CN114943178A - Three-dimensional geological model modeling method and device and computer equipment - Google Patents
Three-dimensional geological model modeling method and device and computer equipment Download PDFInfo
- Publication number
- CN114943178A CN114943178A CN202210545188.2A CN202210545188A CN114943178A CN 114943178 A CN114943178 A CN 114943178A CN 202210545188 A CN202210545188 A CN 202210545188A CN 114943178 A CN114943178 A CN 114943178A
- Authority
- CN
- China
- Prior art keywords
- stratum
- points
- dimensional geological
- dimensional
- geological model
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T17/00—Three dimensional [3D] modelling, e.g. data description of 3D objects
Abstract
The invention discloses a three-dimensional geological model modeling method, which comprises the following steps: the method comprises the steps of obtaining drilling data of a plurality of drilling position points of a research area, conducting resampling and normalization processing on the drilling data, utilizing a trained stacking machine learning model to predict the research area to obtain a three-dimensional geological model, obtaining a geological data database of the research area, utilizing the thickness and the occurrence of a rock stratum to conduct precision evaluation on the three-dimensional geological model, and conducting iterative correction on the three-dimensional geological model. The invention integrates the stacking machine learning algorithm, simultaneously integrates the geoscience knowledge acquired from the geological data database, can acquire a model result meeting the requirement of the geoscience knowledge through iterative correction, and the result shows that the classification effect of the integrated stacking machine learning algorithm on the stratum is better than that of a single classifier, and the three-dimensional geological model obtained through iterative correction of the geoscience knowledge database has higher precision and the detail expression capability of the model is enhanced.
Description
Technical Field
The invention relates to the technical field of three-dimensional geological modeling, in particular to a three-dimensional geological model modeling method, a three-dimensional geological model modeling device and computer equipment.
Background
With the progress of scientific technology and the development of human society, the need for underground land and space has become a necessary trend for the historical development of cities. The development of regional geological survey, urban geological exploration, geological environment evaluation, urban underground space development, safety utilization and other works cannot be separated from the three-dimensional geological model.
Because the three-dimensional geological model is mostly based on the simulation result obtained by performing feature deduction and spatial interpolation on structural data such as drilling holes, profiles and the like, the data are always sparse for complex geological modeling tasks. Therefore, in the modeling process, geological modeling personnel usually need to use a large amount of experience knowledge to adjust the trend of a geological surface, the thickness of a geological body and the relationship among the geological bodies; in addition, in the conventional three-dimensional geological modeling method, a modeling framework integrating modeling and model correction is lacked, and the geological modeling correction process is mostly corrected once; the three-dimensional geological model has low precision and insufficient detail expression.
Disclosure of Invention
In order to solve the above problems, an embodiment of the present invention provides a three-dimensional geological model modeling method, which specifically includes the following steps:
step S1: acquiring drilling data of a plurality of drilling position points of a research area, wherein the drilling data comprises three-dimensional coordinates of a stratum boundary and stratum attributes;
step S2: resampling the drilling data of each drilling position point to enable the stratum boundaries of each stratum to have the same number of interpolation points, obtaining three-dimensional coordinates and stratum attributes of the interpolation points, and performing normalization processing on the stratum boundaries and the three-dimensional coordinates of the interpolation points;
step S3: training a stacking machine learning model by using the normalized three-dimensional coordinates of the stratum boundary and the interpolation point and the stratum attributes of the stratum boundary and the interpolation point, and predicting the stratum attribute of each grid point on a gridding three-dimensional geological modeling area of the research area by using the trained stacking machine learning model to obtain a three-dimensional geological model;
step S4: obtaining a geological data database of a research area, and obtaining the thickness and the attitude of each rock stratum at a plurality of position points in the research area by the geological data database;
step S5: performing precision evaluation on the three-dimensional geological model by using the thickness and the attitude of the rock stratum, screening out partial position points of the three-dimensional geological model, which do not meet the requirements on the modeling precision of a research area, as error position points and partial position points which meet the requirements as non-error position points according to the precision evaluation result, wherein the error position points and the non-error position points are grid points on the top surface of a gridded three-dimensional geological modeling area;
step S6: for each error position point, selecting a set number of multiple non-error position points in a 360-degree direction along a direction gradually far away from the error position point, obtaining boundary grid points of corresponding stratum boundaries under the multiple non-error position points in the three-dimensional geological model, interpolating three-dimensional coordinates of the boundary grid points with the same stratum attributes by using a spline surface fitting interpolation algorithm, calculating to obtain corrected boundary grid points of the corresponding stratum boundaries under the error position point, and correcting the stratum attributes of grids under the error position point in the three-dimensional geological model according to the corrected boundary grid points to obtain a corrected three-dimensional geological model;
step S7: and returning to the step S5 to perform iterative correction on the three-dimensional geological model until the precision evaluation result of the corrected three-dimensional geological model meets the set requirement, stopping iterative correction, and outputting the three-dimensional geological model.
Further, in step S2, the normalized formula is:
P * ＝(P-μ)/σ
wherein P represents coordinate data of stratum boundaries or interpolation points in different dimensions, mu represents the mean value of the coordinate data of all the stratum boundaries and the interpolation points in different dimensions, sigma represents the standard deviation of the coordinate data of all the stratum boundaries and the interpolation points in different dimensions, and P represents the standard deviation of the coordinate data of all the stratum boundaries and the interpolation points in different dimensions * Representing different dimensions of normalized stratigraphic boundaries or interpolation pointsCoordinate data in degrees.
