distribution of geological bodies for 3D geological modeling, which is used to optimize the description of the physical properties of geological bodies.

[0033] In addition, obtaining information such as the physical properties and rock types of geological bodies through drilling can also provide key data for subsequent modeling. The core samples obtained during the drilling process are the most intuitive materials. By observing the core samples, the physical characteristics such as the rock types, colors, structures, and textures of strata at different depths can be understood in detail. For example, observing the core samples can distinguish different sedimentary rocks such as sandstone, shale, and limestone, as well as igneous rocks and metamorphic rocks. For sedimentary rocks, structures such as bedding and cross-bedding can also be seen, and this information helps to determine the sedimentary environment and the sedimentary sequence of the strata. The mineral composition in the core samples can also provide important clues. For example, a core sample containing specific metal minerals can indicate the location of potential mineral resources. By analyzing the mineral assemblage of the core samples, the hydrothermal activities or metamorphic processes during the geological history period can also be inferred. Therefore, drilling data is an important criterion for verifying the accuracy of the three-dimensional geological model. Compare the information such as the stratigraphic interfaces, lithology distributions, and geological structures predicted by the model with the actual observations from the drilling. If the model is inconsistent with the drilling data, for example, the predicted stratigraphic thickness in the model is different from the thickness revealed by the drilling, or there is a deviation in the fault location in the model compared with the fault location found in the drilling, then the model needs to be adjusted and calibrated. Drilling data is crucial for identifying and modeling faults. If a dislocation or fracture zone of the strata is found in the drilling core samples, or sudden changes (such as sudden resistivity changes or abnormal acoustic time differences) are observed in the geophysical logging curves, these may be signs of the existence of a fault. By positioning the fault through multiple drill holes, the strike and dip of the fault can be determined. In the three-dimensional geological model, the fault can be represented as a surface with a certain geometric shape, and its position and shape are constrained by the drilling data. For example, based on the position and throw information of the drill hole passing through the fault, an accurate fault model can be constructed to show the extension of the fault in three-dimensional space and its cutting of the strata.

[0034] Specifically, after obtaining the geological information of the area to be studied through the above various methods, a surface model can be constructed for the surface of each geological body (such as faults, stratigraphic interfaces, etc.) according to the geological information. By connecting these surfaces, a three-dimensional surface model is formed.

[0035] This method regards geological structures as objects composed of a series of surfaces, and these surfaces can be stratigraphic interfaces, fault surfaces, contact surfaces of intrusions, etc. By accurately describing and combining these surfaces, a three-dimensional surface model that can reflect the spatial morphology and mutual relationships of real geological bodies can be constructed.

[0036] Optionally, according to the geological information, mathematical methods are used to construct the stratigraphic interface. For example, for a simple horizontal stratum, a planar stratigraphic interface can be constructed by giving the top and bottom depths and the planar range of the stratum. For an inclined stratum, parameters such as the strike, dip, and dip angle of the stratum need to be considered, and usually, spatial geometric algorithms (such as plane equations) are used to construct the stratigraphic interface.

[0037] Optionally, a fault plane is constructed based on parameters such as the position, strike, dip, and throw of the fault. The fault plane can be regarded as a spatial plane with a specific geometric shape, and the construction of the fault plane is achieved by substituting the geometric parameters of the fault into the plane equation. For example, in three-dimensional space, if the strike of a fault is northeast-southwest, the dip is southeast, and the dip angle is 60°, then a geometric model of the fault plane can be constructed using these parameters.

[0038] For the contact surfaces between intrusions and surrounding rocks, between different lithologic bodies, etc., their geometric characteristics can also be determined according to the geological information obtained from field geological surveys and geophysical explorations, and then appropriate mathematical methods are used for construction. For example, for a granite body intruding into sedimentary rocks, by observing its contact relationship with the surrounding rocks (such as intrusive contact, sedimentary contact, etc.), the shape and position of the contact surface are determined, and then the contact surface is constructed.

[0039] After constructing the surfaces of each geological body, these surfaces are combined and spliced to form a complete three-dimensional surface model.

[0040] Optionally, for stratigraphic interfaces, splicing is carried out in the sedimentary order of the strata to ensure that the older strata are below and the newer strata are above. For example, when constructing a three-dimensional surface model containing multiple sedimentary strata, the Cambrian stratigraphic interface, Ordovician stratigraphic interface, etc. are spliced in sequence according to the geological age order.

[0041] Optionally, for the combination of the fault plane and the stratigraphic interface, the cutting effect of the fault on the strata needs to be considered. When the fault plane intersects the stratigraphic interface, according to the nature of the fault (normal fault, reverse fault or strike-slip fault) and the throw, corresponding dislocation and deformation processing are carried out on the strata. For example, for a normal fault, the strata above the fault plane are displaced downward according to the throw to match the strata below the fault plane, so as to truly reflect the cutting and dislocation effects of the fault on the strata.

[0042] Furthermore, the surface model is voxelized to obtain an initial geological voxel model.

[0043] Specifically, after completing the surface modeling, using Geological Object Computer Aided Design software, the voxel size is determined according to the size of the area to be studied, the complexity of the geological structure, and the required model accuracy. If the area to be studied is large and the accuracy requirement is not extremely high, the voxel size can be relatively large; conversely, if the area to be studied is small and the geological structure is complex and fine representation is required, the voxel size should be small. For example, in a preliminary study of a large sedimentary basin, the voxel side length can be set to dozens of meters, while in a study of a small ore vein, the voxel side length may be only a few centimeters.

