Machine learning based method and system for three-dimensional geological modeling of fossil localities

Abstract

The application provides a kind of fossil locality three-dimensional geological modeling method and system based on machine learning, it is related to computer three-dimensional modeling technical field, first, the geological observation data set of fossil locality is acquired, then the structure correlation mapping processing is carried out to geological observation data set, the geological structure correlation graph is obtained, to express the correlation of different observation point geological structure description information and the corresponding relationship of fossil distribution and geological structure, then the pre-trained geological structure evolution model is called to carry out three-dimensional space evolution simulation to geological structure correlation graph, generate initial three-dimensional geological framework model, then optimization adjustment is carried out based on initial three-dimensional geological framework model and preset geological constraint rule, obtain the optimized three-dimensional geological model of fossil locality, which contains the spatial distribution boundary of each rock layer, the three-dimensional coordinate range of fossil enrichment area and the spatial correlation of geological structure and fossil distribution, to provide comprehensive and accurate information for geological research.

Description

Technical Field

[0001] This invention relates to the field of computer 3D modeling technology, and more specifically, to a method and system for 3D geological modeling of fossil sites based on machine learning. Background Technology

[0002] In the field of geological research, the study of fossil sites is crucial for understanding Earth's history, biological evolution, and paleoenvironmental changes. Traditional methods of fossil site research primarily rely on field investigations and manual mapping by geologists. This involves detailed recording of the geological structures at various observation points and comprehensive analysis of fossil distribution. However, these methods have several limitations. Firstly, manual recording and mapping are inefficient, requiring significant manpower and time for the collection and processing of large-scale fossil site data. Secondly, traditional methods struggle to comprehensively and accurately represent the three-dimensional geological structure of fossil sites. Geologists are forced to rely on experience and two-dimensional data to construct three-dimensional models in their minds, inevitably leading to biases in the understanding of geological structures and consequently affecting the accurate judgment of fossil distribution patterns and their origins.

[0003] With the development of computer technology, some geological modeling methods based on simple numerical simulations have gradually emerged. However, these methods often only consider limited geological factors, such as focusing only on stratigraphic thickness or the simple morphology of geological structures, failing to fully consider the complex relationships between various elements of the geological structure, as well as the intrinsic connection between fossil distribution and geological structure. Therefore, the established geological models have low accuracy, cannot truly reflect the actual geological conditions of fossil sites, and are difficult to meet the needs of modern geological research for high-precision, three-dimensional visualization geological models. Summary of the Invention

[0004] In view of the aforementioned problems, and in conjunction with the first aspect of the present invention, the present invention provides a method for three-dimensional geological modeling of fossil sites based on machine learning, the method comprising:

[0005] A set of geological observation data of fossil sites is obtained. The set of geological observation data includes geological structure description information of multiple observation points and corresponding fossil distribution records. The geological structure description information of each observation point consists of stratigraphic lithological characteristics, geological structural characteristics and rock layer contact relationship characteristics.

[0006] The geological observation dataset is subjected to structural association mapping to obtain a geological structure association map. The geological structure association map is used to represent the association between geological structure description information of different observation points and the correspondence between fossil distribution records and geological structure description information.

[0007] The pre-trained geological structure evolution model is invoked to perform three-dimensional spatial evolution simulation processing on the geological structure association map to generate an initial three-dimensional geological framework model of the fossil site. The initial three-dimensional geological framework model includes stratigraphic interface distribution characteristics and geological structural spatial distribution characteristics.

[0008] Based on the initial three-dimensional geological framework model and the preset geological constraint rules, the model is optimized and adjusted to obtain the optimized three-dimensional geological model of the fossil site. The geological constraint rules are used to limit the range of stratigraphic thickness variation and geological structural morphology parameters.

[0009] Output the optimized three-dimensional geological model of the fossil site. The optimized three-dimensional geological model of the fossil site includes the spatial distribution boundaries of each rock layer, the three-dimensional coordinate range of the fossil enrichment area, and the spatial correlation between geological structure and fossil distribution.

[0010] In another aspect, the present invention also provides a three-dimensional geological modeling system for fossil sites based on machine learning, including a processor and a machine-readable storage medium connected to the processor. The machine-readable storage medium is used to store programs, instructions or code, and the processor is used to execute the programs, instructions or code in the machine-readable storage medium to implement the above-mentioned method.

[0011] Based on the above, this invention acquires a geological observation dataset containing detailed geological structure descriptions and corresponding fossil distribution records from multiple observation points. By performing structural correlation mapping on this dataset, a geological structure correlation map is obtained. This map clearly presents the correlations between geological structure descriptions at different observation points and the correspondence between fossil distribution records and geological structure descriptions, thus deeply exploring the inherent laws between geological structure and fossil distribution. A pre-trained geological structure evolution model is used to simulate the three-dimensional spatial evolution of the geological structure correlation map, generating an initial three-dimensional geological framework model containing stratigraphic interface distribution characteristics and spatial distribution characteristics of geological structures. This realizes the transformation from two-dimensional data to a three-dimensional model, intuitively displaying the geological structure of fossil sites. Based on the initial three-dimensional geological framework model and preset geological constraint rules, model optimization and adjustment are performed, further improving the model's accuracy and reliability, ensuring that the generated three-dimensional geological model conforms to actual geological conditions. The final optimized 3D geological model of the fossil site not only includes the spatial distribution boundaries of each rock layer and the 3D coordinate range of the fossil-rich area, but also presents the spatial correlation between geological structure and fossil distribution. It provides geological researchers with comprehensive, accurate and intuitive geological information, which helps to deeply understand the geological evolution process, biological evolution law and paleoenvironmental changes of the fossil site. Attached Figure Description

[0012] Figure 1This is a schematic diagram of the execution flow of the three-dimensional geological modeling method for fossil sites based on machine learning provided in an embodiment of the present invention.

[0013] Figure 2 This is a schematic diagram of exemplary hardware and software components of the machine learning-based three-dimensional geological modeling system for fossil sites provided in this embodiment of the invention. Detailed Implementation

[0014] The present invention will now be described in detail with reference to the accompanying drawings. Figure 1 This is a flowchart illustrating a machine learning-based three-dimensional geological modeling method for fossil sites according to an embodiment of the present invention. The following is a detailed description of this machine learning-based three-dimensional geological modeling method for fossil sites.

[0015] Step S110: Obtain a set of geological observation data of the fossil site. The set of geological observation data includes geological structure description information of multiple observation points and corresponding fossil distribution records. The geological structure description information of each observation point consists of stratigraphic lithological characteristics, geological structural characteristics and rock layer contact relationship characteristics.

[0016] In this embodiment, a specific fossil site is selected as the research object, within which multiple observation points are distributed. When acquiring the geological observation data set, geological investigation is required for each observation point. For example, regarding stratigraphic lithology, information such as the rock color, structure, and composition of the rock strata at each observation point needs to be recorded. For instance, some observation points show grayish-white rocks with a clastic structure, mainly composed of quartz and feldspar; others show dark gray rocks with a muddy structure containing more clay minerals.

[0017] Regarding geological structural features, it is necessary to observe and record the folding conditions at each observation point, such as the bending direction of the folds, the dip angle of the rock strata, the presence or absence of faults, the morphology of the fault surfaces, the distribution density and orientation of joints, etc. For example, at one observation point, there are obvious folds, and the rock strata dip in a southeast direction at a relatively large angle; at another observation point, there is a fault with a relatively straight fault surface and obvious displacement of the rock strata on both sides.

[0018] Determining the characteristics of rock strata contact relationships requires identifying the contact type between different rock strata: whether it is a conformable contact, an unconformable contact, or an intrusive contact. For example, if the interface between two adjacent rock strata is smooth and the strata have the same attitude, it is a conformable contact; while in other places, there are obvious erosion surfaces between the rock strata, and the upper and lower rock strata have different attitudes, which is an unconformable contact.

[0019] Simultaneously, each observation point needs to record the corresponding fossil distribution, including the type, quantity, and preservation condition of the fossils. For example, one observation point may have found a large number of trilobite fossils that are relatively well-preserved; another observation point may have found a small number of brachiopod fossils, some of which are somewhat damaged. The information from multiple observation points is then compiled to generate a geological observation data set for the fossil site.

[0020] Step S120: Perform structural association mapping processing on the geological observation data set to obtain a geological structure association map. The geological structure association map is used to represent the association between geological structure description information of different observation points and the correspondence between fossil distribution records and geological structure description information.

[0021] In this embodiment, after obtaining the geological observation data set, structural correlation mapping processing is performed on it. The purpose is to explore the intrinsic connection between geological structures at different observation points, as well as the correspondence between fossil distribution and geological structures, so as to construct a geological structure correlation map that can show these correspondences.

[0022] Step S121: Extract the stratigraphic lithology features, geological structural features, and rock layer contact relationship features from the geological structure description information of each observation point in the geological observation data set, and generate stratigraphic lithology feature set, geological structural feature set, and rock layer contact relationship feature set respectively.

[0023] From the geological observation data set, the stratigraphic lithological characteristics of each observation point are extracted one by one. These stratigraphic lithological characteristics are then compiled and organized to form a stratigraphic lithological characteristic set. This stratigraphic lithological characteristic set contains information such as the color, texture, and composition of rocks from all observation points. For example, the stratigraphic lithological characteristics of observation point A are grayish-white, clastic, and contain quartz and feldspar; those of observation point B are dark gray, argillaceous, and contain clay minerals, etc., and these are all included in the stratigraphic lithological characteristic set.

[0024] Similarly, the geological structural features of each observation point are extracted, such as relevant parameters of folds, characteristics of faults, and joint conditions, and aggregated into a set of geological structural features. For example, the bending direction and dip angle of the folds at observation point C, and the morphology of the fault surface and displacement at observation point D are all included in this set of geological structural features.

