A quality assessment method for geological 3D models

Abstract

This invention provides a quality assessment method for 3D geological models. The method includes: S11, analyzing historical geological project documents to extract assessment indicators and data requirements, and determining relevant modeling source data based on the geological project type and objectives; S12, learning from the assessment schemes of successful historical geological project documents to generate a 3D geological model assessment scheme that includes the scope and methods of validation data; S13, monitoring and analyzing the 3D geological model modeling process based on the 3D geological model assessment scheme; S14, determining the weights of each assessment indicator in the 3D geological model assessment scheme to determine the quality level of the 3D geological model, and continuously adjusting the weight allocation of the assessment indicators; S15, determining whether the 3D geological model meets the acceptance requirements based on its quality level. This invention provides a systematic assessment method for the quality assessment of 3D geological models, thereby promoting the application and development of 3D geological models in various fields.

Description

Technical Field

[0001] This invention relates to the field of three-dimensional modeling technology, and in particular to a method for quality assessment of geological three-dimensional models. Background Technology

[0002] With the continuous advancement of computer technology, Geographic Information Systems (GIS), and 3D geological modeling technology, 3D geological models are playing an increasingly important role in digital earth construction, smart mine development, and three-dimensional surveying and monitoring of natural resources. However, due to factors such as differences in the accuracy and errors of the source data for 3D geological modeling, the complexity and variability of geological conditions, and the diversity of modeling methods, 3D geological models exhibit multidimensional uncertainties in quality. This poses a challenge to model quality assessment and thus limits their wider promotion and application.

[0003] Currently, although extensive research has been conducted both domestically and internationally on the uncertainties of 3D geological models, a unified technical guideline for quality assessment of these models remains lacking. This makes it difficult to provide effective references for the accuracy, quality, and applicability of various types of 3D geological models in practical applications. Summary of the Invention

[0004] Therefore, the purpose of this invention is to provide a quality assessment method for three-dimensional geological models, offering a systematic assessment method for the quality evaluation of three-dimensional geological models, thereby promoting the application and development of three-dimensional geological models in various fields.

[0005] To achieve the above-mentioned objectives, this invention provides a method for quality assessment of geological three-dimensional models, the system comprising:

[0006] S11. Analyze historical geological project documents, extract evaluation indicators and data requirements, and determine the relevant modeling source data required for the evaluation indicators based on the geological project type and objectives.

[0007] S12. Learn from the evaluation schemes of successful geological project documents in the past, and combine them with the current geological project type, objectives, evaluation indicators and relevant modeling source data to generate a three-dimensional geological model evaluation scheme that includes the scope and methods of validation data.

[0008] S13. Based on the three-dimensional geological model evaluation scheme, the real-time data and the three-dimensional geological model images generated during the three-dimensional geological model modeling process are monitored and analyzed.

[0009] S14. Determine the weight of each evaluation indicator in the three-dimensional geological model evaluation scheme based on historical evaluation data, summarize the weights to obtain the score of the three-dimensional geological model, determine the quality level of the three-dimensional geological model, and continuously adjust the weight allocation of the evaluation indicators based on the quality level of the three-dimensional geological model.

[0010] S15. Determine whether the three-dimensional geological model meets the acceptance requirements based on its quality level. If yes, automatically generate a three-dimensional geological model quality assessment report. If no, update the modeling source data and repeat steps S12 to S14 until the quality level of the three-dimensional geological model meets the acceptance requirements.

[0011] Furthermore, step S11 specifically includes:

[0012] Document conversion processing is performed based on the characteristics of historical geological project documents. The main text content of each document is extracted, and the evaluation indicators and data requirements of the main text content of each document are labeled to establish a geological modeling corpus. The geological modeling corpus includes a list of defined geological terms.

[0013] Identify and associate the rows and columns in the list of geological terms with the labeled evaluation indicators and data requirements, and generate an analysis module based on the identified rows and columns. The analysis module is used to identify or recognize the attribute values ​​of the corresponding attributes of the rows and columns.

[0014] Based on the geological project type and objectives, the analysis module automatically extracts the evaluation indicators and data requirements for this 3D geological modeling from the current geological project documents. The geological modeling corpus is updated with each extracted geological project document and the identified evaluation indicators and data requirements.

[0015] Furthermore, in step S12, the evaluation scheme for learning from the successful quality project documentation includes:

[0016] Based on the evaluation schemes of historically successful geological project documents, obtain a set N of evaluation schemes, perform statistical analysis on the historical evaluation data in the set N, screen out the main factors affecting the modeling of the three-dimensional geological model, and establish the correspondence between the modeling of the three-dimensional geological model and the main factors.

[0017] The information gain algorithm is used to evaluate the main factors that affect the modeling of the three-dimensional geological model. The feature with the largest information gain among the main factors is selected as the main feature. The identified main features are classified into categories, and the main features are used as the first-level features of each category. The first-level features are used as the root nodes of the decision tree.

[0018] Based on each primary feature root node, secondary features are selected sequentially for splitting, and child nodes of each node are recursively constructed until all feature levels are expressed, at which point the decision tree construction ends, and all evaluation schemes in the set N of evaluation schemes are used to build decision trees and are marked as a decision tree set.