[0044] Divide the three-dimensional space containing the surface model in units of the selected voxel size, which is approximately equivalent to filling the entire three-dimensional space region with multiple small cubes to form a regular voxel grid. For example, for a cuboid-shaped area to be studied, divide it in the three directions of length, width, and height according to the voxel size to determine the spatial coordinate range of each voxel.

[0045] For each surface in the surface model (such as a formation interface, a fault surface, etc.), judge its intersection with the voxel grid. Since this is a complex geometric judgment process, spatial geometric algorithms are usually used to implement it. For example, for a planar formation interface, by calculating the spatial position relationship between its plane equation and the vertices of the voxels, determine whether the voxel intersects with the formation interface. The vertex coordinates of the voxel can be substituted into the equation of the surface, and according to the calculation result (such as greater than, less than, or equal to zero), determine which side of the surface the vertex is located on. If some vertices of a voxel are on one side of the surface and other vertices are on the other side, it can be judged that the voxel intersects with the surface.

[0046] After determining that the voxel intersects with the surface, determine the range of voxel attribute assignment according to the attributes of the surface. For example, for a formation interface, the lithologies on both sides are different. According to the position and lithology information of the surface, the voxels can be divided into different lithology regions. If a voxel is completely inside a certain formation, assign the corresponding lithology attribute to the voxel according to the lithology of the formation; if the voxel crosses the formation interface, determine the mixed lithology attribute of the voxel according to the proportion relationship of the lithologies on both sides of the interface (such as by calculating the volume proportion of the part where the voxel intersects with the interface).

[0047] According to the intersection judgment between the surface and the voxel and the determination of the range of attribute assignment, assign geological attributes to each voxel. The attribute assignment can be carried out in various ways, such as direct assignment (assigning values according to the single attribute of the intersecting surface), interpolation assignment (for voxels spanning multiple attribute regions, determining their attribute values through interpolation methods). For example, for the lithology attribute, assign lithology categories such as sandstone and shale to the voxels according to the formation lithology information in the surface model; for physical attributes (such as porosity, permeability, etc.), according to the known formation physical property data, combined with the intersection situation between the voxel and the surface, assign values to the voxels through linear interpolation or other appropriate interpolation methods.

[0048] After completing the attribute assignment, an initial geological voxel model can be obtained.

[0049] S120: Model the level set function for each geological interface respectively to obtain multiple initial voxel level set functions.

[0050] The level set function is an implicit surface representation method that describes the boundary of a geometric shape by defining a high-dimensional function. In the initial geological voxel model, the level set function is usually defined in three-dimensional space, and its zero level set function (the set of points with a function value of zero) represents the boundary of the geological body, such as the formation interface, fault plane, etc.

[0051] In this embodiment, the initial geological voxel model can be represented as a matrix, and the matrix value of each row can be represented as (x coordinate, y coordinate, z coordinate, lithology id).

[0052] Specifically, the initial geological voxel model includes the geological interfaces of multiple geological bodies. First, the voxels corresponding to each geological interface are extracted separately, and then the level set function is modeled for each geological interface separately to obtain multiple initial voxel level set functions.

[0053] The expression of the initial voxel level set function can be represented by formula (1): where, is the Euclidean distance from each voxel in the area to be studied to the corresponding geological interface.

[0054] The calculation formula of the Euclidean distance is shown in formula (2): ; where, , are the x-axis coordinate, y-axis coordinate, and z-axis coordinate of the voxel; similarly, , are the x-axis coordinate, y-axis coordinate, and z-axis coordinate of the connection point of the voxel perpendicular to the geological interface.

[0055] Select a certain direction of the geological interface as the positive direction ( ), when the voxel is located in the positive direction of the geological interface, the distance is positive; when the voxel is located in the negative direction of the geological interface ( ), the distance is negative; when the voxel is located on the geological interface, the distance is 0.

[0056] When there are multiple geological interfaces in the area to be studied, the level set function needs to be modeled for each complex geological interface separately to obtain multiple initial voxel level set functions, and finally form a matrix, where S is the number of geological interfaces.

[0057] S130: Based on each initial voxel level set function, perform forward gravity anomaly calculation to obtain theoretical gravity anomaly information.

[0058] Forward gravity anomaly calculation is an important part of 3D geological modeling and geological structure optimization. It refers to the process of predicting the gravity anomaly generated by a geological body at a certain point on the ground or in space through theoretical calculations based on parameters such as the shape, size, density, and spatial position of the underground geological body, that is, calculating the theoretically generated gravity change according to the constructed underground geological structure model.

[0059] The principle of forward gravity anomaly calculation is Newton's law of universal gravitation. According to Newton's law of universal gravitation, the gravitational force between two point masses is shown in formula (3): where is the gravitational constant; and are the masses of the two point masses respectively; is the distance between the two point masses.

[0060] For a geological body with a certain volume, it can be regarded as a set composed of countless tiny point masses. When calculating the gravity field generated by the geological body, it is necessary to integrate the gravitational forces generated by these tiny point masses.