Further, the step S3 includes the following steps:
step S31: selecting a land parcel with a set thickness under an external rectangle of the research area, and carrying out grid division according to set density in the three-dimensional direction to obtain a grid three-dimensional geological modeling area;
step S32: dividing an original data set consisting of the normalized three-dimensional coordinates of the stratum boundary and the interpolation point and the corresponding stratum attributes of the stratum boundary and the interpolation point into a training set and a testing set according to a set proportion;
step S33: dividing the training set into 5 parts, during each training, taking 1 part as a test set and the other 4 parts as the training set without repetition, and obtaining 5 classifiers according to the classification method;
step S34: for each classification method in the step S33, predicting the extracted test set by using 5 classifiers, and vertically splicing each prediction result in sequence as a feature of a meta classifier; predicting the test set divided in the step S32, and obtaining another characteristic of the meta classifier based on the prediction result;
step S35: two features provided by each classifier can be obtained in step S34, where the 3 base classifiers total 6 features, the 6 features are used as training features of the meta classifier, the corresponding real class is used as a training label, and the test set of the meta classifier is the real label corresponding to the prediction result of the test set partitioned in step S32 in step S34;
step S36: training the meta classifier by using the training set and the test set obtained in the step S35 to obtain a final stacking machine learning model;
step S37: and predicting each grid central point of the gridded three-dimensional geological modeling area by using the trained stacking machine learning model to obtain the stratum attribute of each point to be predicted.
Preferably, the step S5 includes the steps of:
step S51: calculating the geological data database by using the three-dimensional geological modelThe thickness of each rock stratum at a plurality of position points is calculated according to the thickness of the rock stratum calculated by the three-dimensional geological model and the thickness of the rock stratum in the geological data database to obtain a layered thickness error Q and a thickness average relative error D at the plurality of position points MAE Error Q of layered thickness and average relative error D of thickness MAE The calculation formula is as follows:
wherein n is the number of rock layers, D mi Thickness, D, of each formation at each location point calculated for the three-dimensional geological model bi A thickness for each formation at each location in the geological data repository;
step S52: generating a stratum thickness error contour map of a research area on the top surface of a gridding three-dimensional geological modeling area according to the layered thickness errors Q at a plurality of position points by using a Krigin interpolation method in ArcGIS, averagely dividing the errors of the research area into five intervals according to the size values of the errors on the error contour map by using a natural discontinuity point method, taking grid points in a part of areas on the top surface corresponding to two intervals with the largest errors as error position points, and taking grid points in the part of areas on the top surface corresponding to the other three intervals as non-error position points;
step S53: calculating the actual horizontal distance S of the projection of the highest point of each rock stratum and the bottom point of each rock stratum in the research area on the horizontal plane according to the rock stratum attitude in the geological data database i True horizontal distance S i The calculation formula is as follows:
S i ＝L i sina i
wherein L is i Representing the length of the formation, a i Representing a formation dip angle;
using said three-dimensional groundThe texture model calculates a predicted horizontal distance S 'of the projection of the highest point of each rock stratum and the bottom point of each rock stratum in the research area on a horizontal plane' i According to said actual horizontal distance S i And the predicted horizontal distance S' i Calculating to obtain the average relative error S of the distance MAE Mean relative error of distance S MAE The calculation formula is as follows:
step S54: averaging the thickness with a relative error D MAE And said distance average relative error S MAE The average relative error MAE is taken as the average relative error MAE of the whole study area, and the calculation formula of the average relative error MAE is as follows:
preferably, in step S6, the spline surface fitting interpolation algorithm selects a bi-cubic spline interpolation algorithm, and the set number of the corresponding non-error position points is 16.
Preferably, the step S7 includes the steps of:
and returning to the step S51 to perform iterative correction on the three-dimensional geological model, stopping iterative correction when the average relative error MAE is smaller than a set threshold value, and outputting the three-dimensional geological model.
Preferably, the three-dimensional geological model is modified 3 times or so.
Preferably, the following steps are further included after step S3 and step S7, respectively: and visualizing the three-dimensional geological model.
An embodiment of the present invention further provides a three-dimensional geological model modeling apparatus, including:
a first data acquisition module for acquiring borehole data for a plurality of borehole location points of a study area, the borehole data including three-dimensional coordinates of a formation boundary and formation properties;
the data processing module is used for resampling the drilling data of each drilling position point to enable the stratum boundaries of each stratum to have the same number of interpolation points, obtaining the three-dimensional coordinates and the stratum attributes of the interpolation points, and performing normalization processing on the stratum boundaries and the three-dimensional coordinates of the interpolation points;
the machine learning model prediction module is used for training a stacking machine learning model by using the normalized three-dimensional coordinates of the stratum boundary and the interpolation point and the stratum attributes of the stratum boundary and the interpolation point, and predicting the stratum attribute of each grid point on a gridding three-dimensional geological modeling area of the research area by using the trained stacking machine learning model to obtain a three-dimensional geological model;
the second data acquisition module is used for acquiring a geological data database of the research area, and the thickness and the attitude of each rock stratum at a plurality of position points in the research area are obtained from the geological data database;
the precision evaluation module is used for carrying out precision evaluation on the three-dimensional geological model by utilizing the thickness and the attitude of the rock stratum, screening out partial position points of the three-dimensional geological model which do not meet the requirements on the modeling precision of a research area as error position points and partial position points which meet the requirements as non-error position points according to the precision evaluation result, wherein the error position points and the non-error position points are grid points on the top surface of a gridded three-dimensional geological modeling area;
the correction module is used for selecting a set number of non-error position points in the direction gradually far away from the error position points in the 360-degree direction for each error position point, obtaining boundary grid points of corresponding stratum boundaries in the three-dimensional geological model, located right below the non-error position points, interpolating three-dimensional coordinates of the boundary grid points with the same stratum attributes by using a spline surface fitting interpolation algorithm, calculating to obtain corrected boundary grid points of the corresponding stratum boundaries right below the error position points, and correcting the stratum attributes of grids located right below the error position points in the three-dimensional geological model according to the corrected boundary grid points to obtain a corrected three-dimensional geological model;
and the iterative output module is used for iteratively correcting the three-dimensional geological model until the precision evaluation result of the corrected three-dimensional geological model meets the set requirement, stopping iterative correction and outputting the three-dimensional geological model.