[0025] For the characteristics of rock strata contact relationships, the types of rock strata contact involved at each observation point are extracted to form a set of rock strata contact relationship characteristics. For example, information such as the conformable contact at observation point E and the unconformable contact at observation point F are collected into this set of rock strata contact relationship characteristics.

[0026] Step S122: Perform similarity comparison processing on the rock types in the stratigraphic lithology feature set, calculate the matching degree of rock types at different observation points, and generate a stratigraphic lithology similarity matrix. The elements in the stratigraphic lithology similarity matrix represent the degree of similarity of stratigraphic lithology features between two observation points.

[0027] After generating the stratigraphic lithological feature set, it is necessary to compare the similarity of the rock types in it to obtain the degree of similarity between rock types at different observation points.

[0028] Step S1221: Convert the rock type of each observation point in the stratigraphic lithology feature set into a standardized rock type code. The standardized rock type code is generated based on a preset rock classification system, and each rock type corresponds to a unique code value.

[0029] A pre-defined rock classification system is established, covering common rock types and assigning a unique code value to each rock type. For example, granite is coded as G001, sandstone as S002, and shale as Y003, etc.

[0030] Then, based on this classification system, the rock type of each observation point in the stratigraphic lithology feature set is converted into the corresponding standardized code. For example, if the rock at a certain observation point is identified as sandstone, it is converted into code S002; if the rock at another observation point is shale, it is converted into code Y003.

[0031] Step S1222: Construct a rock type matching dictionary, which contains matching weights between different rock types. The matching weights are determined based on the genetic correlation of rock types, and the more closely related the genetic correlation of rock types, the higher the matching weights.

[0032] Analyze the genetic relationships between various rock types and construct a rock type matching dictionary. For example, sandstone and conglomerate both belong to clastic rocks and have a close genetic relationship, so their matching weight is set to a higher value; while sandstone and marble, one is a sedimentary rock and the other is a metamorphic rock, have a relatively distant genetic relationship, so their matching weight is set to a lower value.

[0033] In the dictionary, the rock type code is used as the key, and the corresponding value is the matching weight with other rock type codes. For example, the matching weight of S002 (sandstone) and L004 (conglomerate) is 0.8, and the matching weight of S002 and D005 (marble) is 0.3, etc.

[0034] Step S1223: For any two observation points, extract the standardized rock type codes of the two observation points, query the rock type matching dictionary to obtain the corresponding matching weights, and use the matching weights as the matching degree of the rock types of the two observation points.

[0035] Choose any two observation points, such as observation point A and observation point B, and extract their standardized rock type codes, let's say S002 and L004 respectively. Then look up the matching weights corresponding to these two codes in the rock type matching dictionary, and if we get 0.8, then the rock type matching degree between these two observation points is 0.8.

[0036] For example, the code for observation point C is Y003 and the code for observation point D is D005. The dictionary query yields a matching weight of 0.2, which represents their rock type matching degree.

[0037] Step S1224: According to the numbering order of the observation points, fill the matching degree of rock types between all two observation points into the corresponding positions of the matrix to generate the stratigraphic lithology similarity matrix. The stratigraphic lithology similarity matrix is ​​a square matrix, and the number of rows and columns of the stratigraphic lithology similarity matrix is ​​equal to the number of observation points. The diagonal elements of the stratigraphic lithology similarity matrix represent the matching degree of rock types of the same observation point, and their values ​​are preset maximum values.

[0038] The observation points are numbered in a certain order, such as from 1 to N (N is the total number of observation points). Then, an N×N square matrix is ​​constructed, where the rows and columns of the matrix correspond to the numbers of each observation point.

[0039] For the position in the i-th row and j-th column of the matrix, fill in the rock type matching degree between observation point i and observation point j. For the diagonal position, i.e., the i-th row and i-th column, fill in the preset maximum value, such as 1.0, because the rock types of the same observation point must be completely matched.

[0040] For example, given three observation points, numbered 1, 2, and 3, with matching degrees of 0.8 between 1 and 2, 0.5 between 1 and 3, and 0.6 between 2 and 3, the stratigraphic lithology similarity matrix would be: [

[0042] [1.0, 0.8, 0.5],

[0043] [0.8, 1.0, 0.6],

[0044] [0.5, 0.6, 1.0] ]

[0046] Step S123: Perform correlation analysis on the fold morphology, fault orientation and joint density in the set of geological structural features to determine the causal relationship between the geological structural features of different observation points and generate a geological structural correlation matrix. The elements in the geological structural correlation matrix represent the correlation strength between the geological structural features of two observation points.

[0047] For fold morphology within a set of geological structural features, parameters such as bending direction, axial plane attitude, and limb dip angle of folds at different observation points are compared. If folds at two observation points show strong consistency in these parameters, it indicates that they may belong to the same fold system and are closely related in genesis.

[0048] For the fracture strike, analyze the strike angle and extension direction of the fracture at different observation points. If the strikes are similar and can be roughly connected into a straight line in space, it indicates that these fractures may have the same origin and a high degree of correlation.

[0049] Regarding joint density, the distribution density of joints at different observation points is calculated. If the density values ​​are close and the joint orientation and attitude are similar, it indicates that their geological structural environments may be similar and the correlation is strong.

[0050] Based on these analysis results, a correlation strength value is assigned to each pair of observation points, generating a geological structure correlation matrix. This geological structure correlation matrix is ​​also a square matrix, with the number of rows and columns equal to the number of observation points; the larger the element value, the higher the correlation strength.

[0051] For example, the fold morphology, fault orientation, and joint density of observation points M and N are very similar, and the correlation strength between them is 0.9; while the geological structural characteristics of observation points O and P are quite different, and the correlation strength is 0.2.

[0052] Step S124: Based on the conformable contact identifier, unconformable contact identifier, and intrusive contact identifier in the rock stratum contact relationship feature set, construct a rock stratum contact relationship network for different observation points. In the rock stratum contact relationship network, nodes represent observation points, and edges represent the rock stratum contact relationship type between two observation points.

[0053] The contact identifiers involved in each observation point are extracted from the set of rock strata contact relationship characteristics. For example, conformable contact is represented by Z, unconformable contact by B, and intrusive contact by Q.

[0054] A network is constructed with observation points as nodes. When the rock strata involved by two observation points have a contact relationship, an edge is connected between the corresponding two nodes, and the contact relationship type is marked on the edge.

[0055] For example, if the rock strata at observation point 1 and observation point 2 are in conformable contact, then the edge between them is marked with Z; if the rock strata at observation point 2 and observation point 3 are in unconformable contact, then the edge is marked with B.

[0056] Step S125: Combine the stratigraphic lithology similarity matrix, geological structure correlation matrix, and rock stratum contact relationship network to generate a geological structure correlation map containing node attributes and edge attributes. The node attributes are the geological structure description information of the observation point and the corresponding fossil distribution record, and the edge attributes are the similarity degree in the stratigraphic lithology similarity matrix, the correlation strength in the geological structure correlation matrix, and the contact relationship type in the rock stratum contact relationship network.

[0057] The similarity degree in the stratigraphic lithology similarity matrix and the correlation strength in the geological structure correlation matrix are used as additional attributes of the edges in the stratigraphic contact relationship network, which together with the original contact relationship type constitute the edge attributes.

[0058] Each node's attributes include the stratigraphic and lithological characteristics, geological structural characteristics, rock layer contact relationship characteristics, and corresponding fossil distribution records for that observation point.

[0059] Through the above integration, a complete geological structure correlation map is formed, which can comprehensively show the various correlations between different observation points in terms of geological structure, as well as the correspondence between fossil distribution and geological structure.

[0060] Step S130: Call the pre-trained geological structure evolution model to perform three-dimensional spatial evolution simulation processing on the geological structure association map to generate an initial three-dimensional geological framework model of the fossil site. The initial three-dimensional geological framework model includes stratigraphic interface distribution features and geological structural spatial distribution features.

[0061] The pre-trained geological structure evolution model is trained using a large amount of geological sample data. It can analyze and process the geological structure correlation map and simulate the evolution of geological structures in three-dimensional space.

[0062] The constructed geological structure correlation map is input into the geological structure evolution model. The geological structure evolution model interprets and calculates various information in the geological structure correlation map, and then generates an initial three-dimensional geological framework model. This three-dimensional geological framework model can roughly present the distribution of stratigraphic interfaces and the spatial distribution of geological structures in the fossil site.

[0063] Step S131: Input the geological structure association map into the map parsing layer of the pre-trained geological structure evolution model, parse the node attributes and edge attributes in the geological structure association map, and obtain node attribute feature vectors and edge attribute feature vectors. The node attribute feature vectors contain the geological structure description information of the observation points and the feature values ​​of the corresponding fossil distribution records. The edge attribute feature vectors contain the feature values ​​of the similarity degree in the stratigraphic lithology similarity matrix, the association strength in the geological structure association matrix, and the contact relationship type in the rock layer contact relationship network.

[0064] After the geological structure correlation map enters the map analysis layer of the geological structure evolution model, this map analysis layer will analyze the node attributes and edge attributes in the map one by one. For node attributes, the geological structure description information of the observation point (such as various characteristic parameters of strata lithology, various indicators of geological structure, and types of rock layer contact relationships) and fossil distribution records (fossil types, quantities, etc.) are converted into a series of feature values. These feature values ​​are combined to form the node attribute feature vector.

[0065] For example, the node attribute feature vector of a certain observation point may contain multiple dimensions of values ​​such as rock color feature value, structure feature value, fold tilt angle feature value, fossil quantity feature value, etc.