[0061] Assume that the density of a geological body is , and the volume is , then the mass of the geological body is ; in a rectangular coordinate system, for any point on the ground, the vertical component of the gravity (i.e., the gravity anomaly value) generated by the geological body can be calculated by integral formula (4): where are the coordinates of the point mass inside the geological body.

[0062] Assume that the side length of a voxel is 1, the density corresponding to this voxel is , the central coordinate of this voxel is , and the coordinate of the observation point is , then according to Newton's law of universal gravitation, the vertical component of the gravity (i.e., the gravity anomaly value) generated by this single voxel at the observation point can be calculated by formula (5): For the initial geological voxel model composed of voxels, the total gravity anomaly value (i.e., the theoretical gravity anomaly information) is the sum of the vertical components of the gravity generated by all voxels at the observation point, that is, as shown in formula (6): Optionally, in actual calculations, based on each initial voxel level set function and the prior knowledge of the area to be studied, the lithology of each three-dimensional voxel in the area to be studied can be assigned, and a density value can be assigned to each voxel according to the existing lithology characteristics to support the forward gravity anomaly calculation.

[0063] Optionally, since the number of voxels in the initial geological voxel model is very large, a loop structure can be used through programming to traverse all voxels, perform forward gravity anomaly calculations, calculate the contribution of each voxel and accumulate it, and finally obtain the theoretical gravity anomaly information of the initial geological voxel model.

[0064] S140: Optimize each initial voxel level set function based on the genetic algorithm to obtain a coarse-resolution geological voxel model.

[0065] Since in the process of forward gravity anomaly calculation, it is necessary to perform forward gravity calculation on each voxel value, the calculation amount is extremely large, and when the number of geological interfaces in the area to be studied increases, the calculation complexity will increase exponentially. Therefore, in this embodiment, the voxels are further optimized by coarse resolution (such as kilometer-level resolution) to reduce the number of voxels and improve the calculation efficiency.

[0066] Specifically, use the genetic algorithm to optimize each initial voxel level set function with low spatial resolution in the area to be studied to obtain a coarse-resolution geological voxel model.

[0067] S150: Optimize the coarse-resolution geological voxel model with fine resolution based on the actual gravity anomaly information and the theoretical gravity anomaly information to obtain the target geological model.

[0068] After rapid optimization by the genetic algorithm, a coarse-resolution geological voxel model can be obtained. The coarse-resolution geological voxel model can reasonably reflect the spatial distribution of multiple geological interfaces. However, due to its low spatial resolution, it is not sufficient to support work with high requirements for model resolution such as minefield exploration. Therefore, it is necessary to further optimize the coarse-resolution geological voxel model with fine resolution (such as 50m to 100m) using the actual gravity anomaly information and the theoretical gravity anomaly information to obtain the final target geological model.

[0069] The modeling optimization method provided in this embodiment, after constructing an initial geological voxel model based on the geological information of the area to be studied, performs level set function modeling on the initial geological voxel model to obtain multiple initial voxel level set functions, then uses the genetic algorithm to perform coarse-resolution optimization on each initial voxel level set function to obtain a coarse-resolution geological voxel model, and finally uses the actual gravity anomaly information and the theoretical gravity anomaly information obtained through gravity anomaly forward modeling to perform fine-resolution optimization on the coarse-resolution geological voxel model to obtain the target geological model. In the above manner, after obtaining the initial geological voxel model, first perform coarse-resolution optimization through level set function modeling and the genetic algorithm, and then perform fine-resolution optimization through the theoretical gravity anomaly information obtained through gravity anomaly forward modeling, which can effectively improve the resolution of the model, ensure the accuracy and reliability of the target geological model, and enable the target geological model to accurately reflect the true geological structure of the area to be studied.

[0070] In some embodiments, based on the genetic algorithm, optimizing each initial voxel level set function to obtain a coarse-resolution geological voxel model includes: generating an initial population of the genetic algorithm; the initial population includes multiple individuals, and one individual is an initial voxel level set function; based on the elitist retention strategy and the tournament selection strategy, selecting multiple individuals from the initial population as parent individuals; performing population crossover operations on the multiple parent individuals to obtain a parent population; performing Cauchy mutation and Gaussian mutation on the parent population to obtain multiple offspring individuals; performing population update based on the multiple offspring individuals to obtain an updated population; the updated population includes multiple updated individuals, and one updated individual is an optimized voxel level set function; based on the updated population, determining whether a preset termination condition is satisfied; if the preset termination condition is not satisfied, then return to the step of selecting multiple individuals from the initial population as parent individuals based on the elitist retention strategy and the tournament selection strategy until the preset termination condition is satisfied, and based on the updated population, generating a coarse-resolution geological voxel model.

[0071] Before performing optimization using the genetic algorithm, to ensure the rationality of the geological voxel model, parameter constraint conditions need to be preset: the depth range of the geological voxel model is limited to 100 to 5000 meters, and the density difference is controlled within 1.0 g / cm³ to 3.0 g / cm³ to conform to the regional tectonic characteristics. It can be understood that the input parameters of the genetic algorithm are all the initial voxel level set functions within the area to be studied. For the case where there may be N geological bodies within the area to be studied, all parameters need to be concatenated to form a high-dimensional solution space. For example, the number of parameters of a double-geological-body model is Q×N, where Q is the number of voxels included in a single initial voxel level set function.