Embodiments of the present invention also provide a computer device comprising a processor and a memory, a three-dimensional geological model modeling program stored on the memory and executable on the processor, the modeling program, when executed by the processor, implementing the steps of the modeling method as described above.
Compared with the prior art, the invention has the beneficial effects that: the method integrates the stacking machine learning algorithm, simultaneously integrates the geoscience knowledge acquired from the geological data database, can acquire a model result meeting the requirement of the geoscience knowledge through iterative correction, and the result shows that the classification effect of the integrated stacking machine learning algorithm on the stratum is better than that of a single classifier, and the three-dimensional geological model obtained through iterative correction of the geoscience knowledge database has higher precision and the detail expression capability of the model is enhanced.
Drawings
FIG. 1 is a flow chart of a modeling method of the present invention;
FIG. 2 is a schematic diagram of the resampling principle of the modeling method of the present invention;
FIG. 3 is a schematic diagram of the principle of the stacking machine learning model training of the modeling method of the present invention;
FIG. 4 is a schematic diagram of a three-dimensional geological model predicted by a stacking machine learning model of the modeling method of the present invention;
FIG. 5 is an error contour plot of the modeling method of the present invention;
FIG. 6 is a schematic diagram of a method of obtaining a plurality of non-error location points for the modeling method of the present invention;
FIG. 7 is a variation diagram of an error contour plot of an iterative correction process of the modeling method of the present invention;
FIG. 8 is a schematic representation of a three-dimensional geological model resulting from iterative correction of the modeling method of the present invention;
FIG. 9 is a detailed comparison of the predicted three-dimensional geological model and the three-dimensional geological model obtained by iterative correction of the modeling method of the present invention;
FIG. 10 is a schematic diagram of the structure of the modeling apparatus of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
Example 1
The invention provides a three-dimensional geological model modeling method. Referring to fig. 1, fig. 1 is a flow chart of a reconstruction method according to the present invention; the method comprises the following steps:
step S1: borehole data is acquired for a plurality of borehole location points of a study area, the borehole data including three-dimensional coordinates of formation boundaries and formation properties.
The drilling data is obtained by drilling holes into the underground by a plurality of drilling position points selected by a surveying staff on the ground of a research area, drilling different depths into one position point step by step, measuring the formation attribute arrangement of the same position point from top to bottom, and then determining the three-dimensional coordinates of different formation boundaries. As shown in fig. 2, the left bar chart in the figure is a stratigraphic structure diagram obtained by drilling at one position point, three stratums a, B and C are distributed from top to bottom from the ground point (x, y and H) to the lowest point (x, y and H3), the three-dimensional coordinates of the stratum boundary of the a and B are (x, y and H1), and the three-dimensional coordinates of the stratum boundary of B, C are (x, y and H2). In this embodiment, the attribute of the stratigraphic boundary is marked as the attribute of the stratigraphic layer above the stratigraphic boundary.
Step S2: and resampling the drilling data of each drilling position point to enable the stratum boundaries of each stratum to have the same number of interpolation points, obtaining the three-dimensional coordinates and the stratum attributes of the interpolation points, and carrying out normalization processing on the stratum boundaries and the three-dimensional coordinates of the interpolation points.
Because the thicknesses of different stratums are different, the numerical value levels of different dimensional coordinates of three-dimensional coordinates of sampling position points are greatly different, particularly, the coordinates in the depth direction are much larger than the magnitude of the coordinates in the transverse direction and the longitudinal direction, in order to ensure the accuracy of machine learning, the drilling data of each drilling position point needs to be resampled, the same number of data points of different stratums are ensured, the three-dimensional coordinate data of all the data points are normalized, and the magnitude of the three-dimensional coordinates of all the data points is ensured to be close.
As shown in fig. 2, a plurality of points (x, y, H1 ') a, (x, y, H2') a …, etc. are inserted between the ground point (x, y, H) and the formation boundary point (x, y, H1), a plurality of points (x, y, H3 ') a, (x, y, H4') B …, etc. are inserted between the formation boundary point (x, y, H1) and the formation boundary point (x, y, H2), a plurality of points (x, y, H5 ') a, (x, y, H6') C …, etc. are inserted between the formation boundary point (x, y, H2) and the formation boundary point (x, y, H3), ensuring that each formation has the same number of interpolation points.
Specifically, the normalized formula is as follows:
P * ＝(P-μ)/σ
wherein P represents coordinate data of stratum boundaries or interpolation points in different dimensions, mu represents the mean value of the coordinate data of all the stratum boundaries and the interpolation points in different dimensions, sigma represents the standard deviation of the coordinate data of all the stratum boundaries and the interpolation points in different dimensions, and P represents the standard deviation of the coordinate data of all the stratum boundaries and the interpolation points in different dimensions * And representing the normalized coordinate data of different dimensions of the stratum boundary or the interpolation point.