[0066] For edge attributes, the degree of lithological similarity, the strength of geological structural correlation, and the type of contact relationship are also converted into feature values ​​and combined to form an edge attribute feature vector. For example, the edge attribute feature vector of a certain edge contains feature values ​​of similarity, correlation strength, and contact relationship type.

[0067] Step S132: The spatial correlation modeling layer of the geological structure evolution model is used to perform spatial correlation modeling on the node attribute feature vector and edge attribute feature vector to generate a spatial correlation feature matrix. The spatial correlation feature matrix is ​​used to represent the positional relationship of different observation points in three-dimensional space and the correlation of geological structure features.

[0068] The node attribute feature vector and the edge attribute feature vector are input into the spatial association modeling layer. The role of this spatial association modeling layer is to explore the spatial positional relationship between different observation points and the correlation between their geological structural features.

[0069] Step S1321: Input the node attribute feature vector and edge attribute feature vector into the feature fusion unit of the spatial association modeling layer, and perform element-wise weighted fusion processing on the node attribute feature vector and edge attribute feature vector to generate a fused feature vector. The weights of the element-wise weighted fusion processing are preset according to the degree of influence of node attributes and edge attributes on spatial association.

[0070] The feature fusion unit receives node attribute feature vectors and edge attribute feature vectors, and performs weighted processing on corresponding elements in the two vectors according to preset weights. For example, for a certain element in the node attribute feature vector and the corresponding element in the edge attribute feature vector, they are multiplied by their respective weights and then combined to form an element in the fused feature vector.

[0071] The weights are set based on the influence of node attributes and edge attributes on spatial correlation; those with greater influence have relatively higher weight values. After the above processing, the resulting fused feature vector contains both the attribute information of the node itself and the correlation information with other nodes.

[0072] Step S1322: Perform spatial coordinate mapping processing on the fused feature vector, mapping the feature value of each observation point to the corresponding three-dimensional spatial coordinates, generating a set of spatial feature points with spatial coordinate information, wherein the three-dimensional spatial coordinates are determined based on the geographic coordinate system of the fossil site.

[0073] Based on the geographic coordinate system of the fossil site, a unique three-dimensional spatial coordinate (x, y, z) is determined for each observation point. Then, each feature value in the fused feature vector is correlated with this three-dimensional spatial coordinate, so that each feature value has clear spatial location information.

[0074] For example, if the three-dimensional coordinates of observation point A are (x1, y1, z1), each eigenvalue in its fused feature vector is mapped to these coordinates, forming a spatial feature point. The spatial feature points of all observation points are collected together to form a set of spatial feature points.

[0075] Step S1323: Calculate the Euclidean distance and direction vector between any two spatial feature points in the set of spatial feature points through the distance calculation unit of the spatial association modeling layer, and generate a distance matrix and a direction matrix. The elements in the distance matrix represent the straight-line distance between two spatial feature points, and the elements in the direction matrix represent the direction angle between two spatial feature points.

[0076] The distance calculation unit iterates through all point pairs in the set of spatial feature points and calculates the Euclidean distance between each pair. The calculation method is to derive the straight-line distance from the three-dimensional coordinates of the two points through the corresponding geometric relationships, and then arrange these distance values ​​in point pair order to form a distance matrix.

[0077] Simultaneously, the direction vector between each pair of spatial feature points is calculated, and then the direction angle, such as the angle relative to true north, is determined. These direction angles are then organized into a direction matrix.

[0078] For example, given spatial feature points P(x2, y2, z2) and Q(x3, y3, z3), the Euclidean distance between them is calculated to be d, and the direction angle is θ. Then, fill in d in the corresponding position in the distance matrix and θ in the corresponding position in the direction matrix.

[0079] Step S1324: Combine the distance matrix, direction matrix and fused feature vector to construct a spatial correlation function, which is used to describe the relationship between the feature values ​​of spatial feature points as a function of distance and direction.

[0080] Analyze the relationships between distance values ​​in the distance matrix, direction angles in the direction matrix, and eigenvalues ​​in the fused feature vector. For example, observe the trend of eigenvalue changes as distance increases; and the differences in eigenvalues ​​in different directions.

[0081] Based on these observed relationships, a spatial correlation function is constructed. This spatial correlation function reflects how the eigenvalues ​​between spatial feature points change with their distance and direction; for example, points that are closer together and in the same direction are likely to have more similar eigenvalues.

[0082] Step S1325: Based on the spatial correlation function, interpolate the set of spatial feature points to supplement the feature values ​​of unobserved points, generate a spatial feature grid covering the entire fossil site, and convert the spatial feature grid into a matrix form to obtain a spatial correlation feature matrix. The rows and columns of the spatial correlation feature matrix correspond to the planar coordinates of the fossil site, and the matrix elements correspond to the feature values.

[0083] By using the constructed spatial correlation function, interpolation calculations are performed on areas within the fossil site that lack observation points, inferring the characteristic values ​​of these unobserved points. This method extends the coverage of the characteristic values ​​to the entire fossil site, forming a spatial feature grid.

[0084] Each grid cell in this spatial feature grid corresponds to a specific location within the fossil site and has a corresponding feature value. Representing this spatial feature grid as a matrix yields the spatial correlation feature matrix, where the rows and columns correspond to the planar coordinates (x, y) of the fossil site, and each matrix element is the feature value corresponding to that location.

[0085] Step S133: Call the time dimension extension module of the geological structure evolution model to perform geological evolution time dimension extension processing on the spatial correlation feature matrix, and generate a spatiotemporal evolution feature matrix containing time series information. Each time series element in the spatiotemporal evolution feature matrix represents the geological structure feature state of the corresponding geological period.

[0086] The time dimension extension module can receive a spatial correlation feature matrix. This module contains multiple sub-units for processing time dimension information. First, the sub-units analyze the feature values ​​in the spatial correlation feature matrix to identify key features that reflect the geological evolution process. These key features may include changes in the sedimentary sequence of rock strata, and traces of the formation and alteration of geological structures.

[0087] Next, based on relevant knowledge of geochronology, the fossil sites are divided into different geological periods. For example, in chronological order from ancient times to the present, they are divided into multiple periods such as the Paleozoic, Mesozoic, and Cenozoic, each with its corresponding time interval.

[0088] Then, based on the identified key features and the divided geological periods, the spatial correlation feature matrix is ​​expanded. For each geological period, according to the geological evolution law of that period, the feature values ​​in the spatial correlation feature matrix are adjusted and deduced to obtain the corresponding geological structural feature state of that period.

[0089] These characteristic states from different geological periods are arranged chronologically to form a spatiotemporal evolution characteristic matrix. For example, the first group of elements in the spatiotemporal evolution characteristic matrix corresponds to the geological structural characteristics of the Paleozoic era, the second group corresponds to the Mesozoic era, and so on. Each element contains geological structural characteristic information of various locations of fossil sites within that period.

[0090] Step S134: Using the three-dimensional construction module of the geological structure evolution model, the spatiotemporal evolution feature matrix is ​​divided into three-dimensional grids, dividing the fossil site into multiple three-dimensional grid units, each of which corresponds to a feature value in the spatiotemporal evolution feature matrix.

[0091] The 3D building module can first determine the 3D spatial extent of the fossil site, and then set the mesh generation precision based on the actual geographic coordinates and depth range. The precision setting needs to comprehensively consider both the model's accuracy and computational efficiency; generally, the smaller the mesh cell size, the higher the model precision, but the greater the computational load.

[0092] According to the set precision, the three-dimensional space of the fossil site is divided into a large number of three-dimensional grid units, each with its own specific spatial coordinate range. For example, the coordinate range of a certain grid unit may be x between x4 and x5, y between y4 and y5, and z between z4 and z5.

[0093] Then, the eigenvalues ​​in the spatiotemporal evolution feature matrix are mapped one-to-one with these three-dimensional grid cells. Each grid cell corresponds to an eigenvalue, which reflects the geological structural characteristics of that grid cell in the corresponding geological period.

[0094] Step S135: Determine the stratigraphy, geological structure and rock contact relationship of each three-dimensional grid cell based on the feature value corresponding to each three-dimensional grid cell, and generate an initial three-dimensional geological framework model containing stratigraphic interface distribution features and geological structure spatial distribution features. The stratigraphic interface distribution features represent the location of the interface between different strata in three-dimensional space, and the geological structure spatial distribution features represent the distribution morphology of folds, faults and joints in three-dimensional space.

[0095] After completing the three-dimensional mesh generation, the specific geological properties of each mesh unit can be determined based on its characteristic values, thereby constructing an initial three-dimensional geological framework model.

[0096] Step S1351: Extract the stratigraphic lithology feature components from the feature values ​​corresponding to each three-dimensional grid cell, compare them with the preset lithology feature benchmark library, determine the rock type of the three-dimensional grid cell, mark its stratigraphic lithology attributes, classify adjacent cells with the same attributes into the same lithology cell, and record the coordinate information of each lithology cell.

[0097] The preset lithological characteristic benchmark library contains the characteristic component ranges of various known rock types. For example, the characteristic components of granite may be within a certain range, while the characteristic components of sandstone may be within another range.

[0098] For each 3D grid cell, the stratigraphic lithology feature component is extracted from the feature values ​​and compared with the range of feature components for each rock type in the benchmark library. If the component falls within the range of a certain rock type, the grid cell is labeled as that rock type.

[0099] For example, if the stratigraphic lithological characteristic component of a certain grid cell falls within the characteristic component range of sandstone, it is marked as sandstone. Then, adjacent grid cells with the same sandstone attribute are grouped into one lithological unit, and the coordinate information of all grid cells contained in that lithological unit is recorded.

[0100] Step S1352: Extract the geological structural feature components from the feature values, identify fold feature parameters, fault feature parameters, and joint feature parameters, determine the fold location to which the unit belongs based on the fold feature parameters, and classify units that are in the same fold location and are spatially continuous into the same fold structural unit; determine whether the unit belongs to a fault zone based on the fault feature parameters, and classify units distributed along the fault extension trajectory into the same fault structural unit; calculate the degree of joint development of the unit based on the joint feature parameters, classify units with the same degree of development and that are adjacent into the same joint structural unit, and record the coordinate information and corresponding feature parameters of each structural unit.

[0101] Geological structural feature components are extracted from the feature values ​​of each grid cell. These components include feature parameters related to folds, faults, joints, etc.

[0102] Fold characteristic parameters may include the axial direction of the fold and the inclination angle of the wing. Based on these parameters, it can be determined whether a mesh element belongs to the core, wing, or turning point of the fold. Mesh elements belonging to the same fold region and spatially continuous are grouped into a single fold construction unit, and the coordinate information and corresponding fold characteristic parameters of this unit are recorded.

[0103] Fracture characteristic parameters may include the strike, dip angle, and displacement of the fracture surface. Based on these parameters, it is determined whether a mesh element is located within a fracture zone. Mesh elements distributed along the fracture extension trajectory and belonging to the fracture zone are grouped into a single fracture structural element, and their coordinate information and fracture characteristic parameters are recorded.

[0104] Joint characteristic parameters may include joint density, orientation, and length. The degree of joint development is calculated based on these parameters, and a comprehensive index is obtained by integrating these parameters. Adjacent mesh cells with the same degree of development are grouped into a single joint construction unit, and their coordinate information and joint characteristic parameters are recorded.

[0105] Step S1353: Extract the rock layer contact relationship feature components from the feature values, identify the contact type identifier, determine the contact relationship type between adjacent three-dimensional mesh units, extract the contact surfaces of adjacent units with the same contact type and located at the boundary of lithological units as rock layer contact interfaces, and record the coordinates and contact relationship type of the contact interfaces.

[0106] Extract the rock layer contact relationship feature components from the feature values ​​of each grid cell, and identify the contact type identifiers, such as the corresponding identifiers for conformable contact, unconformable contact, and intrusive contact.

[0107] Examine the contact relationship between each grid cell and its adjacent grid cells, and determine the contact relationship type based on the contact type identifiers of both. For example, if the contact identifier of grid cell A is conformable contact, and the contact identifier of the adjacent grid cell B is also conformable contact, and they are located at the boundary of different lithological cells, then the contact surface between them is extracted as a conformable contact interface, and the coordinates of the interface and the contact relationship type are recorded.

[0108] Step S1354: Perform spatial overlay analysis on all lithological units, determine the vertical positional relationship based on the three-dimensional coordinate depth, divide the stratigraphic units by combining the rock layer contact interface, each stratigraphic unit is composed of lithological units with the same vertical positional relationship and connected by the integrated contact interface, and record the composition and distribution range of the stratigraphic units.

[0109] All lithological units are ordered according to their depth in three-dimensional coordinates to determine their vertical relationship. For example, lithological units with smaller depths are located above, and those with larger depths are located below.

[0110] Based on the previously extracted rock strata contact interfaces, lithological units that are connected by conformable contact interfaces and have consistent vertical relationships are classified as a single stratigraphic unit. For example, if lithological unit 1 is located above and lithological unit 2 is located below, and they are connected by conformable contact interfaces with consistent vertical relationships throughout the distribution range, they are classified as the same stratigraphic unit. The composition of this stratigraphic unit and its three-dimensional coordinate range are then recorded.

[0111] Step S1355: Extract the top and bottom surfaces of each stratigraphic unit as stratigraphic interfaces. The top surface is the contact interface with the smallest depth in the unit, and the bottom surface is the contact interface with the largest depth. Record the coordinates of the stratigraphic interfaces to form the stratigraphic interface distribution characteristics.

[0112] For each stratigraphic unit, among its contact interfaces, identify the interface with the shortest depth as the top surface of the stratigraphic unit and the interface with the longest depth as the bottom surface.

[0113] Record the three-dimensional coordinate information of the top and bottom surfaces. This coordinate information together constitutes the stratigraphic interface distribution characteristics, which show the location of the interface between different strata in three-dimensional space.

[0114] Step S1356: Integrate the coordinate information and characteristic parameters of fold, fault and joint structural units to determine their three-dimensional spatial distribution morphology and form the spatial distribution characteristics of geological structures.

[0115] The coordinate information and characteristic parameters of the previously divided fold, fracture, and joint structural units are integrated. By analyzing this information, the three-dimensional spatial distribution morphology of folds, fractures, joints, and their orientation are reconstructed.

[0116] For example, connecting the coordinates of folded structural units reveals the overall bending trend of the folds; connecting the coordinates of faulted structural units in series reveals the extension path of the faults in three-dimensional space. The integrated information forms the spatial distribution characteristics of geological structures.

[0117] Step S1357: Integrate the distribution characteristics of stratigraphic interfaces and the spatial distribution characteristics of geological structures to ensure spatial consistency, supplement discontinuous areas, and make the stratigraphic interfaces and geological structures present a continuous distribution state, combining them to form an initial three-dimensional geological framework model.

[0118] The distribution characteristics of stratigraphic interfaces and the spatial distribution characteristics of geological structures are correlated to check whether they are consistent in spatial location. For example, when a fold structure passes through a stratigraphic interface, the morphology of the stratigraphic interface should match the morphology of the fold.

[0119] For areas with discontinuities, reasonable supplementation should be made based on the surrounding characteristics and trends to ensure that the stratigraphic interfaces and geological structures are continuously distributed throughout the fossil site.

[0120] Through the above association, integration and supplementation, the distribution characteristics of stratigraphic interfaces and the spatial distribution characteristics of geological structures are combined to form an initial three-dimensional geological framework model.

[0121] Step S140: Based on the initial three-dimensional geological framework model and the preset geological constraint rules, perform model optimization and adjustment to obtain an optimized three-dimensional geological model of the fossil site. The geological constraint rules are used to limit the range of stratigraphic thickness variation and geological structural morphology parameters.

[0122] The initial three-dimensional geological framework model may contain some inconsistencies with actual geological laws. It needs to be optimized and adjusted according to the preset geological constraints to improve the accuracy and rationality of the model.

[0123] Step S141: Obtain preset geological constraint rules, which include constraints on the range of stratigraphic thickness variation, constraints on geological structural morphology parameters, and constraints on the correlation between fossil distribution and geological structure. The constraints on the range of stratigraphic thickness variation limit the maximum and minimum thickness of different strata. The constraints on geological structural morphology parameters limit the range of curvature of folds, the range of extension length of faults, and the range of density of joints. The constraints on the correlation between fossil distribution and geological structure limit the correspondence between specific fossil types and the corresponding lithology and geological structure of strata.

[0124] The pre-defined geological constraint rules are based on extensive geological research and practical experience. The constraint on the range of stratum thickness variation specifies the maximum and minimum achievable thickness for different strata; for example, the thickness of a certain stratum cannot be less than h1 or greater than h2.

[0125] Geological structural morphological parameters constrain the curvature of folds, the extension length of faults, and the density of joints, setting corresponding ranges. For example, the curvature of folds should be between c1 and c2, and the extension length of faults should be between l1 and l2.

[0126] The correlation between fossil distribution and geological structure clarifies the strata lithology and geological structural environment in which a particular fossil species usually appears. For example, a certain type of trilobite fossil is mostly distributed in sandstone strata and in areas close to the folds and limbs.

[0127] Step S142: Input the initial three-dimensional geological framework model and the geological constraint rules into the model optimization module, extract the stratigraphic interface distribution features and geological structural spatial distribution features in the initial three-dimensional geological framework model, and obtain the feature set to be optimized.

[0128] After receiving the initial 3D geological framework model and geological constraint rules, the model optimization module can analyze the model and extract the stratigraphic interface distribution features and the spatial distribution features of geological structures. These features are the objects that need to be optimized and adjusted, and together they constitute the set of features to be optimized.

[0129] Step S143: Calculate the formation thickness in the set of features to be optimized, compare the calculated formation thickness with the maximum and minimum values ​​in the formation thickness variation range constraint, and mark the formation areas that exceed the constraint range as thickness anomaly areas.

[0130] The thickness of each stratigraphic unit is calculated based on the distribution characteristics of the stratigraphic interfaces. The calculation method is to subtract the average depth of the top surface from the average depth of the bottom surface of the stratigraphic unit.

[0131] The calculated thickness is compared with the maximum and minimum values ​​in the stratigraphic thickness variation range constraint. If the thickness of a stratigraphic unit is less than the minimum value or greater than the maximum value, the area where the stratigraphic unit is located is marked as a thickness anomaly area.

[0132] Step S144: Extract the geological structural morphological parameters from the feature set to be optimized, compare the extracted curvature of folds, extension length of fractures and density of joints with the corresponding ranges in the constraints of the geological structural morphological parameters, and mark the geological structural regions that exceed the constraints as structural anomaly regions.

[0133] Parameters such as the curvature of folds, the extension length of faults, and the density of joints are extracted from the spatial distribution characteristics of geological structures.

[0134] These parameters are compared with the corresponding ranges in the geological structure morphology parameter constraints. If the curvature of a fold is not between the specified c1 and c2, or the extension length of a fault exceeds the range of l1 to l2, or the density of joints in a region is not within the set range, then the region where the geological structure is located is marked as a structural anomaly region.

[0135] Step S145: Analyze the correspondence between fossil distribution records and geological structure description information in the initial three-dimensional geological framework model, compare it with the correspondence in the fossil distribution and geological structure association constraints, and mark the areas that do not meet the association constraints as association anomaly areas.

[0136] View the fossil distribution records for each region in the initial 3D geological framework model, as well as the corresponding geological structure description information (such as stratigraphy, geological structure type, etc.).