[0072] In the specific calculation process, the genetic algorithm indirectly updates the geometric shape of the geological body in the area to be studied by adjusting all the initial voxel level set functions in the area to be studied, thereby affecting the lithology and density of all voxels in the area to be studied, and further affecting the simulated gravity anomaly value in the area to be studied.

[0073] Understandably, before using the genetic algorithm for optimization, it is also necessary to design an error evaluation system and establish a multi-objective evaluation function with the gravity anomaly fitting degree as the core. In this embodiment, the population fitness is used as the evaluation index, and the population fitness includes the following error terms: (1) Data fitting term: Calculate the theoretical gravity anomaly information of the initial geological voxel model through the three-dimensional gravity forward algorithm, and use the improved Tikhonov regularization method to calculate the root mean square error (RMSE) between the actual gravity anomaly information and the theoretical gravity anomaly information, as shown in formula (7): Among them, is the root mean square error between the actual gravity anomaly information and the theoretical gravity anomaly information; is the weight parameter used to adjust the weights of different error terms; is the number of observed gravity anomaly values; is the th theoretical gravity anomaly value in the theoretical gravity anomaly information; is the th actual gravity anomaly value in the actual gravity anomaly information.

[0074] Optionally, the Gauss-Newton iteration method is introduced to accelerate the forward calculation to ensure that the forward calculation time of a single initial geological voxel model is controlled within 5 seconds.

[0075] (2) Model complexity penalty term: Calculate the curvature change rate and density jump gradient between adjacent geological bodies, and impose a quadratic penalty on the cases of morphological mutations or drastic changes in physical properties to avoid non-physical oscillation solutions. The expression of the model complexity penalty term is shown in formula (8): Among them, is the model complexity penalty value; is the regularization coefficient used to adjust the weights of different error terms; represents the square of the gradient norm of the initial voxel level set function, that is, the square of the norm of the three-dimensional space gradient of the initial voxel level set function of the th geological body; is the absolute value of the change in the density difference between adjacent geological bodies.

[0076] (3) Prior knowledge fusion term: If there is borehole data or seismic interpretation results in the prior data, a spatial distance constraint function can be set to impose an exponential penalty on the solution that deviates from the known control points. The expression of the prior knowledge fusion term is shown in Equation (9): where, is the prior knowledge fusion value; is the weight parameter used to adjust the weights of different error terms; is the lithology result of the th geological body interpreted by the known borehole data or seismic interpretation results (i.e., the actual lithology result); is the lithology result of the th geological body optimized by the genetic algorithm.

[0077] By weighted summation of the three error terms, the three error terms can be combined into a fitness value , as shown in Equation (10): where, .

[0078] Further, the population fitness optimized by the genetic algorithm is as shown in Equation (11): where, is a preset coefficient used to control the decay of individual age; is the individual age.

[0079] Optionally, .

[0080] It can be understood that before using the genetic algorithm for optimization, various hyperparameters of the genetic algorithm need to be set: (1) Population initialization: The initial population is generated by Latin hypercube sampling to ensure uniform coverage of the parameter space. For example, the initial population size is set to 50 - 200, and the initial population size can be dynamically adjusted according to the parameter dimension.

[0081] (2) Selection mechanism: Combining the elitist retention strategy (retaining the top 5% of the fittest individuals) and the tournament selection strategy, the selection pressure coefficient gradually decays from 2.0 to 1.2 to balance exploration and exploitation.

[0082] (3) Crossover operation: For geometric parameters, simulated binary crossover (SBX) is used, with the distribution index η = 15; for physical property parameters, arithmetic crossover is used, and the mixing ratio α randomly varies within the range of 0.4 to 0.6; an adaptive crossover probability is set, with pc = 0.9 in the initial stage and decreasing to 0.7 in the later stage.

[0083] (4) Mutation strategy: Use a hierarchical mutation mechanism, which includes global mutation and local fine-tuning; global mutation means performing Cauchy perturbation on random parameters with a probability of 5%, and the scale parameter γ = 0.1; local fine-tuning means performing Gaussian mutation (σ = 2% of the parameter range) on the selected individuals, and the probability linearly decreases from 0.2 to 0.05 with the number of iterations.

[0084] (5) Constraint handling: Use the projection repair method to handle out-of-bounds parameters, reset the out-of-bounds values to the nearest feasible boundary, and apply an additional 10% penalty at the same time.

[0085] After the above preparations are completed, the genetic algorithm can be used for optimization. Please refer to Figure 2 , Figure 2 which is a schematic diagram of the optimization process of the genetic algorithm provided by the present invention.

[0086] As Figure 2 shown, in this embodiment, after setting the hyperparameters of the genetic algorithm, an initial population of the genetic algorithm is generated. The initial population includes multiple individuals, and one individual is an initial voxel level set function; according to the population fitness of the initial population, it is judged whether the preset termination condition is satisfied.

[0087] If the preset termination condition is not satisfied, perform a population selection operation on the initial population: Based on the elite retention strategy and the tournament selection strategy, select multiple individuals from the initial population as parent individuals.