Step S3: and training a stacking machine learning model by using the normalized three-dimensional coordinates of the stratum boundary and the interpolation point and the stratum attributes of the stratum boundary and the interpolation point, and predicting the stratum attribute of each grid point on a gridding three-dimensional geological modeling area of the research area by using the trained stacking machine learning model to obtain the three-dimensional geological model.
Specifically, the method comprises the following steps:
step S31: and selecting a land parcel with a set thickness under the circumscribed rectangle of the research area to perform grid division in the three-dimensional direction according to the set density to obtain a grid three-dimensional geological modeling area. Specifically, the method is carried out on the land parcel according to the set density in the transverse direction (X direction), the longitudinal direction (Y direction) and the vertical direction (Z direction)Mesh division to obtain an N x *N y *N z Of the gridded three-dimensional geological modeling area, N x Indicating the number of grids in the transverse direction (X direction), N y Indicating the number of grids in the longitudinal direction (Y direction), N z The number of grids in the vertical direction (Z direction) is represented, the central point of each grid is used as a point to be predicted, and stratum attributes of the point to be predicted need to be predicted by using a trained stacking machine learning model.
Step S32: and dividing an original data set consisting of the normalized three-dimensional coordinates of the stratum boundary and the interpolation point and the corresponding stratum attributes of the stratum boundary and the interpolation point into a training set and a testing set according to a set proportion. Specifically, the original data set is divided into 80%: the 20% ratio is divided into training and test sets.
Step S33: the training set is divided into 5 parts, 1 part is taken as a test set and the other 4 parts are taken as the training set without repetition during each training, and 5 classifiers can be obtained according to the classification method.
Step S34: for each classification method in step S33, 5 classifiers are used to predict the extracted test set, and each prediction result (i.e., classification probability) is vertically spliced in sequence as a feature of a meta classifier; the test set divided in step S32 is also predicted, and another feature of the meta classifier is obtained based on the prediction result.
Step S35: in step S34, two features provided by each classifier are obtained, the total of 6 features of the 3 base classifiers, the 6 features are used as training features of the meta classifier, the corresponding real class is used as a training label, and the test set of the meta classifier is the real label corresponding to the prediction result of the test set partitioned in step S32 in step S34.
Step S36: and (5) training the meta classifier by using the training set and the test set obtained in the step S35 to obtain a final stacking machine learning model, as shown in fig. 3.
Step S37: and predicting each grid central point of the gridded three-dimensional geological modeling area by using the trained stacking machine learning model to obtain the stratum attribute of each point to be predicted.
Preferably, after step S3, the method further includes the following steps: and visualizing the three-dimensional geological model.
Specifically, the three-dimensional coordinates of the point to be predicted and the corresponding stratum attributes are input into GMS software to generate a visualized three-dimensional geological model, as shown in fig. 4.
Step S4: and acquiring a geological data database of the research area, and acquiring the thickness and the attitude of each rock stratum at a plurality of position points in the research area from the geological data database.
Specifically, the geological data database is formed by data extracted from geological reports and geological yearbooks, and the data comprises lithology, rock stratum thickness, attitude, topography and landform and the like.
Step S5: and carrying out precision evaluation on the three-dimensional geological model by utilizing the thickness and the attitude of the rock stratum, screening out partial position points of the three-dimensional geological model which do not meet the requirements on the modeling precision of the research area as error position points and partial position points which meet the requirements as non-error position points according to the precision evaluation result, wherein the error position points and the non-error position points are grid points on the top surface of the gridded three-dimensional geological modeling area.
Specifically, the method comprises the following steps:
step S51: calculating the thickness of each rock stratum at a plurality of positions in the geological data database by using the three-dimensional geological model, and calculating the thickness of the rock stratum calculated by the three-dimensional geological model and the thickness of the rock stratum in the geological data database to obtain the layering thickness error Q and the average relative thickness error D at the plurality of positions MAE Error Q of layered thickness, average relative error D of thickness MAE The calculation formula is as follows:
wherein n is the number of rock layers, D mi Thickness, D, of each formation at each location point calculated for the three-dimensional geological model bi Is the thickness of each formation at each location in the geological data repository.
Step S52: generating a stratum thickness error contour map of a research area on the top surface of the gridded three-dimensional geological modeling area according to the layered thickness errors Q at a plurality of position points by using a Krigin interpolation method in ArcGIS, averagely dividing the errors of the research area into five sections according to the size values of the errors on the error contour map by using a natural discontinuity point method, taking grid points in a part of the area on the top surface corresponding to two sections with the largest errors as error position points, and taking grid points in the part of the area on the top surface corresponding to the other three sections as non-error position points, as shown in FIG. 5.