[0137] The above correspondence is compared with the content of the fossil distribution and geological structure association constraint. If the fossil species in a certain area do not match the stratigraphy or geological structure of that area, for example, fossils that are usually only found in sandstone strata are found in shale strata, then the area is marked as an anomalous association area.

[0138] Step S146: Based on the thickness anomaly region, structural anomaly region and associated anomaly region, perform local adjustment processing on the initial three-dimensional geological framework model, adjust the stratigraphic interface distribution characteristics and geological structural spatial distribution characteristics of the anomaly region, so that the adjusted characteristics conform to the geological constraint rules.

[0139] For different types of anomalous areas, corresponding adjustment measures should be taken to make them conform to geological constraint rules.

[0140] Step S1461: For the thickness anomaly region, calculate the difference between the current formation thickness of the thickness anomaly region and the maximum and minimum values ​​in the formation thickness variation range constraint, and determine the adjustment direction and adjustment range of the formation interface based on the difference. The adjustment direction and adjustment range are set based on the sign and magnitude of the difference.

[0141] For areas with abnormal thickness, if the current thickness is greater than the maximum value, calculate the difference between the current thickness and the maximum value; this difference is positive. Based on the magnitude of the difference, determine whether the top surface of the formation needs to be raised or the bottom surface lowered, and by how much. The larger the difference, the greater the adjustment.

[0142] If the current thickness is less than the minimum value, calculate the difference between it and the minimum value; this difference is negative. Similarly, based on the magnitude of the difference, determine the direction and extent to which the top surface of the formation is lowered or the bottom surface is raised.

[0143] Step S1462: Move and adjust the stratigraphic interface distribution characteristics of the thickness anomaly area according to the adjustment direction and adjustment range, so that the adjusted stratigraphic thickness is within the constraint of the stratigraphic thickness variation range, and record the three-dimensional coordinates of the adjusted stratigraphic interface as the updated stratigraphic interface distribution characteristics.

[0144] Based on the determined adjustment direction and magnitude, the top or bottom surface of the strata in areas of abnormal thickness is moved. For example, for areas with excessive thickness, the top surface is moved upwards by a certain distance, or the bottom surface is moved downwards by a certain distance, so that the adjusted strata thickness is within the constraints.

[0145] After the adjustment is completed, the new three-dimensional coordinates of the stratigraphic interface are recorded as the updated stratigraphic interface distribution characteristics.

[0146] Step S1463: For the structurally anomalous area, calculate the deviation value between the current geological structural morphological parameters and the constraint range in the geological structural morphological parameter constraints, and determine the deformation adjustment amount of the geological structure based on the deviation value. The deformation adjustment amount includes the curvature adjustment amount of folds, the extension length adjustment amount of fractures, and the density adjustment amount of joints.

[0147] For wrinkles in the constructed anomalous region, calculate the deviation of their current curvature from the upper or lower limit of the constraint range. If the curvature is greater than the upper limit, the deviation is positive, and the curvature needs to be reduced, determining the corresponding curvature adjustment amount; if it is less than the lower limit, the deviation is negative, and the curvature needs to be increased, determining the adjustment amount.

[0148] For fractures, calculate the deviation between the extended length and the constraint range, and determine the length adjustment amount that needs to be shortened or extended based on the deviation value.

[0149] For joints, calculate the deviation between their density and the constraint range, and determine the amount of density adjustment required to increase or decrease the number of joints.

[0150] Step S1464: Adjust the spatial distribution characteristics of the geological structure in the structural anomaly area according to the deformation adjustment amount, so that the adjusted geological structure morphology parameters are within the constraints of the geological structure morphology parameters, and record the three-dimensional morphology of the adjusted geological structure as the updated spatial distribution characteristics of the geological structure.

[0151] Based on the determined deformation adjustment amount, the geological structure of the tectonic anomaly area is adjusted. For example, folds are deformed to bring their curvature within the constraints; the extension trajectory of faults is modified to ensure their length meets the requirements; and the distribution of joints is adjusted to change their density.

[0152] After adjustment, the new three-dimensional morphology of the geological structure is recorded as the updated spatial distribution characteristics of the geological structure.

[0153] Step S1465: For the associated abnormal region, analyze the reason why the fossil distribution record and geological structure description information of the associated abnormal region do not meet the association constraints. If the correspondence between the fossil distribution record and the geological structure description information is incorrect, then correct the correspondence of the associated abnormal region to make it consistent with the correspondence in the fossil distribution and geological structure association constraints.

[0154] Analyze the associated abnormal areas to see if the error is due to incorrect fossil distribution records, incorrect geological structure descriptions, or an incorrect correspondence between the two.

[0155] If the correspondence is incorrect, such as incorrectly assigning a fossil to an unrelated geological structure, then the above correspondence should be corrected to comply with the provisions of the association constraint, such as correctly assigning the fossil to the stratigraphic lithology and geological structure environment in which it usually occurs.

[0156] Step S1466: After adjusting the thickness anomaly region, structural anomaly region and associated anomaly region, perform smooth transition processing on the stratigraphic interface distribution characteristics and geological structural spatial distribution characteristics around the adjusted region to make the geological structural characteristics of the adjusted region and the surrounding region continuous and consistent.

[0157] The adjusted area may have obvious boundaries or abrupt changes with the surrounding areas, requiring smooth transition processing. For example, the adjusted stratigraphic interface and the surrounding unadjusted interface are connected by a gradual transition to make the interface morphology continuous; for geological structures, the parameters of their edge parts are adjusted to gradually integrate them with the characteristics of the surrounding structures.

[0158] Step S1467: Perform an overall check on the initial three-dimensional geological framework model after smoothing the transition, and verify whether the adjusted stratigraphic thickness, geological structural morphology parameters, fossil distribution and geological structure correlation all conform to the geological constraint rules. If there are any non-compliance issues, redetermine the abnormal areas and make adjustments until all features conform to the geological constraint rules.

[0159] A comprehensive review of the model after smoothing was conducted, the stratigraphic thickness was recalculated, geological structural morphology parameters were extracted, and the correlation between fossil distribution and geological structure was compared again to see if it conformed to geological constraints.

[0160] If, during the inspection process, it is found that the thickness of a certain stratigraphic unit still exceeds the constraint range, or the morphological parameters of a certain geological structure do not meet the specified standards, or there is a mismatch between the fossil distribution and the geological structure, then these areas need to be re-marked as abnormal areas and adjusted again according to the process of steps S1461 to S1466.

[0161] For example, during an overall inspection, if it is found that the curvature of a certain adjusted fold structure has improved compared to before, but is still not completely within the constraint range, it is necessary to recalculate the deviation value between the curvature of the fold and the constraint range, determine the new deformation adjustment amount, and adjust it again, while smoothing the transition of the surrounding area.

[0162] This process is repeated until all strata thicknesses, geological structural parameters, and the relationship between fossil distribution and geological structure in the model fully conform to the geological constraint rules. The model obtained at this point is the optimized three-dimensional geological model of the fossil site.

[0163] Step S150: Output the optimized three-dimensional geological model of the fossil site. The optimized three-dimensional geological model of the fossil site includes the spatial distribution boundaries of each rock layer, the three-dimensional coordinate range of the fossil enrichment area, and the spatial correlation between geological structure and fossil distribution.

[0164] After the model has been optimized and adjusted, the optimized three-dimensional geological model of the fossil site needs to be output for subsequent research, analysis and application.

[0165] Step S151: Extract the spatial distribution boundaries of each rock layer in the optimized three-dimensional geological model of the fossil site, and convert the spatial distribution boundary of each rock layer into a three-dimensional polygonal mesh surface. Each three-dimensional polygonal mesh surface is composed of multiple triangular facets, and the vertex coordinates of each triangular facet are determined based on the geographic coordinate system of the fossil site.

[0166] By traversing the optimized three-dimensional geological model of the fossil site, the spatial distribution boundary of each rock layer is identified. The above boundary is the interface between the rock layer and other rock layers or geological bodies, and usually presents a complex three-dimensional morphology.

[0167] To represent these boundaries more accurately, the spatial distribution boundaries of each rock layer are converted into a three-dimensional polygonal mesh. During the conversion, the boundary surface is divided into multiple triangular patches because triangles are stable and can better fit complex curved surface shapes.

[0168] Each triangular facet has three distinct three-dimensional coordinates for its three vertices. These coordinates are determined based on the geographic coordinate system of the fossil site, ensuring the accuracy of the grid facet's spatial location. For example, the spatial distribution boundary of a rock stratum is transformed into a three-dimensional polygonal grid composed of hundreds of triangular facets. The vertex coordinates of each triangular facet (x6, y6, z6), (x7, y7, z7), (x8, y8, z8), etc., correspond to specific locations on the boundary of that rock stratum.

[0169] Step S152: Extract the three-dimensional coordinate range of the fossil enrichment area in the optimized three-dimensional geological model of the fossil site, determine the minimum circumscribed cuboid of each fossil enrichment area, and record the three-dimensional coordinates of the eight vertices of the minimum circumscribed cuboid as the boundary coordinates of the fossil enrichment area.

[0170] In the optimized model, based on the fossil distribution record, areas with a large number of fossils and a relatively concentrated distribution are identified, namely fossil-rich areas.

[0171] For each fossil-rich region, the smallest circumscribed cuboid that can completely enclose the region is determined by calculation. The calculation method is to find the maximum and minimum values ​​of the region in the x, y, and z coordinate axes, and use these values ​​as the boundaries of the cuboid to obtain the smallest circumscribed cuboid.

[0172] The three-dimensional coordinates of the eight vertices of the smallest circumscribed cuboid are recorded. These three-dimensional coordinates together constitute the boundary coordinates of the fossil enrichment region, defining the extent of the fossil enrichment region in three-dimensional space. For example, the coordinates of the eight vertices of the smallest circumscribed cuboid of a certain fossil enrichment region are (x9, y9, z9), (x10, y10, z9), (x9, y10, z9), (x10, y10, z9), (x9, y9, z10), (x10, y10, z10), (x9, y10, z10), (x10, y9, z10).

[0173] Step S153: Extract the spatial correlation between geological structures and fossil distribution in the optimized three-dimensional geological model of the fossil site, and generate a correlation list, which contains the distribution probability of each geological structure type and the corresponding fossil species.

[0174] The spatial relationship between geological structures and fossil distribution in the optimized three-dimensional geological model of fossil sites was analyzed, and the types and quantities of fossils appearing in different geological structure types (such as the core and limbs of folds, near fault zones, and different joint development areas) were counted.

[0175] Based on the statistical results, the distribution probability of each geological structure type and its corresponding fossil species is calculated, that is, the proportion of a certain fossil in a certain geological structure type to the total number of fossils in that geological structure type.

[0176] These geological structure types, corresponding fossil species, and distribution probabilities are compiled into a correlation list. For example, the distribution probability of folded limbs with trilobite fossils is 0.7, and the distribution probability of brachiopod fossils near fault zones is 0.6, etc.

[0177] Step S154: Integrate the three-dimensional polygonal mesh surface, the boundary coordinates of the fossil-rich area, and the correlation list to generate a model output file containing spatial data and attribute data. The spatial data is the boundary coordinates of the three-dimensional polygonal mesh surface and the fossil-rich area, and the attribute data is the correlation list and the stratigraphic lithological characteristics and geological structural characteristics of each rock layer.

[0178] The three-dimensional polygonal mesh surfaces of each rock layer obtained in step S151, the boundary coordinates of the fossil enrichment area obtained in step S152, and the association list generated in step S153 are integrated.

[0179] Among them, the three-dimensional polygonal mesh surface and the boundary coordinates of the fossil-rich area are spatial data, which describe the spatial location and morphology of various geological bodies and fossil-rich areas in the model.

[0180] The list of relationships, as well as the stratigraphic and lithological characteristics of each rock layer (such as rock type, color, structure, etc.) and geological structural characteristics (such as fold parameters, fault parameters, joint parameters, etc.), belong to attribute data, which supplement the geological attribute information corresponding to the spatial data.

[0181] These spatial and attribute data are organized in a specific format to generate a model output file, which fully contains all the information of the optimized three-dimensional geological model of the fossil site.

[0182] Step S155: Perform format conversion processing on the model output file to convert it into a preset 3D model format, which supports display and interactive operation in 3D visualization software.

[0183] Since different 3D visualization software may support different model formats, it is necessary to convert the output file format to ensure that the model can be displayed and interacted with normally in commonly used 3D visualization software.

[0184] The default 3D model format can be a widely used format such as .obj, .stl, .3ds, etc., which are supported by most 3D visualization software.

[0185] During the conversion process, the integrity and accuracy of spatial and attribute data are maintained to ensure that the converted model can accurately present the spatial distribution of each rock layer, the range of fossil enrichment areas, and related attribute information in 3D visualization software.

[0186] Step S156: Store the converted model output file to the specified storage location and generate a model output log. The model output log records the time of output of the three-dimensional geological model of the fossil site, the storage location, and the information on the number of rock layers and fossil types contained in the three-dimensional geological model of the fossil site.

[0187] Save the converted model output file to the specified storage location, which can be the local computer's hard drive, the network server's storage space, etc.

[0188] Simultaneously, a model output log is generated, which records detailed information such as the specific time of model output (year, month, day, hour, and minute); the storage path of the model output file (i.e., the specific folder location); the number of rock strata included in the model (e.g., 20 rock strata); and the number of fossil species (e.g., 5 fossil species). The model output log helps in model management and traceability, facilitating subsequent retrieval and use of the model.

[0189] Step S157: Load the converted model output file into the 3D visualization platform through the 3D visualization interface, and display the optimized 3D geological model of the fossil site in the 3D visualization platform. Users can rotate, scale, and cut the optimized 3D geological model of the fossil site to view the geological structure and fossil distribution from different perspectives.

[0190] Using a 3D visualization interface, the converted model output file is loaded into a 3D visualization platform. The platform can parse the spatial and attribute data from the model file and generate corresponding 3D graphics, intuitively displaying the optimized 3D geological model of the fossil site.

[0191] Users can use the interactive functions provided by the 3D visualization platform to rotate the model and observe the geological structure and fossil distribution from different angles; zoom in or out to view the overall shape or local details; and cut along a plane to view the internal geological structure and fossil distribution.

[0192] Through these interactive operations, researchers, geologists, and other users can gain a deeper understanding of the geological structure characteristics of fossil sites and the distribution patterns of fossils.

[0193] To ensure that the geological structure evolution model can accurately process and analyze the geological structure correlation map and generate a reliable initial three-dimensional geological framework model, the model needs to be pre-trained.

[0194] Step S211: Collect geological sample data, which includes geological observation data from different fossil sites, corresponding geological structure correlation maps, and known three-dimensional geological models.

[0195] A large amount of geological sample data was collected from various geological research literature, geological exploration reports, geological databases, etc. The above sample data should cover fossil site information under different regions and geological conditions.

[0196] Each geological sample should include detailed geological observation data of the fossil site, such as stratigraphic lithology, geological structure, rock contact relationship characteristics, and fossil distribution records at multiple observation points; a geological structure correlation map constructed based on these observation data; and an existing, validated, known three-dimensional geological model of the fossil site, which serves as the target value for training.

[0197] Step S212: Preprocess the collected geological sample data, including data cleaning, standardization, and feature extraction.

[0198] Data cleaning removes noisy, duplicate, and erroneous data from the sample data. For example, it deletes samples with incomplete or contradictory observations.

[0199] Standardization involves converting geological parameters from different sample data according to a unified standard to make them comparable. For example, it involves converting descriptions of rock types from different samples into standardized codes and unifying the units of geological structural parameters.

[0200] Feature extraction involves extracting features from preprocessed data that are meaningful for model training, such as key features of stratigraphy and lithology, and important parameters of geological structures. These features will serve as input features for model training.

[0201] Step S213: Construct the network architecture of the geological structure evolution model, which includes a map parsing layer, a spatial correlation modeling layer, a time dimension extension module, and a three-dimensional construction module.

[0202] The graph parsing layer is used to parse the input geological structure association graph, extract node attribute feature vectors and edge attribute feature vectors, and contains neural network units for processing node attributes and edge attributes.

[0203] The spatial association modeling layer is used to model the spatial association between node attribute feature vectors and edge attribute feature vectors, and generate a spatial association feature matrix. This spatial association modeling layer includes sub-modules such as feature fusion unit and distance calculation unit, and each sub-module is composed of a corresponding neural network layer.

[0204] The time dimension extension module is used to extend the spatial correlation feature matrix in terms of time dimension, generating a spatiotemporal evolution feature matrix. It contains a recurrent neural network unit that processes time series information.

[0205] The 3D construction module is used to perform 3D meshing and geological attribute determination on the spatiotemporal evolution feature matrix, and generate an initial 3D geological framework model. This 3D construction module contains a neural network structure for meshing and attribute judgment.

[0206] Step S214: Use the geological structure association map in the preprocessed geological sample data as input and the known three-dimensional geological model as output label to train the constructed geological structure evolution model.

[0207] During the training process, the geological structure correlation map of each geological sample is input into the geological structure evolution model. After processing at each layer, the model generates a predicted three-dimensional geological model.

[0208] The predicted 3D geological model is compared with the known 3D geological model (output labels), and the loss value between the two is calculated. The loss value reflects the difference between the predicted model and the real model.

[0209] Based on the loss value, the parameters of each layer of the model, such as the weights and biases of the neural network, are adjusted using the backpropagation algorithm to reduce the loss value.

[0210] Step S215: Set the number of training iterations and the learning rate. During the training process, periodically verify the performance of the model. When the prediction accuracy of the model reaches the preset threshold, stop training to obtain the pre-trained geological structure evolution model.

[0211] Set an appropriate number of iterations, which is the number of times the parameters are updated using training samples during model training. Simultaneously, set a learning rate, which determines the extent of parameter adjustments. An excessively large learning rate may lead to model instability, while a learning rate that is too small will slow down the training process.

[0212] During training, at regular intervals, validation samples are used to verify the model's performance, and the model's prediction accuracy on the validation samples is calculated.

[0213] When the model's prediction accuracy on the validation samples reaches a preset threshold, such as 90%, it indicates that the model has good predictive ability. At this point, training is stopped, the model parameters are saved, and a pre-trained geological structure evolution model is obtained.

[0214] Figure 2 Schematic diagrams are shown of exemplary hardware and software components of a machine learning-based 3D geological modeling system 100 for fossil sites, which can implement the ideas of this application, according to some embodiments of this application. For example, a processor 120 can be used on the machine learning-based 3D geological modeling system 100 for fossil sites and to perform the functions in this application.

[0215] The machine learning-based 3D geological modeling system 100 for fossil sites can be a general-purpose server or a special-purpose server, both of which can be used to implement the machine learning-based 3D geological modeling method for fossil sites of this application. Although only one server is shown in this application, for convenience, the functions described in this application can be implemented in a distributed manner on multiple similar platforms to balance the load.

[0216] For example, a machine learning-based 3D geological modeling system 100 for fossil sites may include a network port 110 connected to a network, one or more processors 120 for executing program instructions, a communication bus 130, and various forms of storage media 140, such as a disk, ROM, or RAM, or any combination thereof. Exemplarily, the machine learning-based 3D geological modeling system 100 for fossil sites may also include program instructions stored in ROM, RAM, or other types of non-transitory storage media, or any combination thereof. The methods of this application can be implemented according to these program instructions. The machine learning-based 3D geological modeling system 100 for fossil sites also includes an I / O interface 150 between the computer and other input / output devices.

[0217] For ease of explanation, only one processor is described in the machine learning-based 3D geological modeling system for fossil sites 100. However, it should be noted that the machine learning-based 3D geological modeling system for fossil sites 100 of this application may also include multiple processors, and therefore the steps performed by one processor described in this application may also be performed jointly or individually by multiple processors. For example, if the processor of the machine learning-based 3D geological modeling system for fossil sites 100 performs steps A and B, it should be understood that steps A and B may also be performed jointly by two different processors or individually by one processor. For example, the first processor performs step A, the second processor performs step B, or the first processor and the second processor jointly perform steps A and B.

[0218] Furthermore, this embodiment of the invention also provides a readable storage medium, wherein computer-executable instructions are preset in the readable storage medium, and when the processor executes the computer-executable instructions, the above-mentioned three-dimensional geological modeling method for fossil sites based on machine learning is implemented.

[0219] It should be noted that, in order to simplify the description of the present invention and thus help to understand one or more embodiments of the invention, multiple features may sometimes be grouped into one embodiment, drawing or description thereof in the foregoing description of the embodiments of the present invention.

Claims

1. A method for three-dimensional geological modeling of fossil sites based on machine learning, characterized in that, The method includes: A set of geological observation data of fossil sites is obtained. The set of geological observation data includes geological structure description information of multiple observation points and corresponding fossil distribution records. The geological structure description information of each observation point consists of stratigraphic lithological characteristics, geological structural characteristics and rock layer contact relationship characteristics. The geological observation dataset is subjected to structural association mapping to obtain a geological structure association map. The geological structure association map is used to represent the association between geological structure description information of different observation points and the correspondence between fossil distribution records and geological structure description information. A pre-trained geological structure evolution model is invoked to perform three-dimensional spatial evolution simulation processing on the geological structure association map, generating an initial three-dimensional geological framework model of the fossil site. The initial three-dimensional geological framework model includes stratigraphic interface distribution features and geological structural spatial distribution features. The process includes: inputting the geological structure association map into the map parsing layer of the pre-trained geological structure evolution model, parsing the node attributes and edge attributes in the geological structure association map, and obtaining node attribute feature vectors and edge attribute feature vectors. The node attribute feature vectors contain geological structure description information of the observation points and feature values ​​of the corresponding fossil distribution records. The edge attribute feature vectors contain feature values ​​of the similarity degree in the stratigraphic lithology similarity matrix, the association strength in the geological structural association matrix, and the contact relationship type in the rock layer contact relationship network. The spatial correlation modeling layer of the geological structure evolution model performs spatial correlation modeling on the node attribute feature vector and edge attribute feature vector to generate a spatial correlation feature matrix. The spatial correlation feature matrix is ​​used to represent the positional relationship of different observation points in three-dimensional space and the correlation of geological structure features. The time dimension extension module of the geological structure evolution model is invoked to extend the spatial correlation feature matrix in terms of geological evolution time dimension, generating a spatiotemporal evolution feature matrix containing time series information. Each time series element in the spatiotemporal evolution feature matrix represents the geological structure feature state of the corresponding geological period. The spatiotemporal evolution feature matrix is ​​divided into three-dimensional grids using the three-dimensional construction module of the geological structure evolution model, dividing the fossil site into multiple three-dimensional grid units, each of which corresponds to a feature value in the spatiotemporal evolution feature matrix. Based on the characteristic values ​​corresponding to each three-dimensional grid cell, the stratigraphy, geological structure, and rock contact relationship of the three-dimensional grid cell are determined, and an initial three-dimensional geological framework model containing stratigraphic interface distribution characteristics and geological structure spatial distribution characteristics is generated. The stratigraphic interface distribution characteristics represent the position of the interface between different strata in three-dimensional space, and the geological structure spatial distribution characteristics represent the distribution morphology of folds, faults, and joints in three-dimensional space. Based on the initial three-dimensional geological framework model and the preset geological constraint rules, the model is optimized and adjusted to obtain the optimized three-dimensional geological model of the fossil site. The geological constraint rules are used to limit the range of stratigraphic thickness variation and geological structural morphology parameters. Output the optimized three-dimensional geological model of the fossil site. The optimized three-dimensional geological model of the fossil site includes the spatial distribution boundaries of each rock layer, the three-dimensional coordinate range of the fossil enrichment area, and the spatial correlation between geological structure and fossil distribution.

2. The method for three-dimensional geological modeling of fossil sites based on machine learning according to claim 1, characterized in that, The process of performing structural correlation mapping on the geological observation data set to obtain a geological structural correlation map includes: Extract the stratigraphic lithological features, geological structural features, and rock layer contact relationship features from the geological structure description information of each observation point in the geological observation data set, and generate stratigraphic lithological feature set, geological structural feature set, and rock layer contact relationship feature set, respectively; The rock types in the stratigraphic lithology feature set are compared for similarity. The matching degree of rock types at different observation points is calculated to generate a stratigraphic lithology similarity matrix. The elements in the stratigraphic lithology similarity matrix represent the degree of similarity of the stratigraphic lithology features of two observation points. Correlation analysis is performed on the fold morphology, fault orientation and joint density in the set of geological structural features to determine the causal relationship between geological structural features at different observation points and generate a geological structural correlation matrix. The elements in the geological structural correlation matrix represent the correlation strength between the geological structural features of two observation points. Based on the conformable contact identifier, unconformable contact identifier, and intrusive contact identifier in the rock stratum contact relationship feature set, a rock stratum contact relationship network is constructed for different observation points. In the rock stratum contact relationship network, nodes represent observation points, and edges represent the rock stratum contact relationship type between two observation points. By combining the stratigraphic lithology similarity matrix, the geological structure correlation matrix, and the rock stratum contact relationship network, a geological structure correlation map containing node attributes and edge attributes is generated. The node attributes are the geological structure description information of the observation point and the corresponding fossil distribution record, while the edge attributes are the similarity degree in the stratigraphic lithology similarity matrix, the correlation strength in the geological structure correlation matrix, and the contact relationship type in the rock stratum contact relationship network.

3. The method for three-dimensional geological modeling of fossil sites based on machine learning according to claim 2, characterized in that, The process of performing similarity comparison on rock types in the stratigraphic lithological feature set, calculating the matching degree of rock types at different observation points, and generating a stratigraphic lithological similarity matrix includes: The rock type of each observation point in the stratigraphic lithology feature set is converted into a standardized rock type code. The standardized rock type code is generated based on a preset rock classification system, and each rock type corresponds to a unique code value. Construct a rock type matching dictionary, which contains matching weights between different rock types. The matching weights are determined based on the genetic correlation of rock types, and the more closely related the genesis of a rock type, the higher its matching weight. For any two observation points, extract the standardized rock type codes of the two observation points, query the rock type matching dictionary to obtain the corresponding matching weights, and use the matching weights as the matching degree of the rock types of the two observation points; According to the numbering order of the observation points, the matching degree of rock types between all two observation points is filled into the corresponding positions of the matrix to generate the stratigraphic lithology similarity matrix. The stratigraphic lithology similarity matrix is ​​a square matrix, and the number of rows and columns of the stratigraphic lithology similarity matrix is ​​equal to the number of observation points. The diagonal elements of the stratigraphic lithology similarity matrix represent the matching degree of rock types of the same observation point, and their values ​​are preset maximum values.

4. The method for three-dimensional geological modeling of fossil sites based on machine learning according to claim 1, characterized in that, The spatial correlation modeling layer of the geological structure evolution model performs spatial correlation modeling on the node attribute feature vectors and edge attribute feature vectors to generate a spatial correlation feature matrix, including: The node attribute feature vector and the edge attribute feature vector are input into the feature fusion unit of the spatial association modeling layer. The node attribute feature vector and the edge attribute feature vector are subjected to element-wise weighted fusion processing to generate a fused feature vector. The weights of the element-wise weighted fusion processing are preset according to the degree of influence of node attributes and edge attributes on spatial association. The fused feature vector is processed by spatial coordinate mapping, which maps the feature value of each observation point to the corresponding three-dimensional spatial coordinates to generate a set of spatial feature points with spatial coordinate information. The three-dimensional spatial coordinates are determined based on the geographic coordinate system of the fossil site. The distance calculation unit of the spatial association modeling layer calculates the Euclidean distance and direction vector between any two spatial feature points in the set of spatial feature points, generating a distance matrix and a direction matrix. The elements in the distance matrix represent the straight-line distance between two spatial feature points, and the elements in the direction matrix represent the directional angle between two spatial feature points. By combining the distance matrix, the direction matrix, and the fused feature vector, a spatial correlation function is constructed. This spatial correlation function is used to describe the relationship between the feature values ​​of spatial feature points as a function of distance and direction. Based on the spatial correlation function, the spatial feature point set is interpolated to supplement the feature values ​​of unobserved points, generating a spatial feature grid covering the entire fossil site. The spatial feature grid is then converted into a matrix form to obtain a spatial correlation feature matrix, where the rows and columns of the spatial correlation feature matrix correspond to the planar coordinates of the fossil site, and the matrix elements correspond to the feature values.

5. The method for three-dimensional geological modeling of fossil sites based on machine learning according to claim 1, characterized in that, The process involves determining the stratigraphy, geological structure, and rock contact relationships of each three-dimensional grid cell based on its corresponding feature values, thereby generating an initial three-dimensional geological framework model that includes stratigraphic interface distribution characteristics and spatial distribution characteristics of geological structures. Extract the stratigraphic lithology feature components from the feature values ​​corresponding to each three-dimensional grid cell, compare them with the preset lithology feature benchmark library, determine the rock type of the three-dimensional grid cell, mark its stratigraphic lithology attributes, classify adjacent cells with the same attributes into the same lithology cell, and record the coordinate information of each lithology cell. Geological structural features are extracted from the feature values ​​to identify fold, fault, and joint features. Based on the fold features, the fold location of the unit is determined, and units that are spatially continuous and belong to the same fold location are grouped into the same fold structural unit. Based on the fault features, it is determined whether the unit belongs to a fault zone, and units distributed along the fault extension trajectory are grouped into the same fault structural unit. Based on the joint features, the degree of joint development of the unit is calculated, and units with the same degree of development and that are adjacent are grouped into the same joint structural unit. The coordinate information and corresponding feature parameters of each structural unit are recorded. Extract the rock layer contact relationship feature components from the feature values, identify the contact type identifier, determine the contact relationship type between adjacent three-dimensional mesh units, extract the contact surfaces of adjacent units with the same contact type and located at the boundary of lithological units as rock layer contact interfaces, and record the coordinates and contact relationship type of the contact interfaces. Spatial overlay analysis was performed on all lithological units. The vertical positional relationship was determined based on the three-dimensional coordinate depth. Stratigraphic units were divided in combination with the rock layer contact interface. Each stratigraphic unit is composed of lithological units with the same vertical positional relationship and connected by the integrated contact interface. The composition and distribution range of the stratigraphic units were recorded. The top and bottom surfaces of each stratigraphic unit are extracted as stratigraphic interfaces. The top surface is the contact interface with the smallest depth in the unit, and the bottom surface is the contact interface with the largest depth. The coordinates of the stratigraphic interfaces are recorded to form the stratigraphic interface distribution characteristics. By integrating the coordinate information and characteristic parameters of fold, fault and joint structural units, their three-dimensional spatial distribution morphology is determined, forming the spatial distribution characteristics of geological structures; By integrating the distribution characteristics of stratigraphic interfaces and the spatial distribution characteristics of geological structures, ensuring spatial consistency, and supplementing discontinuous areas, the stratigraphic interfaces and geological structures are presented in a continuous distribution state, thus forming an initial three-dimensional geological framework model.

6. The method for three-dimensional geological modeling of fossil sites based on machine learning according to claim 1, characterized in that, The optimization and adjustment process based on the initial three-dimensional geological framework model and preset geological constraint rules yields an optimized three-dimensional geological model of the fossil site, including: Obtain preset geological constraint rules, which include constraints on the range of stratigraphic thickness variation, constraints on geological structural morphology parameters, and constraints on the correlation between fossil distribution and geological structure. The constraints on the range of stratigraphic thickness variation limit the maximum and minimum thickness of different strata. The constraints on geological structural morphology parameters limit the range of curvature of folds, the range of extension length of faults, and the range of density of joints. The constraints on the correlation between fossil distribution and geological structure limit the correspondence between specific fossil species and the corresponding lithology and geological structure of strata. The initial three-dimensional geological framework model and the geological constraint rules are input into the model optimization module to extract the stratigraphic interface distribution features and geological structural spatial distribution features in the initial three-dimensional geological framework model, thereby obtaining the set of features to be optimized. The formation thickness in the set of features to be optimized is calculated, and the calculated formation thickness is compared with the maximum and minimum values ​​in the constraint of the formation thickness variation range. Formation areas that exceed the constraint range are marked as thickness anomaly areas. Geological structural morphological parameters are extracted from the set of features to be optimized. The curvature of the extracted folds, the extension length of the fractures, and the density of the joints are compared with the corresponding ranges in the constraints of the geological structural morphological parameters. Geological structural areas that exceed the constraints are marked as structural anomaly areas. The correspondence between fossil distribution records and geological structure description information in the initial three-dimensional geological framework model is analyzed, and compared with the correspondence in the fossil distribution and geological structure association constraints. Regions that do not meet the association constraints are marked as association anomaly regions. Based on the thickness anomaly region, structural anomaly region and associated anomaly region, the initial three-dimensional geological framework model is locally adjusted to adjust the stratigraphic interface distribution characteristics and geological structural spatial distribution characteristics of the anomaly region so that the adjusted characteristics conform to the geological constraint rules. Repeat the steps of extracting the feature set to be optimized, marking abnormal areas, and performing local adjustments until there are no abnormal areas in the initial three-dimensional geological framework model, to obtain the optimized three-dimensional geological model of the fossil site. The optimized three-dimensional geological model of the fossil site includes the spatial distribution boundaries of each rock layer, the three-dimensional coordinate range of the fossil enrichment area, and the spatial correlation between geological structure and fossil distribution.

7. The method for three-dimensional geological modeling of fossil sites based on machine learning according to claim 6, characterized in that, The initial three-dimensional geological framework model is locally adjusted based on the thickness anomaly region, structural anomaly region, and associated anomaly region. This adjustment modifies the stratigraphic interface distribution characteristics and spatial distribution characteristics of geological structures within the anomaly regions, ensuring that the adjusted features conform to the geological constraint rules. This includes: For the thickness anomaly region, the difference between the current formation thickness of the thickness anomaly region and the maximum and minimum values ​​in the formation thickness variation range constraint is calculated. The adjustment direction and adjustment range of the formation interface are determined based on the difference. The adjustment direction and adjustment range are set based on the sign and magnitude of the difference. The stratigraphic interface distribution characteristics of the thickness anomaly region are moved and adjusted according to the adjustment direction and adjustment range, so that the adjusted stratigraphic thickness is within the constraint of the stratigraphic thickness variation range, and the three-dimensional coordinates of the adjusted stratigraphic interface are recorded as the updated stratigraphic interface distribution characteristics. For the structurally anomalous region, based on the range in the constraints of the geological structural morphology parameters, the deviation value between the current geological structural morphology parameters of the structurally anomalous region and the constraint range is calculated. Based on the deviation value, the deformation adjustment amount of the geological structure is determined. The deformation adjustment amount includes the curvature adjustment amount of folds, the extension length adjustment amount of fractures, and the density adjustment amount of joints. The spatial distribution characteristics of the geological structure in the structural anomaly area are deformed and adjusted according to the deformation adjustment amount, so that the adjusted geological structure morphological parameters are within the constraints of the geological structure morphological parameters, and the three-dimensional morphology of the adjusted geological structure is recorded as the updated spatial distribution characteristics of the geological structure. For the aforementioned anomalous region, analyze the reasons why the fossil distribution record and geological structure description information in the anomalous region do not meet the association constraints. If the correspondence between the fossil distribution record and the geological structure description information is incorrect, then correct the correspondence of the anomalous region to make it consistent with the correspondence in the fossil distribution and geological structure association constraints. After adjusting the thickness anomaly region, structural anomaly region and associated anomaly region, the distribution characteristics of stratigraphic interfaces and the spatial distribution characteristics of geological structures around the adjusted region are smoothed to make the geological structure characteristics of the adjusted region and the surrounding region continuous and consistent. The initial three-dimensional geological framework model after smoothing transition is checked as a whole to verify whether the adjusted stratigraphic thickness, geological structural morphology parameters, fossil distribution and geological structure correlation all conform to the geological constraint rules. If there are any non-compliance, the abnormal areas are re-identified and adjusted until all features conform to the geological constraint rules.

8. The method for three-dimensional geological modeling of fossil sites based on machine learning according to claim 1, characterized in that, The output of the optimized three-dimensional geological model of the fossil site includes: The spatial distribution boundaries of each rock layer in the optimized three-dimensional geological model of the fossil site are extracted, and the spatial distribution boundary of each rock layer is converted into a three-dimensional polygonal mesh surface. Each three-dimensional polygonal mesh surface is composed of multiple triangular facets, and the vertex coordinates of each triangular facet are determined based on the geographic coordinate system of the fossil site. Extract the three-dimensional coordinate range of the fossil enrichment area from the optimized three-dimensional geological model of the fossil site, determine the minimum bounding cuboid of each fossil enrichment area, and record the three-dimensional coordinates of the eight vertices of the minimum bounding cuboid as the boundary coordinates of the fossil enrichment area. Extract the spatial correlation between geological structures and fossil distribution in the optimized three-dimensional geological model of the fossil site, and generate a correlation list, which contains the distribution probability of each geological structure type and the corresponding fossil species; The three-dimensional polygonal mesh, the boundary coordinates of the fossil-rich area, and the list of relationships are integrated to generate a model output file containing spatial data and attribute data. The spatial data is the boundary coordinates of the three-dimensional polygonal mesh and the fossil-rich area, and the attribute data is the list of relationships and the stratigraphic lithology and geological structure characteristics of each rock layer. The model output file is converted into a preset 3D model format, which supports display and interactive operation in 3D visualization software. The converted model output file is stored in the specified storage location, and a model output log is generated. The model output log records the time of output of the three-dimensional geological model of the fossil site, the storage location, and the information on the number of rock layers and fossil types contained in the three-dimensional geological model of the fossil site. The converted model output file is loaded into the 3D visualization platform through the 3D visualization interface. The optimized 3D geological model of the fossil site is displayed in the 3D visualization platform. Users can rotate, scale, and cross-section the optimized 3D geological model of the fossil site to view the geological structure and fossil distribution from different perspectives.

9. A three-dimensional geological modeling system for fossil sites based on machine learning, characterized in that, The system includes a processor and a memory, the memory and the processor being connected. The memory is used to store programs, instructions, or code, and the processor is used to execute the programs, instructions, or code in the memory to implement the machine learning-based three-dimensional geological modeling method for fossil sites as described in any one of claims 1-8.

Images