[0088] Specifically, according to the elite retention strategy, perform elite retention on the initial population, directly copy the individuals with the top 5% fitness rankings as parent individuals; at the same time, apply the tournament selection strategy, randomly select N individuals from the initial population, and select the individual with the highest fitness from the N individuals as a parent individual; repeat this selection process until the sum of the number of parent individuals of the elite retention strategy and the number of parent individuals of the tournament selection strategy reaches the number of parent individuals to be retained.

[0089] Optionally, the value of N can be calculated by formula (12): where, is the number of iterations; is the number of parent individuals to be retained.

[0090] Further, perform a population crossover operation on multiple parent individuals to obtain a parent population.

[0091] Optionally, use simulated binary crossover (SBX) for the geometric parameters of the parent individuals, with the distribution index η = 15; use arithmetic crossover for the heat physical property parameters of the parent individuals, and the mixing ratio Randomly vary within the range of 0.4 to 0.6.

[0092] Optionally, the expression of the arithmetic crossover operation is shown in Formula (13): where is the crossover rate (i.e., the mixing ratio).

[0093] Optionally, the expression of the crossover rate is shown in Formula (14): where is the maximum number of iterations.

[0094] Furthermore, perform Cauchy mutation and Gaussian mutation on the parental population to obtain multiple offspring individuals.

[0095] Specifically, first perform Cauchy mutation on the parental population for global exploration. The mutation formula of Cauchy mutation is shown in Formula (15): where is the Cauchy mutation amplitude constant, fixed at to balance the perturbation intensity.

[0096] After completing the Cauchy mutation, perform Gaussian mutation for local exploitation. The mutation formula of Gaussian mutation is shown in Formula (16): where is the standard deviation of the Gaussian distribution, proportional to the domain (adaptability).

[0097] Optionally, the triggering probability of Cauchy mutation is 5%, and the triggering probability of Gaussian mutation is 0.15×(1−t / maxt), where t is the current iteration number and maxt is the total number of iterations.

[0098] In addition, when performing Cauchy mutation and Gaussian mutation, boundary checking is required, and reflection correction is performed on the out-of-bounds parameters. The correction formula is shown in Formula (17): where and are the parameter constraint conditions.

[0099] Furthermore, perform population update based on multiple offspring individuals to obtain the updated population.

[0100] Specifically, the update is achieved by combining multiple offspring individuals and the elite individuals retained by the elite retention strategy to obtain the updated population; at the same time, implement age increment, that is , and remove duplicate individuals in the updated population. If the parameter difference between any two individuals is less than , then randomly perturb one of them.

[0101] Among them, the updated population includes multiple updated individuals, and an updated individual is an optimized voxel level set function.

[0102] Furthermore, based on the updated population, determine whether the preset termination condition is satisfied.

[0103] Specifically, through the following three convergence conditions, determine whether to jump out of the loop: (1) Absolute convergence criterion: Monitor the relative change rate of the fitness of the optimal individual in the updated population to determine whether the genetic algorithm has entered a stagnant state; when the improvement amplitude of the optimal solution of the population for 20 consecutive generations is less than 0.1%, it is considered that the updated population has converged, the preset termination condition is satisfied, and the loop can be jumped out.

[0104] (2) Relative convergence criterion: Measure the average change amplitude of the parameters of the entire updated population to prevent the genetic algorithm from prematurely terminating at local extrema; when the average moving distance of the updated population is less than 1% of the parameter space dimension D, it is considered that the updated population has converged, the preset termination condition is satisfied, and the loop can be jumped out.

[0105] (3) Maximum iteration limit: Set the maximum number of iterations, evaluate the convergence trend of the updated population every 50 generations. If the fitness is still significantly improving (for example, the growth rate per generation > 0.5%), it is considered that the updated population has not converged, and 50 generations will be automatically extended.

[0106] If the preset termination condition is not satisfied, it is necessary to recalculate the population fitness, return to the step of selecting multiple individuals as parent individuals from the initial population based on the elitist retention strategy and the tournament selection strategy, until the preset termination condition is satisfied, and generate a coarse-resolution geological voxel model based on the updated population.

[0107] Optionally, adopt a parallel architecture, such as a master-slave MPI parallel architecture, divide the population into multiple subgroups for distributed computing on different computing nodes, and design a dynamic task scheduling algorithm to automatically balance the computing workload according to the node load to accelerate the iteration speed of the genetic algorithm.

[0108] Optionally, use the CUDA architecture to parallelize the core algorithm of gravity forward modeling, and decompose the three-dimensional integral calculation into 4096 thread blocks.

[0109] In some embodiments, an optimized voxel level set function includes multiple parameters; based on the updated population, a coarse-resolution geological voxel model is generated, including: determining the population fitness of the updated population; based on the contribution degree of each optimized voxel level set function to the population fitness, freezing partial parameters of each optimized voxel level set function respectively to obtain multiple voxel level set functions with reduced dimensions; and based on the Markov chain Monte Carlo method, performing uncertainty quantification on the multiple voxel level set functions with reduced dimensions to generate a coarse-resolution geological voxel model.