Step S53: calculating the actual horizontal distance S of the projection of the highest point of each rock stratum and the bottom point of each rock stratum in the research area on the horizontal plane according to the rock stratum attitude in the geological data database i True horizontal distance S i The calculation formula is as follows:
S i ＝L i sina i
wherein L is i Denotes the length of the formation, a i Representing a formation dip angle;
calculating a predicted horizontal distance S 'projected on a horizontal plane from the highest point of each rock stratum to the bottom point of each rock stratum in a research area by using the three-dimensional geological model' i According to said actual horizontal distance S i And the predicted horizontal distance S' i Calculating to obtain the average relative error S of the distance MAE Mean relative error of distance S MAE The calculation formula is as follows:
step S54: averaging the thickness with a relative error D MAE And said distance average relative error S MAE Adding and takingAveraging is taken as the average relative error MAE of the whole study area, and the average relative error MAE is calculated as follows:
step S6: for each error position point, selecting a set number of multiple non-error position points in the 360-degree direction along the direction gradually far away from the error position point, obtaining boundary grid points of corresponding stratum boundaries in the three-dimensional geological model, located right below the multiple non-error position points, interpolating three-dimensional coordinates of the boundary grid points with the same stratum attributes by using a spline surface fitting interpolation algorithm, calculating to obtain corrected boundary grid points of the corresponding stratum boundaries right below the error position point, and correcting the stratum attributes of grids in the three-dimensional geological model, located right below the error position point, according to the corrected boundary grid points to obtain the corrected three-dimensional geological model.
As shown in fig. 6, a method of selecting a set number of a plurality of non-error position points in a direction of 360 degrees in a direction gradually away from the error position point is as follows: and connecting every three adjacent grid points in the grid points corresponding to the plurality of drilling position points on the top surface of the gridding three-dimensional geological modeling area to generate a triangular net, taking a triangle where an error position point M is located as a reference triangle, searching the adjacent triangles layer by 3 sides of the reference triangle, taking the triangle where the M point is located as the reference triangle, taking the triangle where the M point is located at L1 as a second layer triangle, taking the triangle where the M point is located at L2 as a third layer triangle at L3, taking the triangle where the fourth layer is located at L4, and obtaining a plurality of non-error position points in a set number.
The set number of the plurality of non-error position points corresponds to a spline surface fitting interpolation algorithm, preferably, a bicubic spline interpolation algorithm is selected in the embodiment, and the set number of the corresponding plurality of non-error position points is 16.
The stratum boundary is a boundary between two stratums with different attributes, three-dimensional coordinates and stratum attributes of boundary grid points can be obtained according to the three-dimensional geological model, and the stratum attributes of grid points right above the boundary grid points and right below another boundary grid point above the boundary grid points in the three-dimensional geological model are the same as the stratum attributes of the boundary grid points. And respectively acquiring boundary grid points with the same stratum attribute under the multiple non-error position points from the three-dimensional geological model, wherein for each same stratum attribute, a boundary grid point is arranged under each non-error position point, and interpolating three-dimensional coordinates of the boundary grid points with the same stratum attribute by using a spline surface fitting interpolation algorithm to obtain coordinates of the boundary points which are under the error position points and correspond to the stratum attribute. When the boundary point corresponds to a grid point, the grid point is a corrected boundary grid point corresponding to the stratum attribute; when the boundary point does not correspond to the grid point, the nearest grid point directly above the boundary point is the modified boundary grid point corresponding to the formation attribute. And endowing the stratum attributes of the grid points which are right above the corrected boundary grid point and right below the other corrected boundary grid point above the corrected boundary grid point in the three-dimensional geological model with the same stratum attributes as those of the corrected boundary grid point, and obtaining the corrected three-dimensional geological model after finishing correcting the stratum attributes of the grids right below each error position point.
Step S7: and returning to the step S5 to perform iterative correction on the three-dimensional geological model until the precision evaluation result of the corrected three-dimensional geological model meets the set requirement, stopping iterative correction, and outputting the three-dimensional geological model.
Specifically, the method comprises the following steps:
and returning to the step S51 to perform iterative correction on the three-dimensional geological model, stopping iterative correction when the average relative error MAE is smaller than a set threshold value, and outputting the three-dimensional geological model.
Preferably, the three-dimensional geological model is corrected more than 3 times. As shown in fig. 7, when the change of the formation thickness error contour line before and after 4 times of three-dimensional geological model correction is compared, the change of the formation thickness error contour line is not obvious after the three-dimensional geological model is corrected for more than 3 times.
Preferably, after step S7, the method further includes the following steps: and visualizing the three-dimensional geological model.
Specifically, the three-dimensional coordinates of the point to be predicted and the corresponding stratum attributes are input into GMS software to generate a visualized three-dimensional geological model, as shown in fig. 8. Comparing the three-dimensional geological model obtained before and after correction as shown in fig. 9, comparing local details a, B, a and b before and after correction, it can be found that the expression of the corrected three-dimensional geological model is enhanced compared with the unmodified model.
The invention provides a three-dimensional geological model modeling method, which integrates a stacking machine learning algorithm, simultaneously integrates geological knowledge acquired from a geological data database, can acquire a model result meeting the requirement of the geological knowledge through iterative correction, and shows that the classification effect of the stacking machine learning algorithm on the stratum is better than that of a single classifier, and the three-dimensional geological model obtained through iterative correction of the geological knowledge database has higher precision and the detail expression capability of the model is enhanced.
Example 2
The invention provides a three-dimensional geological model modeling device, please refer to fig. 10, fig. 10 is a schematic structural diagram of the modeling device 100 of the invention; the modeling apparatus 100 of the present invention includes:
a first data acquisition module 110 is configured to acquire borehole data for a plurality of borehole location points of a study area, the borehole data including three-dimensional coordinates of formation boundaries and formation properties.
The data processing module 120 is configured to resample the borehole data of each borehole position point so that the same number of interpolation points are located between the formation boundaries of each formation, obtain three-dimensional coordinates and formation attributes of the interpolation points, and perform normalization processing on the formation boundaries and the three-dimensional coordinates of the interpolation points.