[0019] Furthermore, in step S12, a three-dimensional geological model evaluation scheme is generated, which includes the scope and method of validation data, specifically including:

[0020] Perform feature clustering analysis on the current geological project types, objectives, evaluation indicators, and related modeling source data, and merge similar features into feature groups;

[0021] Based on feature groups, the complexity of decision trees in the decision tree set is evaluated by the scale complexity measurement algorithm and leaf node distribution. Decision trees whose complexity calculation results are greater than the preset complexity adjustment threshold are selected, and a set of complex decision trees M is established based on the selection results.

[0022] Feature similarity analysis is performed on decision trees and feature groups in the complex decision tree set M. Clustering algorithm is used for iterative calculation to reduce the feature dimension of each complex decision tree and divide similar features into multiple cluster groups.

[0023] Using feature similarity as the weight, the clustering results in each cluster group are weighted and summarized to generate new synthetic features. The information gain of the new synthetic features is re-evaluated to determine a three-dimensional geological model evaluation scheme that includes the scope and method of validation data.

[0024] Furthermore, in step S13, real-time data during the three-dimensional geological modeling process is monitored and analyzed, specifically including:

[0025] Based on project type, objectives and related modeling source data, a three-dimensional spatial distribution map and Voronoi diagram of the source data are established. The density distribution of data points is calculated through kernel density estimation to generate a three-dimensional density map.

[0026] The three-dimensional spatial distribution map of the source data is used to assess the distribution and continuity of the data in three-dimensional space, and whether there are any missing data areas.

[0027] The control region area for each data point is calculated based on the Voronoi diagram to evaluate the uniformity of data coverage in the horizontal direction and the sampling interval of data in the vertical direction.

[0028] The density of modeling data points in the horizontal direction and the coverage integrity of data in the vertical direction are evaluated based on the 3D density map.

[0029] Furthermore, in step S13, the three-dimensional geological model images generated during the three-dimensional geological model modeling process are analyzed, specifically including:

[0030] In three-dimensional space, calculate the cross product of each side of the triangle to determine all possible separation axes, which include the normal vector of each face of the two triangles and the cross product of each side;

[0031] For each separation axis, project all vertices of the two triangles onto the corresponding separation axis, insert the triangle data into the R-tree, and check whether the projection intervals of the two triangles overlap by traversing the spatial region in the R-tree.

[0032] Furthermore, in step S14, the weights of each evaluation index in the three-dimensional geological model evaluation scheme are determined, specifically including:

[0033] The preprocessed historical assessment data are arranged according to assessment indicators and three-dimensional geological model samples to form a decision matrix. The rows of the decision matrix represent different three-dimensional geological model samples, and the columns represent various assessment indicators.

[0034] Calculate the entropy value of each evaluation index, that is, calculate the proportion of the standardized value of the i-th three-dimensional geological model sample under the j-th evaluation index to the sum of the standardized values ​​of all three-dimensional geological model samples for the corresponding evaluation index;

[0035] Based on the entropy value of each evaluation indicator, the weight of each evaluation indicator is calculated, as expressed by:

[0036]

[0037] Among them, w j Let e ​​be the weight of the j-th evaluation indicator. j Let the entropy value of the j-th evaluation index be...

[0038] This is used to normalize the weights, ensuring that the sum of all weights equals 1.

[0039] Furthermore, in step S14, the weight allocation of the evaluation indicators is continuously adjusted, specifically including:

[0040] Historical assessment data, current weight allocation of each assessment indicator, and quality results of the corresponding three-dimensional geological model are integrated into a multi-dimensional vector representing the state space, and the adjustable range of the weight of each assessment indicator is preset.

[0041] When the weights are adjusted and the three-dimensional geological model is re-evaluated, if the evaluation result is closer to the actual quality judgment, a positive reward will be given; if the evaluation result deviates more from the actual quality, a negative reward will be given.

[0042] Initialize the policy network by taking the multidimensional vector, the adjustable range of the weights of each evaluation index, and the reward mechanism as inputs to the policy network, and output a new weight allocation scheme, which includes the selection probability of each weight adjustment action.

[0043] Compared with the prior art, the beneficial effects of the present invention are:

[0044] This invention provides a quality assessment method for 3D geological models. First, it analyzes historical geological project documents, extracts key assessment indicators and data requirements, and, combined with the characteristics of the current project, generates an assessment scheme that includes the scope and methods for validation data, ensuring the scientific validity and relevance of the scheme. Second, it monitors real-time data and model images during the modeling process, determines weights and scores based on historical assessment data, and accurately controls model quality. Furthermore, if acceptance requirements are not met, the data is updated and relevant steps are re-executed to adapt to the needs of complex geological projects, improving modeling efficiency and quality. Finally, it clarifies acceptance criteria and automatically generates a quality assessment report. This invention provides a comprehensive, efficient, and reliable assessment method for 3D geological modeling, which is of great significance for improving the quality and efficiency of 3D geological modeling. Attached Figure Description

[0045] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of this application. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0046] Figure 1 This is a schematic diagram of a quality assessment method for geological three-dimensional models provided in this embodiment. Detailed Implementation

[0047] The present invention will now be described in further detail 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 not intended to limit it. Furthermore, it should be noted that, for ease of description, the accompanying drawings show only the parts relevant to the present invention, and not all of the structures.