[0110] Specifically, after obtaining the updated population, determine the population fitness of the updated population and perform parameter sensitivity analysis: for each optimized voxel level set function, the contribution degree of this optimized voxel level set function to the population fitness can be calculated through the mutation-fitness correlation coefficient of its parameters, identify the key control variables, and freeze partial secondary parameters of this optimized voxel level set function according to the contribution degree to obtain multiple voxel level set functions with reduced dimensions, so as to reduce the subsequent calculation dimension. This process is shown in formula (18): Among them, are the parameters of the optimized voxel level set function; is the population fitness.

[0111] Furthermore, uncertainty propagation is realized: based on the Markov chain Monte Carlo (MCMC) method, uncertainty quantification is performed on the multiple voxel level set functions with reduced dimensions to generate a coarse-resolution geological voxel model.

[0112] Specifically, the Markov chain Monte Carlo method is used to sample in the neighborhood of the optimal solution, construct the posterior probability distribution of the parameters, calculate the anisotropic variation coefficient, draw the 95% confidence interval histogram, and generate a three-dimensional coarse-resolution geological voxel model to achieve visual output.

[0113] Among them, the coarse-resolution geological voxel model is a three-dimensional geological structure probability body, and the uncertainty of the interface position is characterized by different color mappings.

[0114] In some embodiments, the population fitness is determined based on the fitness value, the individual age, and a preset coefficient for controlling the attenuation of the individual age; among them, the fitness value is determined based on the model complexity penalty value, the prior knowledge fusion value, and the root mean square error of the actual gravity anomaly information and the theoretical gravity anomaly information.

[0115] In some embodiments, based on the actual gravity anomaly information and the theoretical gravity anomaly information, the coarse-resolution geological voxel model is optimized at a fine resolution to obtain the target geological model, including: determining the gravity anomaly error based on the actual gravity anomaly information and the theoretical gravity anomaly information; determining the voxel change error based on each initial voxel level set function and each optimized voxel level set function; performing a weighted aggregation operation on the gravity anomaly error and the voxel change error to obtain the total error; respectively determining the error gradient of the total error with respect to the variables of each optimized voxel level set function; and optimizing the coarse-resolution geological voxel model at a fine resolution based on each error gradient to obtain the target geological model.

[0116] After rapid optimization of the coarse resolution through a genetic algorithm, a coarse-resolution geological voxel model is obtained. The coarse-resolution geological voxel model can relatively reasonably reflect the spatial distribution of multiple geological interfaces. However, due to its low spatial resolution, it is not sufficient to support work with high requirements for model resolution such as minefield exploration. Therefore, it is necessary to further optimize the coarse-resolution geological voxel model at a fine resolution (e.g., 50m to 100m) using the actual gravity anomaly information and the theoretical gravity anomaly information to obtain the final target geological model.

[0117] In 3D geological modeling, a reliable shallow geological body model can be constructed based on actual exploration data collected from boreholes and tunnels. However, for the deep structure of the geological body, the actual exploration geological data becomes sparse or even unavailable, making it quite difficult to construct an accurate model. Therefore, this embodiment adopts a compromise method. Based on the use of the shallow model, data related to the deep structure, such as geophysical exploration data, is used to extrapolate the shallow structure to the deep to estimate the overall architecture of the deep structure.

[0118] Specifically, in this embodiment, the gravity anomaly information is used to further optimize the extrapolation model to improve the reliability and resolution of the model.

[0119] Please refer to Figure 3 , Figure 3 which is a schematic flow diagram of the fine-resolution optimization provided by the present invention.

[0120] As Figure 3 shown, first, based on the actual gravity anomaly information and the theoretical gravity anomaly information, the gravity anomaly error is determined. The expression of the gravity anomaly error is as shown in formula (19): where represents the theoretical gravity anomaly value calculated according to the voxel characteristics; represents the actual gravity anomaly value; the physical meanings of the remaining parameters are the same as those in formula (6).

[0121] Further, based on each initial voxel level set function and each optimized voxel level set function, a voxel change error is determined.

[0122] Specifically, the voxel change error is expressed as shown in formula (20): where is a weight function that varies with depth. Generally, this weight is larger when the depth is shallower and smaller when the depth is deeper. Since the geological structure in the shallow layer has a higher certainty, a larger weight is given to limit its change, and vice versa; is the value of the initial voxel level set function; is the value of the optimized voxel level set function in each iteration step; the physical meanings of the remaining parameters are the same as those in formula (6).

[0123] Further, a weighted aggregation operation is performed on the gravity anomaly error and the voxel change error to obtain a total error.

[0124] Specifically, the total error is expressed as shown in formula (21): where is 's weight value, is 's weight value, .

[0125] Further, the error gradient of the total error with respect to the variable of each optimized voxel level set function is determined respectively, and based on each error gradient, the coarse-resolution geological voxel model is optimized to a fine resolution to obtain the target geological model.

[0126] Specifically, after calculating the total error , the error gradient of the total error with respect to the variable of each optimized voxel level set function is calculated respectively; after obtaining the error gradient, the coarse-resolution geological voxel model is optimized to a fine resolution: the loss function is calculated using each error gradient for minimization, a new voxel level set function is updated, and the target geological model is determined according to the new voxel level set function.

[0127] where the expression of the error gradient is as shown in formula (22): where is the error gradient; is the variable matrix to be optimized in each calculation cycle; is a minimum value.