And the machine learning model prediction module 130 is configured to train a stacking machine learning model by using the normalized three-dimensional coordinates of the stratum boundary and the interpolation point and the normalized stratum attributes of the stratum boundary and the interpolation point, and predict the stratum attribute of each grid point on the gridding three-dimensional geological modeling area of the research area by using the trained stacking machine learning model to obtain the three-dimensional geological model.
A second data acquisition module 140 is configured to acquire a geological data database of the research area, from which the thickness and the attitude of each rock formation at a plurality of locations in the research area are obtained.
And the precision evaluation module 150 is used for evaluating the precision of the three-dimensional geological model by using the thickness and the attitude of the rock stratum, screening out partial position points of the three-dimensional geological model which do not meet the modeling precision requirement of the research area as error position points and partial position points which meet the requirement as non-error position points according to the precision evaluation result, wherein the error position points and the non-error position points are grid points on the top surface of the gridded three-dimensional geological modeling area.
The correcting module 160 is configured to, for each error position point, select a set number of multiple non-error position points in a direction gradually away from the error position point in a 360-degree direction, obtain boundary grid points of corresponding formation boundaries in the three-dimensional geological model located directly below the multiple non-error position points, interpolate three-dimensional coordinates of the boundary grid points having the same formation attributes by using a spline surface fitting interpolation algorithm, calculate corrected boundary grid points of corresponding formation boundaries directly below the error position point, and correct, according to the corrected boundary grid points, the formation attributes of grids in the three-dimensional geological model located directly below the error position point, to obtain a corrected three-dimensional geological model.
And the iterative output module 170 is configured to perform iterative correction on the three-dimensional geological model until the precision evaluation result of the corrected three-dimensional geological model meets the set requirement, stop the iterative correction, and output the three-dimensional geological model.
Example 3
The invention also provides computer equipment which comprises a processor, a memory and a three-dimensional geological model modeling program stored on the memory and capable of running on the processor, wherein the modeling program is executed by the processor to realize each process of the three-dimensional geological model modeling method embodiment, and can achieve the same technical effect, and in order to avoid repetition, the description is omitted.
The above-described embodiments of the present invention should not be construed as limiting the scope of the present invention. Any other corresponding changes and modifications made according to the technical idea of the present invention should be included in the protection scope of the claims of the present invention.
Claims (10)
1. A method of modeling a three-dimensional geological model, comprising the steps of:
step S1: acquiring drilling data of a plurality of drilling position points of a research area, wherein the drilling data comprises three-dimensional coordinates of a stratum boundary and stratum attributes;
step S2: resampling the drilling data of each drilling position point to enable the stratum boundaries of each stratum to have the same number of interpolation points, obtaining three-dimensional coordinates and stratum attributes of the interpolation points, and performing normalization processing on the stratum boundaries and the three-dimensional coordinates of the interpolation points;
step S3: training a stacking machine learning model by using the normalized three-dimensional coordinates of the stratum boundary and the interpolation point and the stratum attributes of the stratum boundary and the interpolation point, and predicting the stratum attribute of each grid point on a gridding three-dimensional geological modeling area of the research area by using the trained stacking machine learning model to obtain a three-dimensional geological model;
step S4: obtaining a geological data database of a research area, and obtaining the thickness and the attitude of each rock stratum at a plurality of position points in the research area by the geological data database;
step S5: performing precision evaluation on the three-dimensional geological model by using the thickness and the attitude of the rock stratum, screening out partial position points of the three-dimensional geological model, which do not meet the requirements on the modeling precision of a research area, as error position points and partial position points which meet the requirements as non-error position points according to the precision evaluation result, wherein the error position points and the non-error position points are grid points on the top surface of a gridded three-dimensional geological modeling area;
step S6: for each error position point, selecting a set number of multiple non-error position points in a 360-degree direction along a direction gradually far away from the error position point, obtaining boundary grid points of corresponding stratum boundaries under the multiple non-error position points in the three-dimensional geological model, interpolating three-dimensional coordinates of the boundary grid points with the same stratum attributes by using a spline surface fitting interpolation algorithm, calculating to obtain corrected boundary grid points of the corresponding stratum boundaries under the error position point, and correcting the stratum attributes of grids under the error position point in the three-dimensional geological model according to the corrected boundary grid points to obtain a corrected three-dimensional geological model;
step S7: and returning to the step S5 to perform iterative correction on the three-dimensional geological model until the precision evaluation result of the corrected three-dimensional geological model meets the set requirement, stopping iterative correction, and outputting the three-dimensional geological model.
2. A modeling method in accordance with claim 1, wherein: in step S2, the normalized formula is:
P * ＝(P-μ)/σ
wherein P represents coordinate data of stratum boundaries or interpolation points in different dimensions, mu represents the mean value of the coordinate data of all the stratum boundaries and the interpolation points in different dimensions, sigma represents the standard deviation of the coordinate data of all the stratum boundaries and the interpolation points in different dimensions, and P represents the standard deviation of the coordinate data of all the stratum boundaries and the interpolation points in different dimensions * And representing the normalized coordinate data of different dimensions of the stratum boundary or the interpolation point.