[0048] Reference Figure 1 This embodiment provides a method for quality assessment of geological three-dimensional models, the system comprising:

[0049] S11. Analyze historical geological project documents, extract evaluation indicators and data requirements, and determine the relevant modeling source data required for the evaluation indicators based on the geological project type and objectives.

[0050] S12. Learn from the evaluation schemes of successful geological project documents in the past, and combine them with the current geological project type, objectives, evaluation indicators and relevant modeling source data to generate a three-dimensional geological model evaluation scheme that includes the scope and methods of validation data.

[0051] S13. Based on the three-dimensional geological model evaluation scheme, the real-time data and the three-dimensional geological model images generated during the three-dimensional geological model modeling process are monitored and analyzed.

[0052] S14. Determine the weights of each evaluation indicator in the three-dimensional geological model evaluation scheme based on historical evaluation data, summarize the weights to obtain the score of the three-dimensional geological model, determine the quality level of the three-dimensional geological model, and continuously adjust the weight allocation of the evaluation indicators based on the quality level of the three-dimensional geological model.

[0053] S15. Determine whether the three-dimensional geological model meets the acceptance requirements based on its quality level. If yes, automatically generate a three-dimensional geological model quality assessment report. If no, update the modeling source data and repeat steps S12 to S14 until the quality level of the three-dimensional geological model meets the acceptance requirements.

[0054] In this embodiment, the quality assessment of the 3D geological model includes four parts: assessment of the modeling source data, assessment of geological rationality, assessment of model accuracy, and assessment of topological relationships. The main process includes: Before model construction, the 3D geological model construction unit should submit a model quality assessment request to the assessment unit. Based on the assessment requirements, all necessary materials for the assessment are provided, including but not limited to modeling source data, contracts, task books, acceptance requirements, and model documentation. The assessment unit accepts the assessment task, formulates a model assessment plan, determines the method and scope for reserving verification data, and tracks the entire 3D geological modeling process. The 3D geological modeling source data, geological rationality, model accuracy, and topological relationships are assessed separately. Based on the characteristics and actual conditions of the modeling area, different weights are assigned to each assessment indicator, and the model score is obtained to determine the model quality level. The construction unit modifies and improves the model according to the assessment report and resubmits it to the assessment unit for evaluation until the model meets the acceptance requirements. A 3D geological model quality assessment report is automatically generated by combining template-based report generation technology and Natural Language Generation (NLG) technology.

[0055] Step S11 specifically includes:

[0056] Document conversion processing is performed based on the characteristics of historical geological project documents. The main text content of each document is extracted, and the evaluation indicators and data requirements of the main text content of each document are labeled to establish a geological modeling corpus. The geological modeling corpus includes a list of defined geological terms.

[0057] Identify and associate the rows and columns in the list of geological terms with the labeled evaluation indicators and data requirements, and generate an analysis module based on the identified rows and columns. The analysis module is used to identify or recognize the attribute values ​​of the corresponding attributes of the rows and columns.

[0058] Based on the geological project type and objectives, the analysis module automatically extracts the evaluation indicators and data requirements for this 3D geological modeling from the current geological project documents. The geological modeling corpus is updated with each extracted geological project document and the identified evaluation indicators and data requirements.

[0059] In this embodiment, the main text is extracted from historical geological project documents, and the evaluation indicators and data requirements are accurately labeled, improving the efficiency of information extraction and quickly providing the necessary materials for subsequent geological modeling and other related work. A geological modeling corpus containing a well-defined list of geological terms is established to facilitate unified management and retrieval of geological terms from a large number of geological project documents. By identifying and associating the list of geological terms with the rows and columns corresponding to the labeled evaluation indicators and data requirements, an analysis module is generated, enabling intelligent analysis of the relationships between elements in the document. This allows for rapid and accurate identification of attribute values ​​corresponding to rows and columns, helping to better grasp geological characteristics and patterns. Based on different geological project types and objectives, the analysis module automatically extracts the evaluation indicators and data requirements for the current 3D geological modeling, and the results of each extraction are used to update the geological modeling corpus, ensuring that the corpus is continuously enriched and improved to adapt to the needs of different types and purposes of geological projects.

[0060] Specifically, historical geological project documents from different project stages and research areas are collected, organized, and categorized. Document conversion technology is then used to uniformly convert documents of different formats to extract the main text content. The extracted main text content is annotated to clarify which content belongs to evaluation indicators (such as stability indicators of geological bodies, mineral resource reserve estimation indicators, etc.) and which belongs to data requirements (such as geological exploration data accuracy requirements, physicochemical property data of different geological layers, etc.), and these are also annotated. The annotated document content is organized and summarized, from which defined geological terms are extracted, such as various rock names, geological structure types, ore types, etc., to establish a geological terminology list, and each term is given a clear definition to ensure its accuracy and consistency in the field of geological modeling. The document main text content annotated with evaluation indicators and data requirements, along with the geological terminology list, are integrated to construct a geological modeling corpus. Simultaneously, a storage structure and retrieval method are designed for the corpus to facilitate subsequent queries and use.

[0061] By employing text analysis algorithms and data mining techniques, such as dependency parsing and keyword extraction in natural language processing, the system identifies rows and columns in a list of geological terms that correspond to the labeled evaluation indicators and data requirements. For example, it analyzes the relationship between the row or column containing a geological term in the document and a specific evaluation indicator such as "rock compressive strength" or a data requirement such as "stratum thickness measurement data." Based on the identified rows and columns, an analysis module is developed, utilizing programming languages ​​and relevant data analysis tools. This module is capable of identifying and recognizing the attribute values ​​corresponding to the relevant rows and columns in the document, enabling it to quickly locate and extract key attribute information such as the numerical value of rock compressive strength at a geological point and specific data on stratum thickness.

[0062] When faced with a new geological project document, the first step is to determine its corresponding geological project type and objective, such as whether it is a mineral resource exploration project or a geological modeling project for urban underground space development and utilization. The document is then processed according to the preprocessing and text extraction procedures described above to obtain its main text content. Using the generated analysis module, the main text content of the new document is automatically analyzed to extract evaluation indicators and data requirements relevant to this 3D geological modeling project. For example, for a mine geological modeling project, the analysis module can automatically extract key content such as ore body morphology evaluation indicators and ore grade data requirements. After integrating each extracted geological project document and the identified evaluation indicators and data requirements, the data is fed back into the geological modeling corpus. The corpus is expanded and updated according to established update rules, enabling it to cover more geological project information of different types and objectives, continuously improving the corpus's completeness and usability.

[0063] In step S12, the evaluation scheme for learning from the successful quality project documentation includes:

[0064] Based on the evaluation schemes of historically successful geological projects, an evaluation scheme set N is obtained. Statistical analysis is performed on the historical evaluation data in the evaluation scheme set N to screen out the main factors affecting the modeling of the 3D geological model and establish the correspondence between the 3D geological model modeling and the main factors.

[0065] The information gain algorithm is used to evaluate all the main factors affecting the modeling of the 3D geological model. The feature with the largest information gain among the main factors is selected as the main feature. The identified main features are classified into categories, and the main features are used as the first-level features of each category. The first-level features are used as the root nodes of the decision tree.

[0066] Based on each primary feature root node, secondary features are selected sequentially for splitting, and child nodes of each node are recursively constructed until all feature levels are expressed, at which point the decision tree construction ends, and all evaluation schemes in the set N of evaluation schemes are used to build decision trees and are marked as a decision tree set.

[0067] In this embodiment, through in-depth learning and analysis of historical successful geological project document evaluation schemes, the main factors affecting 3D geological modeling are accurately identified and corresponding relationships are established. This provides clear and crucial reference for subsequent 3D geological modeling, effectively improving modeling accuracy and avoiding interference from irrelevant or secondary factors. The information gain algorithm is used to evaluate the main factor features, scientifically selecting the features with the highest information gain as the main features. This ensures that the selected features have the strongest discriminative power and representativeness for 3D geological modeling, further improving modeling efficiency and quality while reducing unnecessary computation and analysis costs.

[0068] A decision tree is constructed using the primary feature as the root node, and secondary features are selected sequentially for splitting, ultimately forming a complete set of decision trees. This provides a clear, intuitive, and systematic framework for the evaluation and decision-making in 3D geological modeling. The decision tree set, built upon a collection of historical evaluation schemes, provides a referable and replicable model foundation for the formulation of evaluation schemes for subsequent new geological projects and the modeling of 3D geological models. With the accumulation and updating of more geological project data, the decision tree set can be further optimized and improved, enabling the modeling method to continuously adapt to new geological conditions and needs, exhibiting scalability and continuous improvement.

[0069] Specifically, data relevant to the assessment is extracted from collected historical documents, including various geological parameters (such as rock hardness, stratum thickness, groundwater level, etc.), assessment methods (such as drilling, geophysical exploration, etc.), assessment indicators (such as model accuracy, reliability indicators, etc.), and the final modeling results, forming an assessment scheme set N. Statistical methods (such as correlation analysis, analysis of variance, etc.) are used to analyze the historical assessment data in the assessment scheme set N to identify factors significantly correlated with the 3D geological modeling results. For example, analysis revealed that rock type, geological structural complexity, and borehole data density are highly correlated with modeling accuracy. Key aspects of 3D geological modeling (such as model construction and model validation) are linked to the selected main factors in a modeling process, visually demonstrating the influence and degree of each main factor on the modeling process and results, forming a correspondence model between 3D geological modeling and the main factors.

[0070] Information gain algorithms are used to calculate the information gain of each of the main factors influencing the modeling of the 3D geological model. For example, in a geological project, the information gain values ​​of main factors such as rock type, porosity, and permeability are calculated to identify the feature with the highest information gain, such as rock type. Based on the nature and geological significance of the main features, they are categorized. Main features with similar characteristics or belonging to the same geological category are grouped into the same category, and the main feature with the highest information gain (such as rock type) is designated as the first-level feature of that category, becoming the root node of the decision tree. Based on the root node, secondary features are selected sequentially in descending order of information gain to generate child nodes. For example, under the rock type root node, secondary features (such as porosity and mineral composition) related to different rock types are further selected for splitting until all feature levels are effectively represented, completing the construction of a single decision tree. This process is repeated to construct corresponding decision trees for all evaluation schemes in the set N, forming a decision tree set.

[0071] In new 3D geological modeling projects, based on the pre-constructed decision tree set, starting from the root node, judgments and decisions are made sequentially along the branches of the decision tree according to the data characteristics of the actual geological project, quickly determining the evaluation scheme and modeling method that matches the project. As new geological projects are completed and evaluation schemes are accumulated, new data and schemes are incorporated into the evaluation scheme set, and statistical analysis, feature evaluation, and decision tree construction are carried out again to continuously optimize and update the decision tree set.

[0072] In step S12, a three-dimensional geological model evaluation scheme is generated, which includes the scope and methods for validating the data. Specifically, this includes:

[0073] Feature clustering analysis is performed on the current geological project types, objectives, evaluation indicators, and related modeling source data to merge similar features into feature groups.

[0074] Based on feature groups, the complexity of decision trees in the decision tree set is evaluated by the scale complexity measurement algorithm and leaf node distribution. Decision trees whose complexity calculation results are greater than the preset complexity adjustment threshold are selected, and a set of complex decision trees M is established based on the selection results.

[0075] Feature similarity analysis is performed on decision trees and feature groups in a complex decision tree set M. Clustering algorithms are used for iterative calculations to reduce the feature dimension of each complex decision tree and divide similar features into multiple cluster groups.

[0076] Using feature similarity as the weight, the clustering results in each cluster group are weighted and summarized to generate new synthetic features. The information gain of the new synthetic features is re-evaluated to determine a three-dimensional geological model evaluation scheme that includes the scope and method of validation data.

[0077] In this embodiment, cluster analysis is performed on the relevant features of the current geological project to merge similar features into feature groups, thereby reducing the number of features and lowering the complexity of subsequent analysis. A complex decision tree set M is established based on these feature groups, allowing subsequent analysis and modeling to focus on more challenging and complex decision trees, improving the relevance and accuracy of model evaluation, and better meeting the needs of complex geological modeling scenarios. Feature similarity analysis is performed on the decision trees and feature groups in the complex decision tree set M, and iterative calculations using clustering algorithms reduce the feature dimension of each complex decision tree, further simplifying the feature space. Using feature similarity as weight, the clustering results in each cluster group are weighted and aggregated to generate new synthetic features, and information gain is re-evaluated. Finally, a 3D geological model evaluation scheme including the scope and methods of validation data is determined, providing a reliable basis for the accurate evaluation of 3D geological models.

[0078] Specifically, the process involves collecting current geological project types, objectives, evaluation indicators, and relevant modeling source data. The data is then cleaned and standardized to ensure quality and comparability. Features are extracted, such as geological body type, data accuracy requirements, and model resolution. K-Means algorithms are used for cluster analysis, merging similar features into feature groups, for example, merging different types of lithological features into a "lithological feature group." A scale complexity measurement algorithm is selected, and the complexity of each decision tree in the decision tree set is evaluated based on the structure and leaf node distribution, calculating a complexity index value. A preset complexity adjustment threshold is set, and decision trees with complexity calculation results greater than this threshold are selected to establish a complex decision tree set M. For example, a threshold of 0.8 is set, and decision trees with a complexity greater than 0.8 are included in set M.

[0079] Feature similarity analysis is performed on the decision trees and feature groups in the complex decision tree set M to calculate the similarity values ​​between features and determine the similarity relationships between them. Iterative calculations are performed using clustering algorithms to divide similar features into multiple cluster groups based on feature similarity, such as dividing similar geological structural features into "fold structure feature group" and "fault structure feature group." Using feature similarity as a weight, the clustering results in each cluster group are weighted and summarized to generate new synthetic features. For example, for a given cluster group, the synthetic feature value is calculated based on the feature similarity weights. The information gain of the new synthetic features is re-evaluated to determine the contribution of each synthetic feature to the 3D geological model evaluation. Based on the evaluation results and actual needs, a 3D geological model evaluation scheme is determined, including the scope of validation data (such as geological data of a specific region, data of a specific depth range, etc.) and methods (such as cross-validation, hold-out method, etc.), to ensure the effectiveness and reliability of the evaluation scheme.

[0080] In step S13, real-time data during the three-dimensional geological modeling process is monitored and analyzed, specifically including:

[0081] Based on project type, objectives, and relevant modeling source data, a 3D spatial distribution map and Voronoi diagram of the source data are established. The density distribution of data points is calculated through kernel density estimation to generate a 3D density map.

[0082] The three-dimensional spatial distribution map of the source data is used to assess the distribution and continuity of the data in three-dimensional space, as well as whether there are any missing data areas.

[0083] The control area area for each data point is calculated based on the Voronoi diagram, and the coverage uniformity of the data in the horizontal direction and the sampling interval of the data in the vertical direction are evaluated.

[0084] The density of modeling data points in the horizontal direction and the coverage integrity of data in the vertical direction are evaluated based on the 3D density map.

[0085] In this embodiment, ensuring data consistency is crucial during geological modeling due to the diversity of data sources. Data used in modeling should undergo consistency checks. For resolving inconsistencies, the original data should be used as the standard, with borehole data used to correct geological maps, stratigraphy, and profiles. Data consistency checks and corrections should be based on the source data processing, stratigraphic standardization, modeling process, and modeling results descriptions within the 3D geological model evaluation scheme. Evaluating the quality of modeling source data requires considering both horizontal and vertical data distribution, i.e., assessing the distribution of modeling source data in 3D space. This is achieved by establishing a 3D spatial distribution map of the source data. The uniformity of data distribution is assessed by constructing a Voronoi diagram and calculating the control area area for each data point. Finally, a 3D density map is generated by calculating the density distribution of data points using kernel density estimation (KDE).

[0086] Based on the source data's 3D spatial distribution map, Voronoi diagram, and 3D density map, the following evaluations are performed: The uniformity of data coverage in the horizontal direction and the density of modeling data points in the horizontal direction are evaluated. The sampling interval and the completeness of data coverage in the vertical direction are evaluated. The distribution and continuity of data in 3D space are assessed, and any missing data areas are identified. Geological rationality refers to whether the geological model conforms to known geological principles and theories, and is evaluated separately for structural and attribute models. The structural model should conform to geological understanding, and the rationality of geological contact relationships such as conformable contact, unconformable contact, fault contact, and intrusive contact is evaluated. For the attribute model, tectonic restoration should be considered during attribute interpolation to eliminate the influence of faults.

[0087] Model accuracy refers to the degree of conformity between the 3D geological model and the source data, and is evaluated separately for structural and attribute models. The structural model is evaluated based on the source data and modeling documentation to assess its conformity with the source data and its consistency with the required model accuracy. The attribute model is evaluated through mesh generation to assess whether the mesh size is consistent with the required model accuracy; it is also evaluated based on a comprehensive columnar section to assess whether the mesh attributes conform to geological understanding, and the continuity or discreteness of the mesh attributes should match the characteristics of the data. The structural and attribute models should express geological understanding consistently; for example, the understanding of geological objects in engineering, hydrological, and other specialized geological models should be consistent with the framework of the basic geological model. For Quaternary simple tectonic areas with good borehole data, sampling verification (boreholes, profiles, etc.) can be used, including three methods: verification of modeling data, verification of reserved modeling data, and verification using field or on-site physical work (boreholes, measured profiles, etc.).

[0088] In step S13, the three-dimensional geological model images generated during the three-dimensional geological model modeling process are analyzed, specifically including:

[0089] In three-dimensional space, calculate the cross product of each side of the triangle to determine all possible separation axes, which include the normal vector of each face of the two triangles and the cross product of each side.

[0090] For each separation axis, project all vertices of the two triangles onto the corresponding separation axis, insert the triangle data into the R-tree, and check whether the projection intervals of the two triangles overlap by traversing the spatial region in the R-tree.

[0091] In this embodiment, the evaluation of the topological relationships of the 3D geological model needs to be approached from two aspects: macro-topology and internal assessment. Macro-topological assessment evaluates the overall integrity of the model, the closure of geological bodies, and the continuity of geological surfaces. Internal topological assessment is an evaluation of spatial geometry, assessing the closure, shape, self-intersection, and densification of key areas of the mesh within the geological object. For example, the Separating Axis Theorem (SAT) can be used to detect whether two triangles intersect; the R-tree spatial partitioning algorithm can be used to accelerate intersection detection.

[0092] In the process of 3D geological modeling, all possible separation axes are determined by calculating the cross product of each side of a triangle. This includes the normal vector of each face of the two triangles and the cross product of each side. This method allows for a comprehensive and detailed analysis of the relative positional relationship between the two triangles in 3D space. After projecting all vertices of the two triangles onto the corresponding separation axes, checking for overlap between the projection intervals allows for a quick and accurate determination of whether the two triangles have collided. This helps to promptly identify and correct unreasonable structures or geometric conflicts in the 3D geological model, ensuring the accuracy and realism of the model and improving the modeling quality.

[0093] Inserting triangle data into an R-tree and leveraging its spatial indexing capabilities allows for rapid location and querying of spatial regions related to the current separating axis. During R-tree traversal, irrelevant spatial regions can be skipped, reducing unnecessary calculations and comparisons and significantly accelerating collision detection. This is particularly important for large-scale 3D geological modeling, effectively improving modeling efficiency, saving time and computational resources, and making modeling software more fluid and efficient when handling complex geological scenes.

[0094] In step S14, the weights of each evaluation index in the three-dimensional geological model evaluation scheme are determined, specifically including:

[0095] The preprocessed historical assessment data are arranged according to assessment indicators and three-dimensional geological model samples to form a decision matrix. The rows of the decision matrix represent different three-dimensional geological model samples, and the columns represent various assessment indicators.

[0096] Calculate the entropy value of each evaluation index, that is, calculate the proportion of the standardized value of the i-th three-dimensional geological model sample under the j-th evaluation index to the sum of the standardized values ​​of all three-dimensional geological model samples for the corresponding evaluation index.

[0097] Based on the entropy value of each evaluation indicator, the weight of each evaluation indicator is calculated, as expressed by:

[0098]

[0099] Among them, w j Let e ​​be the weight of the j-th evaluation indicator. j Let the entropy value of the j-th evaluation index be... This is used to normalize the weights, ensuring that the sum of all weights equals 1.

[0100] In this embodiment, the weight is determined by calculating the entropy value of each evaluation indicator. The entropy value reflects the dispersion and importance of the evaluation indicators, making the weight allocation more reasonable and scientific, accurately reflecting the actual influence of each evaluation indicator in the 3D geological model evaluation, and improving the reliability and credibility of the evaluation results. The preprocessed historical evaluation data is arranged according to the evaluation indicators and 3D geological model samples to form a decision matrix. This fully integrates historical data resources, ensuring that the evaluation scheme is based on rich data, avoiding the bias caused by insufficient data utilization, and improving the comprehensiveness and accuracy of the evaluation. The rows of the decision matrix represent different 3D geological model samples, such as Sample 1, Sample 2, etc., and the columns represent various evaluation indicators, such as model accuracy, geological body integrity, and construction efficiency. Each element in the matrix corresponds to the specific value of a certain 3D geological model sample under a certain evaluation indicator.

[0101] In step S14, the weight allocation of the evaluation indicators is continuously adjusted, specifically including:

[0102] Historical assessment data, current weight allocation of each assessment indicator, and quality results of the corresponding three-dimensional geological model are integrated into a multi-dimensional vector representing the state space, with the adjustable range of the weight of each assessment indicator preset.

[0103] When the weights are adjusted and the 3D geological model is re-evaluated, a positive reward is given if the evaluation result is closer to the actual quality judgment, and a negative reward is given if the evaluation result deviates more from the actual quality.

[0104] Initialize the policy network by taking the multidimensional vector, the adjustable range of the weights of each evaluation index, and the reward mechanism as inputs to the policy network, and output a new weight allocation scheme, which includes the selection probability of each weight adjustment action.

[0105] In this embodiment, reinforcement learning is used to dynamically adjust the weights of evaluation indicators based on historical evaluation data, current weight allocation, and model quality results. This automatically identifies better weight allocation schemes, making the evaluation results more accurately approximate the actual quality judgment. With continuous adjustment and learning, the policy network can gradually adapt to various complex evaluation scenarios, further improving the versatility and robustness of the evaluation scheme and ensuring the generation of reasonable weight allocation schemes under different circumstances. A reward mechanism is introduced: positive rewards are given when the evaluation result is more accurate after weight adjustment, and negative rewards are given otherwise, providing clear directional guidance and quantitative basis for weight adjustment. The policy network can learn which weight adjustment actions are beneficial to improving evaluation accuracy based on reward signals, thereby generating new weight allocation schemes in a targeted manner, avoiding blind adjustments, and making the adjustment process more efficient and orderly.

[0106] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for quality assessment of geological three-dimensional models, characterized in that, The method includes: S11. Analyze historical geological project documents, extract evaluation indicators and data requirements, and determine the relevant modeling source data required for the evaluation indicators based on the geological project type and objectives. S12. Learn from the evaluation schemes of successful geological project documents in the past, and combine them with the current geological project type, objectives, evaluation indicators and relevant modeling source data to generate a three-dimensional geological model evaluation scheme that includes the scope and methods of validation data. In step S12, the evaluation scheme for learning from the successful quality project documentation includes: Based on the evaluation schemes of historically successful geological project documents, obtain a set N of evaluation schemes, perform statistical analysis on the historical evaluation data in the set N, screen out the main factors affecting the modeling of the three-dimensional geological model, and establish the correspondence between the modeling of the three-dimensional geological model and the main factors. The information gain algorithm is used to evaluate the main factors that affect the modeling of the three-dimensional geological model. The feature with the largest information gain among the main factors is selected as the main feature. The identified main features are classified into categories, and the main features are used as the first-level features of each category. The first-level features are used as the root nodes of the decision tree. Based on each primary feature root node, secondary features are selected sequentially for splitting, and child nodes of each node are recursively constructed until all feature levels are expressed, at which point the decision tree construction ends, and all evaluation schemes in the set N of evaluation schemes are used to build decision trees, which are then labeled as the decision tree set; In step S12, a three-dimensional geological model evaluation scheme is generated, which includes the scope and methods for validating the data. Specifically, this includes: Perform feature clustering analysis on the current geological project types, objectives, evaluation indicators, and related modeling source data, and merge similar features into feature groups; Based on feature groups, the complexity of decision trees in the decision tree set is evaluated by the scale complexity measurement algorithm and leaf node distribution. Decision trees whose complexity calculation results are greater than the preset complexity adjustment threshold are selected, and a set of complex decision trees M is established based on the selection results. Feature similarity analysis is performed on decision trees and feature groups in the complex decision tree set M. Clustering algorithm is used for iterative calculation to reduce the feature dimension of each complex decision tree and divide similar features into multiple cluster groups. Using feature similarity as the weight, the clustering results in each cluster group are weighted and summarized to generate new synthetic features. The information gain of the new synthetic features is re-evaluated to determine the three-dimensional geological model evaluation scheme that includes the scope and method of validation data. S13. Based on the three-dimensional geological model evaluation scheme, the real-time data and the three-dimensional geological model images generated during the three-dimensional geological model modeling process are monitored and analyzed. In step S13, the three-dimensional geological model images generated during the three-dimensional geological model modeling process are analyzed, specifically including: Calculate the cross product of each side of the triangle in three-dimensional space to determine all possible separation axes, which include the normal vector of each face of the two triangles and the cross product of each side; For each separation axis, project all vertices of the two triangles onto the corresponding separation axis, insert the triangle data into the R-tree, and check whether the projection intervals of the two triangles overlap by traversing the spatial region in the R-tree. S14. Determine the weight of each evaluation indicator in the three-dimensional geological model evaluation scheme based on historical evaluation data, summarize the weights to obtain the score of the three-dimensional geological model, determine the quality level of the three-dimensional geological model, and continuously adjust the weight allocation of the evaluation indicators based on the quality level of the three-dimensional geological model. S15. Determine whether the three-dimensional geological model meets the acceptance requirements based on its quality level. If yes, automatically generate a three-dimensional geological model quality assessment report. If no, update the modeling source data and repeat steps S12 to S14 until the quality level of the three-dimensional geological model meets the acceptance requirements.

2. The quality assessment method for geological three-dimensional models according to claim 1, characterized in that, Step S11 specifically includes: Document conversion processing is performed based on the characteristics of historical geological project documents. The main text content of each document is extracted, and the evaluation indicators and data requirements of the main text content of each document are labeled to establish a geological modeling corpus. The geological modeling corpus includes a list of defined geological terms. Identify and associate the rows and columns in the list of geological terms with the labeled evaluation indicators and data requirements, and generate an analysis module based on the identified rows and columns. The analysis module is used to identify or recognize the attribute values ​​of the corresponding attributes of the rows and columns. Based on the geological project type and objectives, the analysis module automatically extracts the evaluation indicators and data requirements for this 3D geological modeling from the current geological project documents. The geological modeling corpus is updated with each extracted geological project document and the identified evaluation indicators and data requirements.

3. The quality assessment method for geological three-dimensional models according to claim 1, characterized in that, In step S13, real-time data during the three-dimensional geological modeling process is monitored and analyzed, specifically including: Based on project type, objectives and related modeling source data, a three-dimensional spatial distribution map and Voronoi diagram of the source data are established. The density distribution of data points is calculated through kernel density estimation to generate a three-dimensional density map. The three-dimensional spatial distribution map of the source data is used to assess the distribution and continuity of the data in three-dimensional space, and whether there are any missing data areas. The control area of ​​each data point is calculated based on the Voronoi diagram to evaluate the uniformity of data coverage in the horizontal direction and the sampling interval of data in the vertical direction. The density of modeling data points in the horizontal direction and the coverage integrity of data in the vertical direction are evaluated based on the 3D density map.

4. The quality assessment method for geological three-dimensional models according to claim 1, characterized in that, In step S14, the weights of each evaluation index in the three-dimensional geological model evaluation scheme are determined, specifically including: The preprocessed historical assessment data are arranged according to assessment indicators and three-dimensional geological model samples to form a decision matrix. The rows of the decision matrix represent different three-dimensional geological model samples, and the columns represent various assessment indicators. Calculate the entropy value of each evaluation index, that is, calculate the proportion of the standardized value of the i-th three-dimensional geological model sample under the j-th evaluation index to the sum of the standardized values ​​of all three-dimensional geological model samples for the corresponding evaluation index; Based on the entropy value of each evaluation indicator, the weight of each evaluation indicator is calculated, as expressed by: in, For the first The weight of each evaluation indicator, For the first The entropy value of each evaluation indicator This is used to normalize the weights, ensuring that the sum of all weights equals 1.

5. The quality assessment method for geological three-dimensional models according to claim 1, characterized in that, In step S14, the weight allocation of the evaluation indicators is continuously adjusted, specifically including: Historical assessment data, current weight allocation of each assessment indicator, and quality results of the corresponding three-dimensional geological model are integrated into a multi-dimensional vector representing the state space, and the adjustable range of the weight of each assessment indicator is preset. When the weights are adjusted and the three-dimensional geological model is re-evaluated, if the evaluation result is closer to the actual quality judgment, a positive reward will be given; if the evaluation result deviates more from the actual quality, a negative reward will be given. Initialize the policy network by taking the multidimensional vector, the adjustable range of the weights of each evaluation index, and the reward mechanism as inputs to the policy network, and output a new weight allocation scheme, which includes the selection probability of each weight adjustment action.

Images