[0128] In the iterative step of the fine-resolution optimization method, the voxel level set function is updated by continuously loop-calculating and using the error gradient of each loop calculation to minimize the loss function. When the absolute value of the difference between the loss function values of two consecutive iterations is less than the iteration stops, and at this time, it is considered that the target geological model has converged to an optimal solution.

[0129] Optionally, after obtaining the target geological model, the lithology of the actual geological borehole data can also be used to verify the lithology results of the model optimization to ensure the accuracy and reliability of the model.

[0130] Optionally, if the spatial resolution of 50m - 100m is still insufficient to meet the requirements, fine-resolution optimization can be continued until the required spatial resolution is refined.

[0131] In some embodiments, based on the geological information of the area to be studied, an initial geological voxel model is constructed, including: based on the geological information of the area to be studied, surface modeling is performed on the surface of each geological body to obtain a surface model; and voxelization modeling is performed on the surface model to obtain an initial geological voxel model.

[0132] The present invention also provides a modeling optimization device for a geological model. Please refer to Figure 4 , Figure 4 which is a schematic structural diagram of the modeling optimization device for the geological model provided by the present invention. In this embodiment, the modeling optimization device for the geological model includes a first construction module 410, a second construction module 420, a gravity anomaly forward modeling module 430, a first optimization module 440, and a second optimization module 450.

[0133] The first construction module 410 is used to construct an initial geological voxel model based on the geological information of the area to be studied; the initial geological voxel model includes geological interfaces of multiple geological bodies.

[0134] The second construction module 420 is used to perform level set function modeling on each geological interface respectively to obtain a plurality of initial voxel level set functions.

[0135] The gravity anomaly forward modeling module 430 is used to perform gravity anomaly forward modeling based on each initial voxel level set function to obtain theoretical gravity anomaly information.

[0136] The first optimization module 440 is used to optimize each initial voxel level set function based on a genetic algorithm to obtain a coarse-resolution geological voxel model.

[0137] The second optimization module 450 is used to perform fine-resolution optimization on the coarse-resolution geological voxel model based on the actual gravity anomaly information and the theoretical gravity anomaly information to obtain a target geological model.

[0138] In some embodiments, the first optimization module 440 is configured to generate an initial population of a genetic algorithm; the initial population includes a plurality of individuals, and an individual is an initial voxel level set function; based on the elitist retention strategy and the tournament selection strategy, a plurality of individuals are selected from the initial population as parental individuals; a population crossover operation is performed on the plurality of parental individuals to obtain a parental population; Cauchy mutation and Gaussian mutation are performed on the parental population to obtain a plurality of offspring individuals; population update is performed based on the plurality of offspring individuals to obtain an updated population; the updated population includes a plurality of updated individuals, and an updated individual is an optimized voxel level set function; based on the updated population, it is determined whether a preset termination condition is satisfied; if the preset termination condition is not satisfied, the process returns to the step of selecting a plurality of individuals from the initial population as parental individuals based on the elitist retention strategy and the tournament selection strategy until the preset termination condition is satisfied, and based on the updated population, a coarse-resolution geological voxel model is generated.

[0139] In some embodiments, an optimized voxel level set function includes a plurality of parameters.

[0140] The first optimization module 440 is configured to determine the population fitness of the updated population; based on the contribution degree of each optimized voxel level set function to the population fitness, partial parameters of each optimized voxel level set function are respectively frozen to obtain a plurality of dimension-reduced voxel level set functions; based on the Markov chain Monte Carlo method, uncertainty quantification is performed on the plurality of dimension-reduced voxel level set functions to generate a coarse-resolution geological voxel model.

[0141] In some embodiments, the population fitness is determined based on a fitness value, an individual age, and a preset coefficient for controlling the attenuation of the individual age; wherein, the fitness value is determined based on a model complexity penalty value, a prior knowledge fusion value, and the root mean square error between the actual gravity anomaly information and the theoretical gravity anomaly information.

[0142] In some embodiments, the second optimization module 450 is configured to determine a gravity anomaly error based on the actual gravity anomaly information and the theoretical gravity anomaly information; determine a voxel change error based on each initial voxel level set function and each optimized voxel level set function; perform a weighted aggregation operation on the gravity anomaly error and the voxel change error to obtain a total error; respectively determine the error gradient of the total error with respect to the variables of each optimized voxel level set function; based on each error gradient, perform fine-resolution optimization on the coarse-resolution geological voxel model to obtain a target geological model.

[0143] In some embodiments, the first construction module 410 is configured to perform surface modeling on the surface of each geological body based on the geological information of the area to be studied to obtain a surface model; perform voxelization modeling on the surface model to obtain an initial geological voxel model.

[0144] The present invention also provides an electronic device. Figure 5 It is a schematic structural diagram of the electronic device provided by the present invention. As Figure 5 shown, the electronic device may include: a processor 510, a communications interface 520, a memory 530, and a communication bus 540. Among them, the processor 510, the communications interface 520, and the memory 530 complete communication with each other through the communication bus 540. The processor 510 may call the logic instructions in the memory 530 to execute the modeling optimization method of the geological model.

[0145] In addition, when the logic instructions in the above-mentioned memory 530 are implemented in the form of software functional units and sold or used as independent products, they may be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, may be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The foregoing storage medium includes: USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical disks, etc., which can store program codes.

[0146] The present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the modeling optimization method of the geological model provided by the above-mentioned various methods is implemented.

[0147] The present invention also provides a computer program product. The computer program product includes a computer program. The computer program can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the modeling optimization method of the geological model provided by the above-mentioned various methods.

[0148] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated. The components shown as units may or may not be physical units, that is, they may be located in one place, or may be distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art can understand and implement it without creative labor.

[0149] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on such an understanding, the essence of the above technical solution, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0150] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for optimizing the modeling of a geological model, characterized in that, Including: Construct an initial geological voxel model based on the geological information of the area to be studied; The initial geological voxel model includes geological interfaces of multiple geological bodies; Perform level set function modeling on each of the geological interfaces respectively to obtain multiple initial voxel level set functions; Based on each of the initial voxel level set functions, conduct forward gravity anomaly calculation to obtain theoretical gravity anomaly information; Based on the genetic algorithm, optimize each of the initial voxel level set functions to obtain a coarse-resolution geological voxel model; Based on the actual gravity anomaly information and the theoretical gravity anomaly information, perform fine-resolution optimization on the coarse-resolution geological voxel model to obtain the target geological model.

2. The modeling optimization method of the geological model according to claim 1, characterized in that The step of optimizing each of the initial voxel level set functions based on the genetic algorithm to obtain a coarse-resolution geological voxel model includes: Generate an initial population of the genetic algorithm; the initial population includes multiple individuals, and one individual is one of the initial voxel level set functions; Based on the elitist retention strategy and the tournament selection strategy, select multiple individuals from the initial population as parent individuals; Perform population crossover operation on the multiple parent individuals to obtain a parent population; Perform Cauchy mutation and Gaussian mutation on the parent population to obtain multiple offspring individuals; Based on the multiple offspring individuals, perform population update to obtain an updated population; the updated population includes multiple updated individuals, and one updated individual is an optimized voxel level set function; Based on the updated population, determine whether the preset termination condition is satisfied; If the preset termination condition is not satisfied, return to the step of selecting multiple individuals from the initial population as parent individuals based on the elitist retention strategy and the tournament selection strategy until the preset termination condition is satisfied, and generate a coarse-resolution geological voxel model based on the updated population.

3. The modeling optimization method of the geological model according to claim 2, characterized in that One optimized voxel level set function includes multiple parameters; The step of generating a coarse-resolution geological voxel model based on the updated population includes: Determine the population fitness of the updated population; Based on the contribution degree of each optimized voxel level set function to the population fitness, freeze part of the parameters of each optimized voxel level set function respectively to obtain multiple dimension-reduced voxel level set functions; Based on the Markov chain Monte Carlo method, perform uncertainty quantification on the multiple dimension-reduced voxel level set functions to generate the coarse-resolution geological voxel model.

4. The modeling optimization method of the geological model according to claim 3, wherein The population fitness is determined based on the fitness value, the individual age, and a preset coefficient for controlling the attenuation of the individual age; Wherein, the fitness value is determined based on the model complexity penalty value, the prior knowledge fusion value, and the root mean square error between the actual gravity anomaly information and the theoretical gravity anomaly information.

5. The modeling optimization method of the geological model according to claim 2, characterized in that, The step of performing fine-resolution optimization on the coarse-resolution geological voxel model based on the actual gravity anomaly information and the theoretical gravity anomaly information to obtain the target geological model includes: Based on the actual gravity anomaly information and the theoretical gravity anomaly information, determine the gravity anomaly error; Determine a voxel change error based on each of the initial voxel level set functions and each of the optimized voxel level set functions; Perform a weighted aggregation operation on the gravity anomaly error and the voxel change error to obtain a total error; Respectively determine the error gradient of the total error with respect to the variables of each of the optimized voxel level set functions; Based on each of the error gradients, perform a fine-resolution optimization on the coarse-resolution geological voxel model to obtain the target geological model.

6. The modeling optimization method of the geological model according to claim 1, wherein The constructing an initial geological voxel model based on the geological information of the area to be studied includes: Perform surface modeling on the surface of each of the geological bodies based on the geological information of the area to be studied to obtain a surface model; Perform voxelization modeling on the surface model to obtain the initial geological voxel model.

7. A modeling optimization device for a geological model, characterized in that Includes: A first construction module for constructing an initial geological voxel model based on the geological information of the area to be studied; The initial geological voxel model includes geological interfaces of a plurality of geological bodies; A second construction module for performing level set function modeling on each of the geological interfaces respectively to obtain a plurality of initial voxel level set functions; A gravity anomaly forward modeling module for performing gravity anomaly forward modeling based on each of the initial voxel level set functions to obtain theoretical gravity anomaly information; A first optimization module for optimizing each of the initial voxel level set functions based on a genetic algorithm to obtain a coarse-resolution geological voxel model; A second optimization module for performing a fine-resolution optimization on the coarse-resolution geological voxel model based on the actual gravity anomaly information and the theoretical gravity anomaly information to obtain the target geological model.

8. An electronic device, comprising a memory, a processor, and a computer program stored on the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the modeling optimization method of the geological model according to any one of claims 1 to 6.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the modeling optimization method of the geological model according to any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the modeling optimization method of the geological model according to any one of claims 1 to 6.

Images