3. A modeling method in accordance with claim 1, wherein: the step S3 includes the following steps:
step S31: selecting a land parcel with a set thickness under an external rectangle of the research area, and carrying out grid division according to set density in the three-dimensional direction to obtain a grid three-dimensional geological modeling area;
step S32: dividing an original data set consisting of the normalized three-dimensional coordinates of the stratum boundary and the interpolation point and the corresponding stratum attributes of the stratum boundary and the interpolation point into a training set and a testing set according to a set proportion;
step S33: dividing the training set into 5 parts, wherein 1 part is not repeatedly taken as a test set and the other 4 parts are taken as training sets during each training, and 5 classifiers can be obtained according to the classification method;
step S34: for each classification method in the step S33, predicting the extracted test set by using 5 classifiers, and sequentially and vertically splicing each prediction result to be used as the characteristic of a meta classifier; predicting the test set divided in the step S32, and obtaining another characteristic of the meta classifier based on the prediction result;
step S35: two features provided by each classifier can be obtained in step S34, where the 3 base classifiers total 6 features, the 6 features are used as training features of the meta classifier, the corresponding real class is used as a training label, and the test set of the meta classifier is the real label corresponding to the prediction result of the test set partitioned in step S32 in step S34;
step S36: training the meta classifier by using the training set and the test set obtained in the step S35 to obtain a final stacking machine learning model;
step S37: and predicting each grid central point of the gridded three-dimensional geological modeling area by using the trained stacking machine learning model to obtain the stratum attribute of each point to be predicted.
4. A modeling method in accordance with claim 1, wherein: the step S5 includes the following steps:
step S51: calculating the thickness of each rock stratum at a plurality of positions in the geological data database by using the three-dimensional geological model, and calculating the thickness of the rock stratum calculated by the three-dimensional geological model and the thickness of the rock stratum in the geological data database to obtain the layering thickness error Q and the average relative thickness error D at the plurality of positions MAE Error Q of layered thickness and average relative error D of thickness MAE The calculation formula is as follows:
wherein n is the number of rock layers, D mi Thickness, D, of each formation at each location point calculated for the three-dimensional geological model bi A thickness for each formation at each location point in the geological data repository;
step S52: generating a stratum thickness error contour map of a research area on the top surface of a gridding three-dimensional geological modeling area according to the layered thickness errors Q at a plurality of position points by using a Krigin interpolation method in ArcGIS, averagely dividing the errors of the research area into five intervals according to the size values of the errors on the error contour map by using a natural discontinuity point method, taking grid points in a part of areas on the top surface corresponding to two intervals with the largest errors as error position points, and taking grid points in the part of areas on the top surface corresponding to the other three intervals as non-error position points;
step S53: calculating the actual horizontal distance S of the projection of the highest point of each rock stratum and the bottom point of each rock stratum in the research area on the horizontal plane according to the rock stratum attitude in the geological data database i True horizontal distance S i The calculation formula is as follows:
S i ＝L i sina i
wherein L is i Representing the length of the formation, a i Representing a formation dip angle;
calculating a predicted horizontal distance S 'projected on a horizontal plane from the highest point of each rock stratum to the bottom point of each rock stratum in a research area by using the three-dimensional geological model' i According to said actual horizontal distance S i And the predicted horizontal distance S' i Calculating to obtain the average relative error S of the distance MAE Distance average relative error S MAE The calculation formula is as follows:
step S54: averaging the thickness to obtain a relative error D MAE And said distance average relative error S MAE The average relative error MAE is taken as the average relative error MAE of the whole study area, and the calculation formula of the average relative error MAE is as follows:
5. a modeling method in accordance with claim 1, wherein: in step S6, a bi-cubic spline interpolation algorithm is selected as the spline surface fitting interpolation algorithm, and the set number of the corresponding non-error position points is 16.
6. The modeling method of claim 4, wherein: the step S7 includes the following steps:
and returning to the step S51 to perform iterative correction on the three-dimensional geological model, stopping iterative correction when the average relative error MAE is smaller than a set threshold value, and outputting the three-dimensional geological model.
7. The modeling method of claim 6, wherein: the number of corrections to the three-dimensional geological model is equal to 3.
8. A modeling method in accordance with claim 1, wherein: the following steps are also included after step S3 and step S7, respectively: and visualizing the three-dimensional geological model.
9. A three-dimensional geological model modeling apparatus, comprising:
a first data acquisition module for acquiring borehole data for a plurality of borehole location points of a study area, the borehole data including three-dimensional coordinates of a formation boundary and formation properties;
the data processing module is used for resampling the drilling data of each drilling position point to enable the stratum boundaries of each stratum to have the same number of interpolation points, obtaining three-dimensional coordinates and stratum attributes of the interpolation points, and normalizing the three-dimensional coordinates of the stratum boundaries and the three-dimensional coordinates of the interpolation points;
the machine learning model prediction module is used for training a stacking machine learning model by utilizing the normalized three-dimensional coordinates of the stratum boundary and the interpolation point and the stratum attributes of the stratum boundary and the interpolation point, and predicting the stratum attribute of each grid point on a gridding three-dimensional geological modeling area of the research area by utilizing the trained stacking machine learning model to obtain a three-dimensional geological model;
the second data acquisition module is used for acquiring a geological data database of the research area, and the thickness and the attitude of each rock stratum at a plurality of position points in the research area are obtained from the geological data database;
the precision evaluation module is used for carrying out precision evaluation on the three-dimensional geological model by utilizing the thickness and the attitude of the rock stratum, screening out partial position points of the three-dimensional geological model which do not meet the requirements on the modeling precision of a research area as error position points and partial position points which meet the requirements as non-error position points according to the precision evaluation result, wherein the error position points and the non-error position points are grid points on the top surface of a gridded three-dimensional geological modeling area;
the correction module is used for selecting a set number of non-error position points in a 360-degree direction along a direction gradually far away from the error position points for each error position point, obtaining boundary grid points of corresponding stratum boundaries positioned under the non-error position points in the three-dimensional geological model, interpolating three-dimensional coordinates of the boundary grid points with the same stratum attributes by using a spline surface fitting interpolation algorithm, calculating to obtain corrected boundary grid points of the corresponding stratum boundaries positioned under the error position points, and correcting the stratum attributes of grids positioned under the error position points in the three-dimensional geological model according to the corrected boundary grid points to obtain a corrected three-dimensional geological model;
and the iterative output module is used for performing iterative correction on the three-dimensional geological model until the precision evaluation result of the corrected three-dimensional geological model meets the set requirement, stopping iterative correction and outputting the three-dimensional geological model.
10. A computer device comprising a processor and a memory, a three-dimensional geological model modeling program stored on the memory and executable on the processor, the modeling program when executed by the processor performing the steps of the modeling method of any of claims 1-8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210545188.2A CN114943178A (en) | 2022-05-19 | 2022-05-19 | Three-dimensional geological model modeling method and device and computer equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210545188.2A CN114943178A (en) | 2022-05-19 | 2022-05-19 | Three-dimensional geological model modeling method and device and computer equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114943178A true CN114943178A (en) | 2022-08-26 |
Family
ID=82906777
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210545188.2A Pending CN114943178A (en) | 2022-05-19 | 2022-05-19 | Three-dimensional geological model modeling method and device and computer equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114943178A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115375867A (en) * | 2022-10-24 | 2022-11-22 | 山东省地质调查院(山东省自然资源厅矿产勘查技术指导中心) | Method, system, equipment and medium for calculating geothermal resource quantity by using grid model |
| CN115619950A (en) * | 2022-10-13 | 2023-01-17 | 中国地质大学（武汉） | Three-dimensional geological modeling method based on deep learning |
| CN116152460A (en) * | 2023-04-14 | 2023-05-23 | 瞳见科技有限公司 | Method, device, terminal and medium for generating rock stratum model based on UE4 |
-
2022
- 2022-05-19 CN CN202210545188.2A patent/CN114943178A/en active Pending
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115619950A (en) * | 2022-10-13 | 2023-01-17 | 中国地质大学（武汉） | Three-dimensional geological modeling method based on deep learning |
| CN115619950B (en) * | 2022-10-13 | 2024-01-19 | 中国地质大学（武汉） | Three-dimensional geological modeling method based on deep learning |
| CN115375867A (en) * | 2022-10-24 | 2022-11-22 | 山东省地质调查院(山东省自然资源厅矿产勘查技术指导中心) | Method, system, equipment and medium for calculating geothermal resource quantity by using grid model |
| CN116152460A (en) * | 2023-04-14 | 2023-05-23 | 瞳见科技有限公司 | Method, device, terminal and medium for generating rock stratum model based on UE4 |
| CN116152460B (en) * | 2023-04-14 | 2024-03-29 | 瞳见科技有限公司 | Method, device, terminal and medium for generating rock stratum model based on UE4 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP2869096B1 (en) | Systems and methods of multi-scale meshing for geologic time modeling | |
| Enge et al. | From outcrop to reservoir simulation model: Workflow and procedures | |
| CN114943178A (en) | Three-dimensional geological model modeling method and device and computer equipment | |
| EP3293552B1 (en) | System and method for editing geological models by switching between volume-based models and surface-based structural models augmented with stratigraphic fiber bundles | |
| US11042676B2 (en) | Representing structural uncertainty in a mesh representing a geological environment | |
| CN105205864B (en) | Geologic structure face threedimensional model method for automatic modeling and system based on multi-source data | |
| US9239398B2 (en) | Method of generating and combining multiple horizons to determine a seismic horizon and its uncertainty | |
| CN112363236B (en) | Gravity field data equivalent source continuation and data type conversion method based on PDE | |
| US10387583B2 (en) | Rotations from gradient directions | |
| CN112150582B (en) | Multi-modal data-oriented geological profile approximate expression method | |
| CN110827405A (en) | Digital remote sensing geological mapping method and system | |
| CN111221035A (en) | Seismic reflection wave slope and gravity anomaly data joint inversion method | |
| US10677948B2 (en) | Context based bounded hydrocarbon formation identification | |
| Choi et al. | Tunneling Analyst: A 3D GIS extension for rock mass classification and fault zone analysis in tunneling | |
| CN107633556A (en) | It is a kind of quantitatively to obtain the probabilistic method of three dimensional ore deposit geological model | |
| Miky et al. | A combined contour lines iteration algorithm and Delaunay triangulation for terrain modeling enhancement | |
| CN108763804A (en) | A kind of reject measures the rock mass broad sense RQD acquisition methods that null field influences | |
| CN112346139A (en) | Gravity data multilayer equivalent source continuation and data conversion method | |
| US11555944B2 (en) | Method for determination of subsoil composition | |
| CN105785444B (en) | Velocity field construction method | |
| CN115469361B (en) | Clastic rock stratum three-dimensional geological modeling method | |
| Liu et al. | Study on a computing technique suitable for true 3D modeling of complex geologic bodies | |
| Wood et al. | Model creation based on digital borehole records and interpreted geological cross‐sections | |
| Lopez et al. | 2.5 D Hexahedral Meshing for Reservoir Simulations | |
| CN116342798A (en) | Tunnel bad geological body three-dimensional modeling method based on advanced forecast data |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |