Geological body type identification method and system based on multi-scale geological correlation network

Abstract

The application discloses a geological body type identification method and system based on a multi-scale geological correlation network, which comprises the following steps: combining a geological knowledge graph with spatial coordinates to generate comprehensive geological feature descriptions of each point; based on this, a multi-scale geological correlation network is constructed by fusing spatial distance and feature similarity; the geological response mode value of each point is calculated by using the network, and a spatially continuous and feature-similar exploration point set is divided according to the geological response mode value; in each set, a few target exploration points are selected based on spatial centrality, feature representativeness and network importance, and a preliminary type identification result is obtained by using a pre-trained model; a weighted fusion mechanism is used to generate the final target geological body type identification result of the set based on the multi-scale geological correlation network; and finally, the result is intelligently propagated to all points in the set based on the similarity of the geological response mode. The method realizes efficient and automatic regional geological body type identification based on the multi-scale geological correlation network.

Description

Technical Field

[0001] This invention belongs to the field of interdisciplinary technology of geological exploration and artificial intelligence, and in particular relates to a method and system for identifying geological body types based on multi-scale geological association networks. Background Technology

[0002] In fields such as mineral resource exploration and engineering geological survey, accurately identifying the types of geological bodies in underground rock formations is crucial for resource assessment, engineering design, and disaster prevention. Traditional geological body type identification based on multi-scale geological correlation networks mainly relies on expert interpretation of data such as borehole cores and well logging curves, which has limitations such as low efficiency, high subjectivity, and difficulty in quantification.

[0003] With the development of exploration technology, the geological data obtained is becoming increasingly diversified, including geophysical logging data, geotechnical parameters, and geochemical indicators. The current technical challenge is how to efficiently and objectively integrate this multi-dimensional data and fully utilize the spatial correlation information between exploration points to achieve accurate and automated identification of geological body types based on multi-scale geological correlation networks.

[0004] Existing automated identification methods typically analyze only the sequence data of a single borehole independently or employ simple spatial interpolation methods, failing to fully explore the deep correlations in characteristics and space between different exploration sites, and thus struggling to accurately characterize the spatial distribution patterns of lithology under complex geological conditions. Therefore, there is an urgent need for a new method for identifying geological body types based on multi-scale geological correlation networks, capable of integrating multi-dimensional features, constructing spatial correlation networks, and performing intelligent reasoning. Summary of the Invention

[0005] The present invention aims to overcome the shortcomings of the prior art and provide a geological body type identification method and system based on a multi-scale geological association network that can comprehensively utilize multi-dimensional exploration data, explicitly model spatial correlation relationships, and achieve efficient and automated identification.

[0006] In a first aspect, the present invention provides a method for identifying geological body types based on a multi-scale geological association network, comprising:

[0007] Obtain geological exploration data for at least one exploration point within a preset exploration depth range, and sort the geological exploration data for the same exploration point to obtain at least one geological exploration data sequence.

[0008] Based on the at least one geological exploration data sequence, a multi-dimensional geological feature fusion analysis is performed to generate a comprehensive geological feature description for each exploration point.

[0009] Based on the comprehensive geological feature descriptions of each exploration point and the three-dimensional spatial coordinates of each exploration point, a multi-scale geological correlation network is constructed among the exploration points.

[0010] Based on the multi-scale geological correlation network, the geological response model value of each exploration point is determined, and the exploration points are divided according to the geological response model value to obtain at least one set of exploration points.

[0011] Obtain the three-dimensional spatial coordinates of each exploration point in a certain set of exploration points, and select at least one target exploration point in the certain set of exploration points according to the three-dimensional spatial coordinates and a preset sequence selection rule.

[0012] The geological exploration data sequence corresponding to the at least one target exploration point is input into a preset geological body type identification model based on a multi-scale geological association network to obtain preliminary geological body type identification results based on a multi-scale geological association network corresponding to the at least one target exploration point.

[0013] Based on the preliminary geological body type identification results based on multi-scale geological association networks, a multi-source decision fusion mechanism is adopted to generate the final target geological body type identification results based on multi-scale geological association networks.

[0014] Based on the similarity of geological response patterns between other exploration sites and the at least one target exploration site, the geological body type identification results of the final target based on the multi-scale geological association network are spatially propagated to obtain other geological body type identification results based on the multi-scale geological association network corresponding to other exploration sites.

[0015] Secondly, the present invention provides a geological body type identification system based on a multi-scale geological association network, comprising:

[0016] The acquisition module is configured to acquire geological exploration data at at least one exploration point within a preset exploration depth range, and sort the geological exploration data at the same exploration point to obtain at least one geological exploration data sequence.

[0017] The generation module is configured to perform multi-dimensional geological feature fusion analysis based on the at least one geological exploration data sequence to generate a comprehensive geological feature description for each exploration point.

[0018] The module is configured to construct a multi-scale geological association network between exploration points based on the comprehensive geological feature descriptions and three-dimensional spatial coordinates of each exploration point.

[0019] The partitioning module is configured to determine the geological response pattern of each exploration point based on the multi-scale geological correlation network, and partition the exploration points according to the geological response pattern to obtain at least one set of exploration points.

[0020] The selection module is configured to obtain the three-dimensional spatial coordinates of each exploration point in a certain set of exploration points, and select at least one target exploration point in the certain set of exploration points according to the three-dimensional spatial coordinates and a preset sequence selection rule.

[0021] The identification module is configured to input the geological exploration data sequence corresponding to the at least one target exploration point into a preset geological body type identification model based on a multi-scale geological association network, so as to obtain the preliminary geological body type identification result based on the multi-scale geological association network corresponding to the at least one target exploration point.

[0022] The fusion module is configured to generate the final target geological body type identification result based on the multi-scale geological association network based on the preliminary geological body type identification results of each preliminary multi-scale geological association network.

[0023] The output module is configured to spatially propagate the geological body type identification results of the final target based on the geological response pattern similarity between other exploration points and the at least one target exploration point, thereby obtaining other geological body type identification results based on the multi-scale geological association network corresponding to other exploration points.

[0024] Thirdly, an electronic device is provided, comprising: at least one processor, and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the steps of the geological body type identification method based on a multi-scale geological association network according to any embodiment of the present invention.

[0025] Fourthly, the present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein when the program instructions are executed by a processor, the processor performs the steps of the geological body type identification method based on a multi-scale geological association network according to any embodiment of the present invention.

[0026] This application presents a geological body type identification method and system based on a multi-scale geological association network. At the data representation level, the method utilizes adaptive depth segmentation and multi-scale (macro, meso, and micro) feature fusion, combined with a geological knowledge graph for confidence weighting, to generate a highly condensed and geologically significant comprehensive feature description, providing a powerful input for subsequent analysis. At the spatial modeling level, it innovatively constructs a multi-scale association network that integrates spatial distance and feature similarity, capable of simultaneously depicting local close associations and regional trend connections, revealing more precisely the spatial-attribute coupling relationships between exploration points that conform to geological laws. Regarding computational efficiency and reliability, the method quantifies the geological response pattern values ​​of points based on network topology, and accordingly divides spatially continuous and feature-uniform point sets. Then, through a multi-criteria fusion strategy (spatial centrality, feature representativeness, and network importance), it selects a few of the most representative target points from each set for high-precision model identification, avoiding the huge overhead of independently modeling all points, thus optimizing computational resources and reducing costs. Finally, by using multi-source weighted decision fusion and a pattern similarity-based intelligent spatial propagation mechanism, the highly reliable target point identification results are reasonably and robustly promoted to the entire region. While ensuring spatial continuity and geological consistency, the identification results of all points are obtained efficiently. Attached Figure Description

[0027] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0028] Figure 1 A flowchart illustrating a geological body type identification method based on a multi-scale geological association network, provided as an embodiment of the present invention;

[0029] Figure 2 This is a structural block diagram of a geological body type identification system based on a multi-scale geological association network, provided in an embodiment of the present invention.

[0030] Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0031] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0032] Please see Figure 1 The diagram shows a flowchart of a geological body type identification method based on a multi-scale geological association network according to this application.

[0033] like Figure 1 As shown, the geological body type identification method based on multi-scale geological association networks specifically includes the following steps:

[0034] Step S101: Obtain geological exploration data of at least one exploration point within a preset exploration depth range, and sort the geological exploration data of the same exploration point to obtain at least one geological exploration data sequence.

[0035] Step S102: Based on the at least one geological exploration data sequence, perform multi-dimensional geological feature fusion analysis to generate a comprehensive geological feature description for each exploration point.

[0036] In this step, based on the changes in the values ​​of various geological feature parameters with exploration depth in the current geological exploration data sequence at the current exploration point, multiple abrupt changes in geological feature depth points are identified. These abrupt changes include:

[0037] Calculate the first-order difference between adjacent sampling depths for each geological feature parameter in the current geological exploration data sequence;

[0038] The sampling point depth whose absolute value of the first-order difference exceeds the preset mutation threshold is marked as the potential mutation depth;

[0039] The potential abrupt change depths of all geological feature parameters are subjected to union processing and depth-near clustering to obtain multiple geological feature abrupt change depth points.

[0040] Using the aforementioned multiple geological feature abrupt change depth points as boundaries, the entire exploration depth range is divided into multiple depth segments;

[0041] Multi-scale fusion feature vectors are extracted from each depth segment, and these feature vectors are arranged in spatial order according to their depth segments to form a three-dimensional feature tensor. The extraction of the multi-scale fusion feature vector for each depth segment includes:

[0042] Within the entire range of a certain depth segment, calculate the mean and variance of the geological feature parameters of all data points within that depth segment to form a feature sub-vector at a certain macroscopic scale.

[0043] A certain depth segment is subdivided into several equal-length sub-windows. The mean value of the feature parameters in each sub-window is calculated. The sub-windows are arranged according to their depth to obtain a sub-window mean sequence. The slope and fluctuation entropy of the sub-window mean sequence are calculated to form a feature sub-vector at a certain mesoscale.

[0044] The number and distribution density of local extreme points of a certain geological exploration data subsequence within a certain depth segment are extracted to form a feature subvector at a certain microscale. Here, a certain geological exploration data subsequence is a geological exploration data subsequence consisting of all data points in the current geological exploration data sequence whose exploration depth falls within a certain depth segment.

[0045] By concatenating a macroscopic feature vector, a mesoscopic feature vector, and a microscopic feature vector at a certain depth segment, a multi-scale fused feature vector at a certain depth segment is obtained.

[0046] The three-dimensional feature tensor is matched with common stratigraphic and lithological combination patterns in a pre-established geological knowledge graph. The degree of matching is calculated, and a pattern matching confidence score is assigned to each depth segment.

[0047] Using the pattern matching confidence as the weight, the multi-scale fusion feature vectors of each depth segment are weighted and averaged to generate a preliminary comprehensive feature vector.

[0048] The three-dimensional spatial coordinates of the current exploration point are converted into a spatial location encoding vector, and the spatial location encoding vector is fused with the comprehensive feature vector at the element level to generate a comprehensive geological feature description of the current exploration point.

[0049] In one specific embodiment, assuming there is a borehole ZK01 in a certain exploration area with a depth range of 0-200 meters, sampling is performed at 1-meter intervals to obtain the measured values ​​of four geological characteristic parameters: density (DEN), longitudinal wave velocity (VP), natural radioactivity intensity (GR), and apparent resistivity (RES). This constitutes a "geological exploration data sequence" containing 200 depth points, with each point having 4 characteristic parameter values.

[0050] Step 1: Identify depth points where geological features change abruptly, and adaptively divide depth segments.

[0051] First, for each feature parameter sequence (DEN, VP, GR, RES), the change value (first-order difference) between adjacent sampling depths is calculated, and a "preset mutation threshold" is set. For example, for the density parameter, this threshold can be set to twice the standard deviation of its total sequence difference value. All difference values ​​are iterated, and sampling point depths whose absolute values ​​exceed the corresponding parameter sequence threshold are marked as "potential mutation depths." For example, at depths of 35 meters, 82 meters, 110 meters, and 156 meters, the difference values ​​of different parameters may significantly exceed the threshold; these points are marked. Next, "depth proximity clustering" is performed on all marked depth points, setting a clustering tolerance (e.g., 5 meters). After clustering, several points originally scattered in the 33-37 meter range are merged into a representative depth (e.g., 35 meters), and points at 82 meters, 110 meters, and 156 meters are also merged separately. Finally, S=4 geological feature mutation depth points are obtained: 35 meters, 82 meters, 110 meters, and 156 meters. Using these four depth points as boundaries, the entire 200-meter borehole is divided into five depth segments: Depth Segment 1 (0-35 meters), Depth Segment 2 (35-82 meters), Depth Segment 3 (82-110 meters), Depth Segment 4 (110-156 meters), and Depth Segment 5 (156-200 meters).

[0052] Step 2: Extract the multi-scale fusion feature vector for each depth segment

[0053] Taking depth section 2 (35-82 meters) as an example:

[0054] Macroscale: Calculate the arithmetic mean and standard deviation of the four parameters DEN, VP, GR, and RES within this segment (approximately 47 sampling points), thus obtaining a macroscale feature vector containing 8 elements (4 means + 4 standard deviations);

[0055] Mesoscale: Depth segment 2 (approximately 47 meters long) is subdivided into several equal-length sub-windows, for example, each sub-window is 5 meters long, resulting in approximately 9 sub-windows (the last one may be less than 5 meters long). Within each sub-window, the average values ​​of four parameters, DEN, VP, GR, and RES, are calculated. Thus, for each parameter, a sequence of 9 average values ​​arranged in depth order is obtained, called the sub-window mean sequence. The linear regression slope (reflecting the overall trend of the parameter's change within the segment) and the information entropy of the sequence value (reflecting the complexity of the parameter's value fluctuation at the sub-window scale) of the sub-window mean sequence of each parameter (DEN, VP, GR, RES) are calculated respectively. This yields a mesoscale feature sub-vector containing 8 elements (4 slopes + 4 entropy values).

[0056] Microscale: For the original DEN, VP, GR, RES data subsequences corresponding to depth segment 2, identify the total number of local maxima and local minima of each parameter curve in this segment, and calculate their distribution density (i.e., the total number of extreme values ​​divided by the segment length of 47 meters). This results in a microscale feature subvector containing 8 elements (the total number and density of extreme points for each of the 4 parameters).

[0057] Finally, the three sub-vectors of macro (8-dimensional), meso (8-dimensional), and micro (8-dimensional) are concatenated in sequence to obtain a 24-dimensional vector, which is the multi-scale fusion feature vector of depth segment 2. This process is repeated for the other four depth segments to obtain five 24-dimensional feature vectors.

[0058] Step 3: Construct a 3D feature tensor and perform knowledge-driven confidence weighting.

[0059] The multi-scale fused feature vectors of these five depth segments are stacked according to their corresponding depth order (depth segment 1, depth segment 2, depth segment 3, depth segment 4, depth segment 5) to form a two-dimensional feature matrix of size 5 (number of depth segments) × 24 (feature dimension) (conceptually, it can be regarded as a tensor with a third dimension of 1). A geological knowledge map is pre-established, which contains typical stratigraphic and lithological combinations in this area (such as "overburden-strongly weathered rock-moderately weathered sandstone-mud interlayers") and their corresponding feature patterns. The above feature matrix of borehole ZK01 is then compared with the knowledge map. Pattern matching is performed on the patterns in the knowledge base (e.g., calculating the cosine similarity between feature vectors). Assuming that the features of depth segment 2 highly match the typical pattern of "moderately weathered sandstone" in the knowledge base, while the feature of segment 4 matches the pattern of "mud interlayer" to a moderate degree, a pattern matching confidence score is assigned to each depth segment (e.g., depth segment 2: 0.95, depth segment 4: 0.7). Then, using these confidence scores as weights, a weighted average is performed on the 24-dimensional feature vectors of the five depth segments to obtain a preliminary 24-dimensional comprehensive feature vector that can represent the comprehensive features of the entire borehole vertical sequence.

[0060] Step 4: Integrate spatial location information to generate the final comprehensive geological feature description.

[0061] The three-dimensional spatial coordinates (X,Y,Z) of borehole ZK01 are obtained. Through a linear transformation and normalization process (e.g., mapping the coordinate values ​​to the [0,1] interval), they are converted into a fixed-length spatial location encoding vector (e.g., a 3D vector). Finally, the 24-dimensional preliminary comprehensive feature vector obtained in the previous step is fused with this 3D spatial location encoding vector (e.g., directly concatenated into a 27-dimensional vector). The resulting 27-dimensional vector is the comprehensive geological feature description of borehole ZK01. It contains the multi-scale and multi-parameter geological characteristics of the borehole in the vertical direction and carries its unique geospatial location information.

[0062] In summary, by performing adaptive depth segmentation based on statistical abrupt change detection on the original data sequence, the extracted feature units are associated with possible geological interfaces, transcending the mechanical nature of fixed-interval segmentation and enhancing the geological indicative significance of the features. Secondly, the feature extraction mechanism at three scales—macro (overall statistics), meso (trend fluctuations), and micro (local extrema)—constructs a comprehensive and complementary information representation, capable of simultaneously capturing background values, variation patterns, and subtle anomalies of stratigraphic properties, overcoming the limitations of low-dimensional information from single statistical quantities. Furthermore, the introduction of geological knowledge graphs for pattern matching and confidence weighting integrates domain prior knowledge into the features in a computable manner. The feature generation process organically combines data-driven and knowledge-guided approaches, significantly enhancing the geological rationality and discriminative power of the generated features. Finally, the absolute spatial coordinate information is encoded and fused with the sequence features, creatively embedding the spatial attributes of points into their feature descriptions, laying a crucial foundation for subsequent analysis of spatial correlations. In summary, the "comprehensive geological feature description" output in this step is a strong representation vector that deeply integrates vertical multi-scale information, geological prior constraints, and absolute spatial location. It provides a highly condensed, geologically meaningful, and highly discriminative data foundation for the entire identification process, serving as the core engine for achieving high-precision, intelligent spatial identification and reasoning.

[0063] Step S103: Based on the comprehensive geological feature description of each exploration point and the three-dimensional spatial coordinates of each exploration point, construct a multi-scale geological correlation network among the exploration points.

[0064] In this step, each exploration point is defined as a network node, and the comprehensive geological feature description vector of each exploration point is defined as the node feature of the network node.

[0065] Calculate the three-dimensional Euclidean distance between any two exploration points based on their three-dimensional spatial coordinates.

[0066] Based on the comprehensive geological feature description vector of the exploration points, calculate the feature vector similarity between any two exploration points;

[0067] Multiple spatial scale levels are preset, and a corresponding spatial neighborhood radius and similarity threshold are set for each spatial scale level. The value of the spatial neighborhood radius increases with the increase of the spatial scale level, and the value of the similarity threshold decreases with the increase of the spatial scale level.

[0068] Establish a connection edge at the current spatial scale level between any two exploration points whose three-dimensional Euclidean distance is no greater than the spatial neighborhood radius corresponding to the current spatial scale level and whose feature vector similarity is greater than the similarity threshold corresponding to the current spatial scale level.

[0069] Based on the three-dimensional Euclidean distance and feature vector similarity, the connection weight of the connecting edge is determined according to a preset weight calculation rule, wherein the connection weight is negatively correlated with the three-dimensional Euclidean distance and positively correlated with the feature vector similarity.

[0070] By integrating the connecting edges established at all spatial scale levels, a multi-scale geological correlation network containing multi-scale connections is formed.

[0071] In one specific embodiment, it is assumed that there are 5 borehole locations in a certain exploration area, numbered ZK01 to ZK05. Each location has generated a 27-dimensional comprehensive geological feature description vector Fi (i=1,...,5) and has known three-dimensional spatial coordinates.

[0072] Step 1: Define network nodes and attributes

[0073] Each exploration point (ZK01-ZK05) is defined as a node in the network. The attributes of each node are represented by its corresponding 27-dimensional comprehensive geological feature description vector.

[0074] Step 2: Calculate the distance and similarity matrix

[0075] Calculate the 3D Euclidean distance between any two nodes. For example, given that the coordinates of ZK01 and ZK02 are (x1, y1, z1) and (x2, y2, z2) respectively, calculate the distance between them. By traversing all pairs of points, a 5×5 symmetric distance matrix D is obtained.

[0076] Calculate the cosine similarity between the feature vectors of any two nodes. For example, the feature vector similarity s12 between the comprehensive geological feature description vector F1 of ZK01 and the comprehensive geological feature description vector F2 of ZK02 is s12 = (F1·F2) / (||F1||·||F2||). Similarly, a 5×5 symmetric similarity matrix S is obtained.

[0077] Step 3: Set multi-scale hierarchical parameters

[0078] Two spatial scale levels are preset (L=2):

[0079] Fine scale: neighborhood radius R1 = 50 meters, similarity threshold θ1 = 0.85.

[0080] Coarse scale: neighborhood radius R2 = 150 meters, similarity threshold θ2 = 0.60.

[0081] In practical applications, more levels and parameter values ​​can be determined based on factors such as exploration network density and geological complexity. The principle for parameter setting is: smaller neighborhood radius and larger similarity threshold at smaller scales, aiming to discover strong local correlations; larger neighborhood radius and smaller similarity threshold at larger scales, aiming to discover weak regional correlations.

[0082] Step 4: Establish connecting edges and calculate weights layer by layer.

[0083] Iterate through each pair of nodes (i,j) for each level. Determine if the connection condition is met:

[0084] Conditional judgment: If the 3D Euclidean distance dij between the i-th node and the j-th node is less than or equal to Rl, and the similarity sij between the comprehensive geological feature description vectors of the i-th node and the j-th node is greater than θl, then establish a hierarchical line between node i and node j. The connecting edge;

[0085] Weight calculation: hierarchy Weight of connecting edges Calculated according to preset rules, the formula is as follows: , hierarchical The neighborhood radius;

[0086] This formula reflects a positive correlation between weights and feature vector similarity, and a negative correlation with the normalized 3D Euclidean distance. The distance factor is 1 when the distance between two points is 0; when the distance equals... When the distance factor is 0, the weight depends entirely on the similarity, but this often... The connection may be at the boundary.

[0087] Example calculation (assuming data):

[0088] Assume that the three-dimensional Euclidean distance between nodes ZK01 and ZK03 is d13 = 40 meters, and the similarity of the feature vectors is s13 = 0.90.

[0089] For fine scale ( =1): d13=40≤R1=50 holds true, s13=0.90>θ1 holds true, therefore a connection is established, and the weights are... ;

[0090] For a rough scale ( =2): d13=40≤R2 holds true, s13=0.90>θ2 holds true, therefore a connection is also established, weights ;

[0091] It is evident that the same pair of nodes has different connection weights at different scales, reflecting the correlation strength at different spatial observation scales.

[0092] Step 5: Integrate to form a multi-scale network

[0093] By aggregating the connections from all levels, a multi-layered network structure is formed. The final network G can be represented as:

[0094] G={V,E1,E2,...,EL}

[0095] Where V is the set of nodes, and El is the set of edges at the l-th level, with each edge recording the connected node pairs and their weights. In this network, there may be no connection between any two nodes, or there may be one or more edges representing different scales of association (with different weights).

[0096] In summary, by introducing a multi-scale hierarchical structure (such as fine-scale and coarse-scale), the correlation between points in terms of local details and regional trends can be simultaneously characterized. The fine-scale network captures the close connections between nearby points with high feature similarity, reflecting the homogeneity and continuity within lithological units. The coarse-scale network reveals the potential connections between distant points with certain feature similarities, suggesting the distribution of a larger geological tectonic background or similar sedimentary environments. This multi-level expression is more in line with the objective law that geological bodies have spatial nesting and hierarchical characteristics. Secondly, the establishment of connection edges is subject to dual constraints (spatial distance and feature similarity), abandoning the coarse assumption that relies solely on geometric distance, giving network connections clear geological significance—only points that are both spatially adjacent and similar in lithological and physical characteristics are strongly correlated, which greatly enhances the geological rationality of the network topology. Furthermore, the calculation of connection weights integrates distance decay and feature similarity, realizing a refined and quantitative characterization of correlation strength, providing accurate input for subsequent quantitative analysis based on network topology.

[0097] Step S104: Based on the multi-scale geological correlation network, determine the geological response model value of each exploration point, and divide the exploration points according to the geological response model value to obtain at least one set of exploration points.

[0098] In this step, all neighboring nodes of a given exploration point at different spatial scale levels are obtained;

[0099] For each spatial scale level, the connection weights of a certain exploration point and all its neighboring nodes at the current level are summed to obtain the intra-level connection strength of the exploration point.

[0100] The multi-scale feature centrality of a certain exploration point is obtained by weighting and summing the intra-level connectivity strength of the exploration point across all spatial scales according to the preset importance coefficients of each level.

[0101] The multi-scale feature centrality of a certain exploration point is normalized to obtain a certain geological response model value for that exploration point.

[0102] Exploration points whose geological response model values ​​are within the same range are grouped into the same set of exploration points.

[0103] For each set of exploration points obtained by division, a spatial connectivity test is performed to ensure that the exploration points in the set are spatially continuous, thus obtaining at least one final set of exploration points.

[0104] In one specific embodiment, it is assumed that in step S103, a multi-scale geological association network containing two scale levels (L=2) has been constructed for six borehole locations (ZK01 to ZK06) in a certain exploration area, and each node has specific neighbors and connection weights at different levels.

[0105] Step 1: Obtain neighboring nodes and calculate hierarchical inner link strength

[0106] Taking node ZK03 as an example:

[0107] At a fine scale ( =1) In the network, assume that ZK03 has three neighbors: ZK01, ZK02, and ZK04, with corresponding connection weights of w31-1=0.18, w32-1=0.22, and w34-1=0.15, respectively.

[0108] Its hierarchical internal connectivity strength I3-1=0.18+0.22+0.15=0.55.

[0109] On a rough scale ( =2) In the network, assume that ZK03 has four neighbors: ZK01, ZK02, ZK04, and ZK05, with corresponding connection weights of w31-2=0.66, w32-2=0.70, w34-2=0.58, and w35-2=0.30, respectively.

[0110] Its hierarchical internal connectivity strength I3-2=0.66+0.70+0.58+0.30=2.24.

[0111] Repeat this calculation for all 6 nodes to obtain the connection strength of each node at both levels.

[0112] Step 2: Calculate the multi-scale feature centrality using weighted methods.

[0113] Preset importance coefficients for the two levels, for example: fine scale ( =1) Coefficient α=0.7, coarse scale ( =2) The coefficient β=0.3 reflects that strong local correlations are more likely to demonstrate the "core" status of a point than weak regional correlations.

[0114] The multi-scale feature centrality of ZK03 .

[0115] Similarly, calculate the multi-scale feature centrality of all nodes.

[0116] Step 3: Normalize to obtain geological response model values.

[0117] Perform max-min normalization on the centrality of all nodes so that its range falls within the interval [0,1]. Assume that the maximum centrality of all multi-scale features is 2.5 and the minimum is 0.2.

[0118] The geological response model value of ZK03 is R3=(C3-0.2) / (2.5-0.2)=(1.057-0.2) / 2.3≈0.373.

[0119] Normalize all nodes to obtain the geological response model value for each point.

[0120] Step 4: Divide the dataset according to the geological response model values

[0121] The K-means clustering algorithm is used, with the geological response pattern values ​​R of all 6 nodes as input features, and the preset number of clusters K=2. After algorithm iteration, the nodes may be divided into two sets:

[0122] Set A (high response mode values, possibly representing the core feature region): {ZK01 (R = 0.82), ZK02 (R = 0.78), ZK04 (R = 0.70)}

[0123] Set B (low response mode values, possibly representing feature transitions or edge regions): {ZK03 (R=0.37), ZK05 (R=0.25), ZK06 (R=0.15)}

[0124] Step 5: Spatial connectivity test and set adjustment

[0125] Perform spatial connectivity checks on each initially partitioned set:

[0126] Calculate the convex hull volume V_A of the three-dimensional spatial coordinates of points in set A and the maximum Euclidean distance L_A_max between pairs of points. Calculate the ratio r_A = V_A / (L_A_max³).

[0127] If r_A is less than a preset spatial compactness threshold (e.g., 0.1), it is considered that the points in the set may be too scattered or discontinuous in space, and there are "enclaves".

[0128] Hypothesis testing revealed that the r_A value of set A was normal, while the r_B value of set B was too low. Further investigation revealed that ZK03, ZK05, and ZK06 in set B were spatially far apart, separated by points in set A.

[0129] Perform set splitting operation: Perform quadratic spatial clustering (such as DBSCAN) based on three-dimensional coordinates on the points of set B, splitting it into two spatially compact subsets B1 and B2.

[0130] Ultimately, we obtained three sets of exploration points that are spatially continuous and have similar internal geological response model values: set A, set B1, and set B2. In practical applications, the preliminary division results will be compared with existing regional geological maps or structural maps. If a set clearly crosses a known geological boundary (such as a fault), the set will be fine-tuned based on that boundary to ensure that the division results conform to geological laws.

[0131] In summary, by introducing geological response pattern values ​​based on multi-scale networks and rigorous spatial connectivity checks, intelligent and rational grouping of exploration points was achieved, resulting in significant multi-dimensional technical effects. First, the defined geological response pattern values, through weighted summation of connection strengths at different spatial scales, quantify the "feature centrality" of an exploration point within the entire network. A point with a high response pattern value indicates that it occupies a pivotal position in the multi-scale spatial network, closely connected to numerous points with similar characteristics. This typically corresponds to the core area of ​​a geological body with homogeneous lithology and stable thickness; conversely, it may represent a transitional zone or a complex lithological area. This measurement method, compared to traditional clustering methods that solely rely on spatial coordinates or original features, more profoundly reveals the structural position of a point within the "spatial distribution pattern of geological attributes." Second, clustering based on these pattern values ​​essentially categorizes points with similar network centrality (i.e., similar geological structural roles), resulting in initial sets with greater intrinsic homogeneity in characteristics. Most importantly, the subsequent spatial connectivity verification and adjustment mechanism enforces the fundamental geological constraint that "the same geological unit should have spatial continuity." It automatically identifies and corrects spatial fragmentation or unreasonable results that may arise from simple numerical clustering (such as grouping spatially distant points with coincidentally similar pattern values ​​together). This ensures that each set of points ultimately divided not only has similar characteristic attributes but also constitutes a continuous region in three-dimensional space. This greatly enhances the geological interpretability and practicality of the clustering results, enabling them to more accurately correspond to real underground geological bodies (such as a continuous sand body or a complete lithological segment). In summary, this step combines complex network topology analysis, numerical clustering, and rigorous spatial logic verification, outputting spatially continuous, internally coordinated, and geologically significant "exploration point sets." This lays an ideal analytical unit foundation for subsequent representative point selection and efficient type identification, representing a crucial transformation step from discrete data points to a coherent understanding of geological bodies.

[0132] Step S105: Obtain the three-dimensional spatial coordinates of each exploration point in a certain set of exploration points, and select at least one target exploration point in the certain set of exploration points according to the three-dimensional spatial coordinates and a preset sequence selection rule.

[0133] In this step, a geometric center of the three-dimensional spatial coordinates of all exploration points in the set of exploration points is calculated, and based on the geometric center, the three-dimensional Euclidean distance from each exploration point in the set of exploration points to the geometric center is calculated.

[0134] The top K exploration points with the smallest three-dimensional Euclidean distance to a certain geometric center are identified as candidate spatial centers, and the geological response model values ​​and comprehensive geological feature description vectors of each candidate spatial center are obtained, where K is a preset positive integer;

[0135] Based on the comprehensive geological feature description vector of all exploration points in the set of exploration points, the principal component analysis method is used to extract the feature direction of the set of exploration points in the comprehensive geological feature space.

[0136] Calculate the projection value of the comprehensive geological feature description vector of each spatial center candidate point in the feature direction, and then normalize and weight the projection values ​​and the corresponding geological response model values ​​to obtain the comprehensive representative score of each spatial center candidate point.

[0137] The candidate spatial center with the highest comprehensive representative score was selected as the first target exploration point.

[0138] Calculate the remaining Euclidean distances between the comprehensive geological feature description vectors of the remaining spatial candidate points and the comprehensive geological feature description vectors of the first target exploration point, and fuse the remaining Euclidean distances with the remaining comprehensive representative scores of the remaining spatial candidate points to obtain the supplementary selection scores of the remaining spatial candidate points.

[0139] Select the remaining spatial candidate points from the top s in the supplementary selection score as target exploration points, thus obtaining at least one target exploration point. The at least one target exploration point includes the spatial center candidate point with the highest comprehensive representative score and the remaining spatial candidate points from the top s in the supplementary selection score.

[0140] In one specific embodiment, it is assumed that a set of exploration points contains 7 borehole points: P1 to P7, each point having known three-dimensional spatial coordinates, geological response pattern values ​​(from step S104), and a comprehensive geological feature description vector (from step S102, assumed to be a 4-dimensional vector for simplification).

[0141] Step 1: Calculate the geometric center of the set and the distance from each point to the geometric center. The three-dimensional spatial coordinates of the 7 points are as follows (unit: meters):

[0142] P1:(100,200,-50),P2:(110,190,-55),P3:(95,210,-48),P4:(105,195,-52),P5:(98,205,-49),P6:(112,185,-58),P7:(102,198,-51);

[0143] Computational geometry center O:

[0144] O_x=(100+110+95+105+98+112+102) / 7=722 / 7≈103.14;

[0145] O_y=(200+190+210+195+205+185+198) / 7=1383 / 7≈197.57;

[0146] O_z=(-50-55-48-52-49-58-51) / 7=-363 / 7≈-51.86;

[0147] Therefore, the geometric center O≈(103.14,197.57,-51.86).

[0148] Calculate the 3D Euclidean distance d_i from each point to O. For example, the distance from P1 to O:

[0149] rice.

[0150] Similar to calculating all distances, let's assume they are sorted from smallest to largest as follows: d4=2.1m, d7=2.3m, d5=3.8m, d1=4.4m, d2=5.2m, d3=6.1m, d6=7.5m.

[0151] Set K=3, and select the three points with the smallest distance as candidate points for the spatial center: P4, P7, P5.

[0152] Step 2: Obtain the geological response model values ​​and feature vectors of candidate points.

[0153] Assume the geological response model values ​​for the three candidate points are: R4=0.85, R7=0.78, and R5=0.72.

[0154] Assume their comprehensive geological feature description vector (4-dimensional) is as follows:

[0155] P4: [0.9, 0.2, 0.6, 0.3];

[0156] P7: [0.8, 0.3, 0.7, 0.4];

[0157] P5: [0.7, 0.4, 0.5, 0.5];

[0158] Step 3: Principal Component Analysis to Extract Feature Directions from the Set

[0159] Principal component analysis (PCA) is performed using the 4-dimensional eigenvectors of all 7 points in the set. The direction (unit vector) of the first principal component (PC1) is calculated and assumed to be: PC1=[0.707,0.000,0.707,0.000] (simplified schematic value, actually calculated).

[0160] Step 4: Calculate the projected values, normalize them, and obtain the comprehensive representative score.

[0161] Calculate the projection (dot product) of the feature vector of each candidate point onto the PC1 direction:

[0162] ;

[0163] ;

[0164] ;

[0165] Max-min normalization was performed on both the projected values ​​and the geological response model values:

[0166] Projection value range: min_proj=0.8484, max_proj=1.0605;

[0167] Calculate the normalized projection value of the feature vector of each candidate point along the PC1 direction:

[0168] proj4_norm=(1.0605-0.8484) / (1.0605-0.8484)=1.0;

[0169] proj7_norm=(1.0605-0.8484) / (1.0605-0.8484)=1.0;

[0170] proj5_norm=(0.8484-0.8484) / (1.0605-0.8484)=0.0;

[0171] Geological response model value range: min_R=0.72, max_R=0.85;

[0172] Calculate the normalized geological response model value of the feature vector of each candidate point in the PC1 direction.

[0173] R4_norm=(0.85-0.72) / (0.85-0.72)=1.0;

[0174] R7_norm=(0.78-0.72) / 0.13≈0.4615;

[0175] R5_norm=(0.72-0.72) / 0.13=0.0;

[0176] Let the weights be: projected value weight w_p = 0.6, geological response model value weight w_r = 0.4. Calculate the comprehensive representative score:

[0177] ;

[0178] ;

[0179] ;

[0180] ;

[0181] Therefore, P4 had the highest overall representative score (1.0) and was selected as the first target exploration site.

[0182] Step 5: Calculate the supplementary selection score and select the remaining target points.

[0183] Suppose we need to select s=1 more target points. The remaining candidate points are P7 and P5.

[0184] First, calculate the Euclidean distance between the remaining candidate points and the comprehensive geological feature description vector of the first target point P4:

[0185] The Euclidean distance between the comprehensive geological feature description vector of P7 and the comprehensive geological feature description vector of P4:

[0186] ,

[0187] The Euclidean distance between the comprehensive geological feature description vector of P5 and the comprehensive geological feature description vector of P4:

[0188] ,

[0189] Distance normalization (maximum-minimum normalization): maximum distance 0.3606, minimum 0.2;

[0190] The normalized Euclidean distance between the comprehensive geological feature description vectors of P7 and P4:

[0191] d47_norm=(0.2-0.2) / (0.3606-0.2)=0;

[0192] The normalized Euclidean distance between the comprehensive geological feature description vectors of P5 and P4:

[0193] d45_norm=(0.3606-0.2) / 0.1606≈1.0;

[0194] The combined representative scores of the remaining candidate points (Score7=0.7846, Score5=0.0) also need to be renormalized (within the remaining candidate points):

[0195] min_Score=0.0,max_Score=0.7846.

[0196] Score7_norm=(0.7846-0.0) / 0.7846=1.0;

[0197] Score5_norm=(0.0-0.0) / 0.7846=0.0;

[0198] Set the fusion weights as follows: distance normalized value weight w_d=0.4, comprehensive representative score normalized value weight w_s=0.6 (points with significant differences in features from the first target point and strong representativeness are encouraged to be selected).

[0199] Calculate supplementary selection scores:

[0200]

[0201] ;

[0202] ;

[0203] Therefore, P7 received the highest supplementary selection score (0.6) and was selected as the second target exploration site.

[0204] Ultimately, the target exploration points selected for this set were P4 and P7.

[0205] In summary, by calculating the spatial geometric center and selecting neighboring points as candidates, it is ensured that the candidate points are spatially closely distributed around the "centroid" of the set. This lays the foundation for selecting target points with representative spatial locations, conforming to the concentrated pattern of the spatial distribution of geological bodies. Secondly, principal component analysis is introduced to extract the main characteristic variation directions of the set as a whole, and the projection of candidate points in these directions is calculated. This achieves an objective quantification of the "characteristic representativeness" of the candidate points, ensuring that the characteristic attributes of the selected target points can explain the characteristic variation of the entire set to the greatest extent, thus representing the typical lithological properties of the set. Furthermore, the characteristic projection values ​​are weighted and fused with the geological response model values ​​that characterize the importance of the point network to obtain a comprehensive representative score. This scoring mechanism organically integrates "spatial proximity," "characteristic typicality," and "network centrality." Three key dimensions ensure that the final selected first target point is not only geographically central and typical in terms of characteristic attributes, but also occupies a core hub position in the global correlation network, making it a truly comprehensive "optimal observation point" that can represent the geological characteristics of the set. Finally, when selecting multiple target points, the difference between the remaining candidate points and the first target point in the feature space (Euclidean distance) is calculated, and then fused with their respective representative scores to obtain a supplementary selection score. This strategy effectively balances "representativeness" and "diversity," avoiding redundancy caused by overly similar features among the selected target points, while ensuring that the supplementary target points themselves have strong representativeness. As a result, the final set of selected target points can cover and represent the feature space variation range of the entire set to the maximum extent with the fewest number of points. In summary, this step, through rigorous mathematical modeling and multi-criteria decision-making, enables the automatic and efficient selection of a concise yet comprehensive "representative set" from the point set. These high-quality target points will provide the most informative input for the subsequent geological body type identification model based on multi-scale geological association networks, significantly improving the accuracy and efficiency of model identification. At the same time, it provides geological interpreters with clear key control points, greatly enhancing the reliability and interpretability of the entire identification process.

[0206] Step S106: Input the geological exploration data sequence corresponding to the at least one target exploration point into the preset geological body type identification model based on multi-scale geological association network to obtain the preliminary geological body type identification result based on multi-scale geological association network corresponding to the at least one target exploration point.

[0207] In this step, a deep learning model based on a Long Short-Term Memory (LSTM) network is constructed, and the deep learning model is trained to obtain a geological body type identification model based on a multi-scale geological association network. Its structure is as follows:

[0208] Input layer: Receives a two-dimensional tensor of shape (sequence length T, number of feature parameters M). For example, sequence length T = 100 (representing 100 depth sampling points), and number of feature parameters M = 4 (such as density, sound velocity, resistivity, natural gamma).

[0209] Hidden layer: Contains 2 layers of LSTM cells, each with 128 hidden cells. LSTM can effectively capture the sequence dependencies and long-term patterns of geological parameters as they change with depth.

[0210] Output layer: A fully connected layer followed by a Softmax activation function. The number of neurons is equal to the predefined number of geological body types, C (e.g., C=3, corresponding to sandstone, mudstone, and limestone). The output is a C-dimensional probability distribution vector, where each element represents the probability of belonging to the corresponding geological body type, and the sum of all elements is 1.

[0211] The training process of the model:

[0212] Data preparation: A borehole dataset containing a large number of known geological body type labels was collected as the training and validation sets. Each sample is a geological exploration data sequence and its corresponding real geological body type label (e.g., "sandstone").

[0213] Data preprocessing: Standardize each data sequence (e.g., perform Z-score standardization on each feature parameter separately) to eliminate the influence of dimensions and accelerate model convergence.

[0214] Model Training: Using the training set data, the LSTM model described above is iteratively trained with classification cross-entropy as the loss function and Adam as the optimizer. During training, the model's performance is monitored using a validation set to prevent overfitting. Training is stopped when the validation set loss no longer decreases significantly, and the optimal model parameters are saved. The final result is a well-trained geological body type recognition model based on a multi-scale geological association network, capable of predicting the probability distribution of geological body types based on input sequences.

[0215] Step S107: Based on the preliminary geological body type identification results based on the multi-scale geological association network, a multi-source decision fusion mechanism is used to generate the final target geological body type identification result based on the multi-scale geological association network.

[0216] In this step, preliminary geological body type identification results based on multi-scale geological association networks are obtained for each target exploration point. A preliminary geological body type identification result based on multi-scale geological association networks includes a probability distribution of a target exploration point belonging to various predefined geological body types.

[0217] Obtain the weight information corresponding to each target exploration point, wherein the weight information includes the comprehensive representative score of the first target exploration point and the supplementary selection score of the other target exploration points;

[0218] The weight information of all target exploration points is normalized to obtain the normalized weight of each target exploration point.

[0219] The probability of a certain geological body type in the preliminary geological body type identification results based on a multi-scale geological association network for a certain target exploration point is multiplied by the normalized weight of the target exploration point to obtain the weighted probability of a certain geological body type.

[0220] The total weighted probability of a certain geological body type is obtained by summing the weighted probabilities of all target exploration points corresponding to a certain geological body type.

[0221] Compare the total weighted probabilities of all predefined geological body types, and determine the geological body type with the highest total weighted probability as the final target geological body type identification result based on the multi-scale geological association network.

[0222] In one specific embodiment, based on the preliminary results obtained from independently identifying geological body types of multiple target exploration points using a multi-scale geological association network in step S106, a weighted fusion mechanism is used to generate the final geological body type identification result based on the multi-scale geological association network, representing the entire set of points. The following is a specific example:

[0223] Suppose that a set of exploration points has selected three target exploration points: P_A (first target point), P_B and P_C, and there are three predefined geological body types: sandstone, mudstone and limestone;

[0224] Step 1: Obtain preliminary geological body type identification results based on multi-scale geological association networks

[0225] The geological exploration data sequence for each target point is input into a pre-set geological body type identification model based on a multi-scale geological correlation network (such as a trained deep learning classifier). The model outputs the probability distribution of each point belonging to different geological body types, assuming the following results:

[0226] Preliminary identification results of P_A: Sandstone probability = 0.70, Mudstone probability = 0.20, Limestone probability = 0.10;

[0227] Preliminary identification results of P_B: sandstone probability = 0.60, mudstone probability = 0.30, limestone probability = 0.10;

[0228] Preliminary identification results for P_C: Sandstone probability = 0.80, Mudstone probability = 0.15, Limestone probability = 0.05;

[0229] Step 2: Obtain the weight information of each target point.

[0230] Based on the selection result of step S105:

[0231] P_A, as the first target point, has a comprehensive representative score of 0.85;

[0232] P_B, as one of the remaining target points, has a supplementary selection score of 0.75.

[0233] P_C, as one of the remaining target points, has a supplementary selection score of 0.65;

[0234] Step 3: Calculate the normalized weights

[0235] The weights of all target points are normalized so that their sum is 1.

[0236] The total weights are 0.85 + 0.75 + 0.65 = 2.25.

[0237] The normalized weight of P_A is W_A' = 0.85 / 2.25 ≈ 0.378;

[0238] The normalized weight of P_B ​​is W_B' = 0.75 / 2.25 ≈ 0.333;

[0239] The normalized weight of P_C is W_C' = 0.65 / 2.25 ≈ 0.289;

[0240] Step 4: Calculate the weighted probability of each geological type.

[0241] For each type of geological body, the identification probability of each target point is multiplied by its normalized weight, and then the sum is obtained for all points to obtain the total weighted probability of that type.

[0242] For sandstone:

[0243] Contributions from P_A: ;

[0244] Contributions from P_B: ;

[0245] Contributions from P_C: ;

[0246] The total weighted probability of sandstone = 0.265 + 0.200 + 0.231 = 0.696;

[0247] For mudstone:

[0248] Contributions from P_A: ;

[0249] Contributions from P_B: ;

[0250] Contributions from P_C: ;

[0251] The total weighted probability of mudstone is 0.076 + 0.100 + 0.043 = 0.219;

[0252] For limestone:

[0253] Contributions from P_A: ;

[0254] Contributions from P_B: ;

[0255] Contributions from P_C: ;

[0256] The total weighted probability of limestone = 0.038 + 0.033 + 0.014 = 0.085;

[0257] Step 5: Compare and determine the final target based on the geological body type identification results of the multi-scale geological association network.

[0258] Compare the total weighted probabilities of the three geological body types:

[0259] Sandstone (0.696) > Mudstone (0.219) > Limestone (0.085)

[0260] Sandstone has the highest total weighted probability.

[0261] Therefore, the final target of this set of exploration sites was determined to be "sandstone" based on the geological body type identification results of the multi-scale geological correlation network. This means that, after synthesizing the identification opinions of three target sites with different representativeness and credibility, the geological unit represented by this set is most likely sandstone.

[0262] In summary, a weighted system based on the "representative strength" of the points was introduced. This weight directly derives from the comprehensive representative score and supplementary selection score scientifically calculated in step S105. This ensures that, in the final decision-making process, target points that are more representative in terms of space, features, and network (i.e., more reliable and typical geological observation points) have greater influence, thus guaranteeing the quality and authority of the decision-making basis from the outset. Secondly, this method explicitly models and quantifies uncertainty at the decision-making level. It does not rely on the "hard labels" output by a single model but fully utilizes the soft information—the probability distribution output by the classification model—through weighted fusion. By integrating multiple probability distributions into a more robust and reliable ensemble probability distribution, this process effectively smooths out the risk of misjudgment that may arise from local noise or model limitations in individual models, significantly enhancing the fault tolerance and anti-interference capability of the decision-making process. Furthermore, this fusion method is mathematically equivalent to a group decision optimization process, aiming to maximize consistency with the identification opinions of all target points while weighting them according to the credibility of each point. The final output, the "total weighted probability," not only indicates the most likely geological body type, but its numerical value also reflects the collective confidence level of the conclusion, providing a valuable reliability metric for subsequent geological interpretation. In summary, this step, through rigorous mathematical fusion, elevates multiple independent but inconsistent local identification opinions into a global geological body type determination conclusion with higher credibility, stronger robustness, and a clear confidence level indication. This ensures that the final output of the entire identification method chain is not a simple accumulation of isolated point conclusions, but a consistent and reliable judgment that has undergone collaborative optimization and can represent the attributes of the entire geological unit, greatly enhancing the geological practical value and decision support effectiveness of the results.

[0263] Step S108: Based on the similarity of geological response patterns between other exploration points and the at least one target exploration point, spatial propagation is performed on the geological body type identification results of the final target based on the multi-scale geological association network to obtain other geological body type identification results based on the multi-scale geological association network corresponding to other exploration points.

[0264] In this step, the absolute difference between the first geological response model value of the first exploration point in the set of exploration points and the geological response model value of each of the at least one target exploration point is calculated to obtain at least one first model difference degree, wherein the first exploration point is any one of the other exploration points;

[0265] Divide at least one first mode difference by the difference between the maximum and minimum values ​​of the geological response mode values ​​of all exploration points in the set of exploration points, and normalize it to obtain the corresponding normalized difference.

[0266] Calculate the reciprocal of each normalized dissimilarity to obtain at least one initial similarity;

[0267] The three-dimensional Euclidean distance between the first exploration point and the at least one target exploration point is obtained, and the at least one initial similarity is corrected based on each three-dimensional Euclidean distance to obtain the geological response pattern similarity between the first exploration point and the at least one target exploration point.

[0268] From the at least one target exploration point, select M target exploration points that have the highest similarity to the final geological response pattern of the first exploration point as reference target exploration points;

[0269] Based on the geological body type corresponding to the reference target exploration point in the geological body type identification result of the final target based on the multi-scale geological association network, and the similarity of the geological response patterns between the first exploration point and each reference target exploration point, a certain propagation probability of the first exploration point belonging to a certain geological body type is calculated.

[0270] The geological body type identification result based on a multi-scale geological association network is used as the geological body type corresponding to the highest propagation probability as the first exploration point.

[0271] In one specific embodiment, assume a set of exploration points includes: target exploration point P1, target exploration point P2, and target exploration point P3, with geological response model values ​​of R1=0.85, R2=0.78, and R3=0.72, respectively, and three-dimensional spatial coordinates of P1(100,150,-50), P2(110,140,-52), and P3(105,145,-51). Step S107 has determined that the final target of this set, based on the geological body type identification result of the multi-scale geological association network, is sandstone.

[0272] Other exploration points (first exploration point): Px, with a geological response model value of Rx=0.65 and three-dimensional spatial coordinates: Px(102,148,-51);

[0273] Step 1: Calculate the first mode difference

[0274] Calculate the absolute difference between Px and the geological response model value for each target point:

[0275] diff_x1 = |0.65 - 0.85| = 0.20;

[0276] diff_x2 = |0.65 - 0.78| = 0.13;

[0277] diff_x3 = |0.65 - 0.72| = 0.07;

[0278] The three first-mode differences were obtained: [0.20, 0.13, 0.07].

[0279] Step 2: Calculate the normalized variance

[0280] The maximum value of the geological response model for all points (P1, P2, P3, Px) in the set is 0.85, the minimum value is 0.65, and the difference is 0.20.

[0281] Divide each first-mode difference by this difference:

[0282] norm_diff_x1 = 0.20 / 0.20 = 1.0;

[0283] norm_diff_x2=0.13 / 0.20=0.65;

[0284] norm_diff_x3=0.07 / 0.20=0.35;

[0285] These values ​​have been normalized in the range [0,1], resulting in the normalized difference.

[0286] Step 3: Calculate the initial similarity

[0287] Calculate the reciprocal of each normalized difference (to avoid division by zero, it is usually added by 1 and the reciprocal is taken, or 1 - normalized difference is used directly as the similarity, the latter is more intuitive here, the smaller the geological difference, the higher the similarity):

[0288] Initial similarity = 1 - normalized difference

[0289] sim_init_x1=1-1.0=0.0;

[0290] sim_init_x2=1-0.65=0.35;

[0291] sim_init_x3=1-0.35=0.65;

[0292] At least one initial similarity is obtained: [0.0, 0.35, 0.65].

[0293] Step 4: Calculate the similarity of geological response patterns (distance correction)

[0294] Calculate the three-dimensional Euclidean distance between Px and each target point:

[0295] ;

[0296] ;

[0297] ;

[0298] Set a distance decay coefficient D0 (e.g., 10 meters). Correct the initial similarity using the formula: Final similarity = Initial similarity / (1 + Distance / D0).

[0299] sim_final_x1=0.0 / (1+3 / 10)=0.0;

[0300] sim_final_x2=0.35 / (1+11.36 / 10)≈0.35 / 2.136≈0.164;

[0301] sim_final_x3=0.65 / (1+4.24 / 10)≈0.65 / 1.424≈0.456;

[0302] The similarity of geological response patterns between Px and each target point is: [0.0, 0.164, 0.456].

[0303] Step 5: Select reference target exploration points

[0304] The two target points with the highest similarity to Px were selected as reference target exploration points. Sorted by similarity: P3 (0.456) > P2 (0.164) > P1 (0.0).

[0305] Therefore, P3 and P2 were selected as reference target exploration points.

[0306] Step 6: Calculate the propagation probability

[0307] It is known that the final target is sandstone, based on the geological body type identification result of the multi-scale geological association network, and all target points (P1, P2, P3) correspond to the type of sandstone.

[0308] The similarity between Px and the reference target points P2 and P3 are sim_x2=0.164 and sim_x3=0.456, respectively.

[0309] Calculate the probability that Px will be propagated as sandstone (propagation probability):

[0310] First, the similarity of the normalized reference points is used as the weight:

[0311] Total similarity = 0.164 + 0.456 = 0.620;

[0312] Weight_w2 = 0.164 / 0.620 ≈ 0.264;

[0313] Weight_w3 = 0.456 / 0.620 ≈ 0.736;

[0314] Since both reference target points P2 and P3 are sandstone, the propagation probability of Px belonging to sandstone is:

[0315] P_sand = weight_w2 1+weight_w3 1 = 0.264 + 0.736 = 1.0 (If there are non-sandstone types in the reference point, the propagation probability is the sum of the corresponding weights).

[0316] Step 7: Determine the geological body type identification results of Px based on multi-scale geological association networks.

[0317] The propagation probability P_sand=1.0 is the highest probability (and unique). Therefore, the geological body type identification result of Px based on the multi-scale geological association network is determined to be sandstone.

[0318] Repeat the above steps for all other exploration points in the set to obtain the identification results for all other points.

[0319] In summary, the method presented in this application, at the data representation level, generates a comprehensive feature description with highly condensed information and clear geological significance by adaptive depth segmentation and multi-scale (macro, meso, and micro) feature fusion, combined with confidence weighting using a geological knowledge graph, providing a powerful input for subsequent analysis. At the spatial modeling level, it innovatively constructs a multi-scale association network that integrates spatial distance and feature similarity, capable of simultaneously depicting local close relationships and regional trend connections, revealing more precisely the spatial-attribute coupling relationships between exploration points that conform to geological laws. Regarding computational efficiency and reliability, the method quantifies the geological response pattern values ​​of points based on network topology, and accordingly divides a set of spatially continuous and feature-uniform points. Then, through a multi-criteria fusion strategy (spatial centrality, feature representativeness, and network importance), it selects a few of the most representative target points from each set for high-precision model identification, avoiding the huge overhead of independently modeling all points, thus optimizing computational resources and reducing costs while increasing efficiency. Finally, by using multi-source weighted decision fusion and a pattern similarity-based intelligent spatial propagation mechanism, the highly reliable target point identification results are reasonably and robustly promoted to the entire region. While ensuring spatial continuity and geological consistency, the identification results of all points are obtained efficiently.

[0320] Please see Figure 2 The diagram shows a structural block diagram of a geological body type identification system based on a multi-scale geological association network according to this application.

[0321] like Figure 2As shown, the geological body type identification system 200 based on multi-scale geological association network includes an acquisition module 210, a generation module 220, a construction module 230, a division module 240, a selection module 250, an identification module 260, a fusion module 270, and an output module 280.

[0322] The system includes the following modules: Acquisition module 210, configured to acquire geological exploration data for at least one exploration point within a preset exploration depth range, and sort the geological exploration data for the same exploration point to obtain at least one geological exploration data sequence; Generation module 220, configured to perform multi-dimensional geological feature fusion analysis based on the at least one geological exploration data sequence to generate a comprehensive geological feature description for each exploration point; Construction module 230, configured to construct a multi-scale geological association network between exploration points based on the comprehensive geological feature descriptions and three-dimensional spatial coordinates of each exploration point; Division module 240, configured to determine the geological response mode of each exploration point based on the multi-scale geological association network, and divide the exploration points according to the geological response mode to obtain at least one set of exploration points; and Selection module 250, configured to acquire the three-dimensional spatial coordinates of each exploration point in a set of exploration points, and select points according to a preset sequence selection rule. Then, at least one target exploration point is selected from the set of exploration points; the identification module 260 is configured to input the geological exploration data sequence corresponding to the at least one target exploration point into a preset geological body type identification model based on a multi-scale geological association network, to obtain preliminary geological body type identification results based on a multi-scale geological association network corresponding to the at least one target exploration point; the fusion module 270 is configured to generate the final target geological body type identification result based on a multi-source decision fusion mechanism based on each preliminary geological body type identification result based on a multi-scale geological association network; the output module 280 is configured to spatially propagate the final target geological body type identification result based on a multi-scale geological association network based on the similarity of geological response patterns between other exploration points and the at least one target exploration point, to obtain other geological body type identification results based on a multi-scale geological association network corresponding to other exploration points.

[0323] It should be understood that Figure 2 The modules and references described in the document Figure 1 The steps described in the text correspond to those in the method described above. Therefore, the operations, features, and corresponding technical effects described above also apply to the method described in the text. Figure 2 The various modules in the document will not be described in detail here.

[0324] In other embodiments, the present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein when the program instructions are executed by a processor, the processor performs the geological body type identification method based on a multi-scale geological association network in any of the above method embodiments.

[0325] In one embodiment, the computer-readable storage medium of the present invention stores computer-executable instructions, which are configured as follows:

[0326] Obtain geological exploration data for at least one exploration point within a preset exploration depth range, and sort the geological exploration data for the same exploration point to obtain at least one geological exploration data sequence.

[0327] Based on the at least one geological exploration data sequence, a multi-dimensional geological feature fusion analysis is performed to generate a comprehensive geological feature description for each exploration point.

[0328] Based on the comprehensive geological feature descriptions of each exploration point and the three-dimensional spatial coordinates of each exploration point, a multi-scale geological correlation network is constructed among the exploration points.

[0329] Based on the multi-scale geological correlation network, the geological response model value of each exploration point is determined, and the exploration points are divided according to the geological response model value to obtain at least one set of exploration points.

[0330] Obtain the three-dimensional spatial coordinates of each exploration point in a certain set of exploration points, and select at least one target exploration point in the certain set of exploration points according to the three-dimensional spatial coordinates and a preset sequence selection rule.

[0331] The geological exploration data sequence corresponding to the at least one target exploration point is input into a preset geological body type identification model based on a multi-scale geological association network to obtain preliminary geological body type identification results based on a multi-scale geological association network corresponding to the at least one target exploration point.

[0332] Based on the preliminary geological body type identification results based on multi-scale geological association networks, a multi-source decision fusion mechanism is adopted to generate the final target geological body type identification results based on multi-scale geological association networks.

[0333] Based on the similarity of geological response patterns between other exploration sites and the at least one target exploration site, the geological body type identification results of the final target based on the multi-scale geological association network are spatially propagated to obtain other geological body type identification results based on the multi-scale geological association network corresponding to other exploration sites.

[0334] Computer-readable storage media may include a program storage area and a data storage area, wherein the program storage area may store an operating system and an application program required for at least one function; the data storage area may store data created based on the use of a geological body type identification system based on a multi-scale geological association network. Furthermore, the computer-readable storage medium may include high-speed random access memory, and may also include memory, such as at least one disk storage device, flash memory device, or other non-volatile solid-state storage device. In some embodiments, the computer-readable storage medium may optionally include memory remotely disposed relative to a processor, which can be connected to the geological body type identification system based on a multi-scale geological association network via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0335] Figure 3 This is a schematic diagram of the structure of the electronic device provided in the embodiment of the present invention, such as... Figure 3 As shown, the device includes a processor 310 and a memory 320. The electronic device may also include an input device 330 and an output device 340. The processor 310, memory 320, input device 330, and output device 340 can be connected via a bus or other means. Figure 3 Taking a bus connection as an example, the memory 320 is the computer-readable storage medium described above. The processor 310 executes various server functions and data processing by running non-volatile software programs, instructions, and modules stored in the memory 320, thereby implementing the geological body type identification method based on a multi-scale geological association network as described in the above-described method embodiment. The input device 330 can receive input digital or character information and generate key signal inputs related to user settings and function control of the geological body type identification system based on a multi-scale geological association network. The output device 340 may include a display screen or other display device.

[0336] The aforementioned electronic device can execute the method provided in the embodiments of the present invention, and has the corresponding functional modules and beneficial effects for executing the method. Technical details not described in detail in this embodiment can be found in the method provided in the embodiments of the present invention.

[0337] In one implementation, the above-described electronic device is applied to a geological body type identification system based on a multi-scale geological association network, serving as a client, and includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to:

[0338] Obtain geological exploration data for at least one exploration point within a preset exploration depth range, and sort the geological exploration data for the same exploration point to obtain at least one geological exploration data sequence.

[0339] Based on the at least one geological exploration data sequence, a multi-dimensional geological feature fusion analysis is performed to generate a comprehensive geological feature description for each exploration point.

[0340] Based on the comprehensive geological feature descriptions of each exploration point and the three-dimensional spatial coordinates of each exploration point, a multi-scale geological correlation network is constructed among the exploration points.

[0341] Based on the multi-scale geological correlation network, the geological response model value of each exploration point is determined, and the exploration points are divided according to the geological response model value to obtain at least one set of exploration points.

[0342] Obtain the three-dimensional spatial coordinates of each exploration point in a certain set of exploration points, and select at least one target exploration point in the certain set of exploration points according to the three-dimensional spatial coordinates and a preset sequence selection rule.

[0343] The geological exploration data sequence corresponding to the at least one target exploration point is input into a preset geological body type identification model based on a multi-scale geological association network to obtain preliminary geological body type identification results based on a multi-scale geological association network corresponding to the at least one target exploration point.

[0344] Based on the preliminary geological body type identification results based on multi-scale geological association networks, a multi-source decision fusion mechanism is adopted to generate the final target geological body type identification results based on multi-scale geological association networks.

[0345] Based on the similarity of geological response patterns between other exploration sites and the at least one target exploration site, the geological body type identification results of the final target based on the multi-scale geological association network are spatially propagated to obtain other geological body type identification results based on the multi-scale geological association network corresponding to other exploration sites.

[0346] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of various embodiments or some parts of embodiments.

[0347] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of various embodiments or some parts of embodiments.

[0348] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for identifying geological body types based on multi-scale geological association networks, characterized in that, include: Obtain geological exploration data for at least one exploration point within a preset exploration depth range, and sort the geological exploration data for the same exploration point to obtain at least one geological exploration data sequence. Based on the at least one geological exploration data sequence, a multi-dimensional geological feature fusion analysis is performed to generate a comprehensive geological feature description for each exploration point. Based on the comprehensive geological feature descriptions of each exploration point and the three-dimensional spatial coordinates of each exploration point, a multi-scale geological correlation network is constructed among the exploration points. Based on the multi-scale geological correlation network, the geological response model value of each exploration point is determined, and the exploration points are divided according to the geological response model value to obtain at least one set of exploration points. Obtain the three-dimensional spatial coordinates of each exploration point in a certain set of exploration points, and select at least one target exploration point in the certain set of exploration points according to the three-dimensional spatial coordinates and a preset sequence selection rule. The geological exploration data sequence corresponding to the at least one target exploration point is input into a preset geological body type identification model based on a multi-scale geological association network to obtain preliminary geological body type identification results based on a multi-scale geological association network corresponding to the at least one target exploration point. Based on the preliminary geological body type identification results based on multi-scale geological association networks, a multi-source decision fusion mechanism is used to generate the final target geological body type identification result based on multi-scale geological association networks. The process of generating the geological body type identification result includes: Obtain preliminary geological body type identification results based on a multi-scale geological association network corresponding to each target exploration point. A preliminary geological body type identification result based on a multi-scale geological association network includes the probability distribution of a target exploration point belonging to various predefined geological body types. Obtain the weight information corresponding to each target exploration point, wherein the weight information includes the comprehensive representative score of the first target exploration point and the supplementary selection score of the other target exploration points; The weight information of all target exploration points is normalized to obtain the normalized weight of each target exploration point. The probability of a certain geological body type in the preliminary geological body type identification results based on a multi-scale geological association network for a certain target exploration point is multiplied by the normalized weight of the target exploration point to obtain the weighted probability of a certain geological body type. The total weighted probability of a certain geological body type is obtained by summing the weighted probabilities of all target exploration points corresponding to a certain geological body type. Compare the total weighted probabilities of all predefined geological body types, and determine the geological body type with the highest total weighted probability as the geological body type identification result based on the multi-scale geological association network as the final target. Based on the similarity of geological response patterns between other exploration sites and the at least one target exploration site, the geological body type identification results of the final target based on the multi-scale geological association network are spatially propagated to obtain other geological body type identification results based on the multi-scale geological association network corresponding to other exploration sites.

2. The geological body type identification method based on a multi-scale geological association network according to claim 1, characterized in that, The step of performing multi-dimensional geological feature fusion analysis based on the at least one geological exploration data sequence to generate a comprehensive geological feature description for each exploration point includes: Based on the changes in the values ​​of various geological feature parameters with exploration depth in the current geological exploration data sequence at the current exploration point, multiple abrupt changes in geological feature depth points are identified. These abrupt changes include: Calculate the first-order difference between adjacent sampling depths for each geological feature parameter in the current geological exploration data sequence; The sampling point depth whose absolute value of the first-order difference exceeds the preset mutation threshold is marked as the potential mutation depth; The potential abrupt change depths of all geological feature parameters are subjected to union processing and depth-near clustering to obtain multiple geological feature abrupt change depth points. Using the aforementioned multiple geological feature abrupt change depth points as boundaries, the entire exploration depth range is divided into multiple depth segments; Multi-scale fusion feature vectors are extracted from each depth segment, and these feature vectors are arranged in spatial order according to their depth segments to form a three-dimensional feature tensor. The extraction of the multi-scale fusion feature vector for each depth segment includes: Within the entire range of a certain depth segment, calculate the mean and variance of the geological feature parameters of all data points within that depth segment to form a feature sub-vector at a certain macroscopic scale. A certain depth segment is subdivided into several equal-length sub-windows. The mean value of the feature parameters in each sub-window is calculated. The sub-windows are arranged according to their depth to obtain a sub-window mean sequence. The slope and fluctuation entropy of the sub-window mean sequence are calculated to form a feature sub-vector at a certain mesoscale. The number and distribution density of local extreme points of a certain geological exploration data subsequence within a certain depth segment are extracted to form a feature subvector at a certain microscale. Here, a certain geological exploration data subsequence is a geological exploration data subsequence consisting of all data points in the current geological exploration data sequence whose exploration depth falls within a certain depth segment. By concatenating a macroscopic feature vector, a mesoscopic feature vector, and a microscopic feature vector at a certain depth segment, a multi-scale fused feature vector at a certain depth segment is obtained. The three-dimensional feature tensor is matched with common stratigraphic and lithological combination patterns in a pre-established geological knowledge graph. The degree of matching is calculated, and a pattern matching confidence score is assigned to each depth segment. Using the pattern matching confidence as the weight, the multi-scale fusion feature vectors of each depth segment are weighted and averaged to generate a preliminary comprehensive feature vector. The three-dimensional spatial coordinates of the current exploration point are converted into a spatial location encoding vector, and the spatial location encoding vector is fused with the comprehensive feature vector at the element level to generate a comprehensive geological feature description of the current exploration point.

3. The geological body type identification method based on a multi-scale geological association network according to claim 1, characterized in that, The construction of a multi-scale geological correlation network among exploration points, based on the comprehensive geological feature descriptions and three-dimensional spatial coordinates of each exploration point, includes: Each exploration point is defined as a network node, and the comprehensive geological feature description vector of each exploration point is defined as the node feature of the network node. Calculate the three-dimensional Euclidean distance between any two exploration points based on their three-dimensional spatial coordinates. Based on the comprehensive geological feature description vector of the exploration points, calculate the feature vector similarity between any two exploration points; Multiple spatial scale levels are preset, and a corresponding spatial neighborhood radius and similarity threshold are set for each spatial scale level. The value of the spatial neighborhood radius increases with the increase of the spatial scale level, and the value of the similarity threshold decreases with the increase of the spatial scale level. Establish a connection edge at the current spatial scale level between any two exploration points whose three-dimensional Euclidean distance is no greater than the spatial neighborhood radius corresponding to the current spatial scale level and whose feature vector similarity is greater than the similarity threshold corresponding to the current spatial scale level. Based on the three-dimensional Euclidean distance and feature vector similarity, the connection weight of the connecting edge is determined according to a preset weight calculation rule, wherein the connection weight is negatively correlated with the three-dimensional Euclidean distance and positively correlated with the feature vector similarity. By integrating the connecting edges established at all spatial scale levels, a multi-scale geological correlation network containing multi-scale connections is formed.

4. The geological body type identification method based on a multi-scale geological association network according to claim 3, characterized in that, The process involves determining the geological response model values ​​for each exploration point based on the multi-scale geological correlation network, and then dividing the exploration points according to these geological response model values ​​to obtain at least one set of exploration points, including: Obtain all neighbor nodes of a given exploration point at different spatial scale levels; For each spatial scale level, the connection weights of a certain exploration point and all its neighboring nodes at the current level are summed to obtain the intra-level connection strength of the exploration point. The multi-scale feature centrality of a certain exploration point is obtained by weighting and summing the intra-level connectivity strength of the exploration point across all spatial scales according to the preset importance coefficients of each level. The multi-scale feature centrality of a certain exploration point is normalized to obtain a certain geological response model value for that exploration point. Exploration points whose geological response model values ​​are within the same range are grouped into the same set of exploration points. For each set of exploration points obtained by division, a spatial connectivity test is performed to ensure that the exploration points in the set are spatially continuous, thus obtaining at least one final set of exploration points.

5. The geological body type identification method based on a multi-scale geological association network according to claim 1, characterized in that, The step of selecting at least one target exploration point from a set of exploration points based on each three-dimensional spatial coordinate using a preset sequence selection rule includes: Calculate a geometric center of the three-dimensional spatial coordinates of all exploration points in the set of exploration points, and based on the geometric center, calculate the three-dimensional Euclidean distance from each exploration point in the set of exploration points to the geometric center. The top K exploration points with the smallest three-dimensional Euclidean distance to a certain geometric center are identified as candidate spatial centers, and the geological response model values ​​and comprehensive geological feature description vectors of each candidate spatial center are obtained, where K is a preset positive integer; Based on the comprehensive geological feature description vector of all exploration points in the set of exploration points, the principal component analysis method is used to extract the feature direction of the set of exploration points in the comprehensive geological feature space. Calculate the projection value of the comprehensive geological feature description vector of each spatial center candidate point in the feature direction, and then normalize and weight the projection values ​​and the corresponding geological response model values ​​to obtain the comprehensive representative score of each spatial center candidate point. The candidate spatial center with the highest comprehensive representative score was selected as the first target exploration point. Calculate the remaining Euclidean distances between the comprehensive geological feature description vectors of the remaining spatial candidate points and the comprehensive geological feature description vectors of the first target exploration point, and fuse the remaining Euclidean distances with the remaining comprehensive representative scores of the remaining spatial candidate points to obtain the supplementary selection scores of the remaining spatial candidate points. Select the remaining spatial candidate points from the top s in the supplementary selection score as target exploration points, thus obtaining at least one target exploration point. The at least one target exploration point includes the spatial center candidate point with the highest comprehensive representative score and the remaining spatial candidate points from the top s in the supplementary selection score.

6. The geological body type identification method based on a multi-scale geological association network according to claim 1, characterized in that, The method of spatially propagating the geological body type identification results of the final target based on the geological response pattern similarity between other exploration points and the at least one target exploration point, using a multi-scale geological association network, yields other geological body type identification results based on the multi-scale geological association network corresponding to other exploration points, including: Calculate the absolute difference between the first geological response model value of the first exploration point in the set of exploration points and the geological response model value of each of the at least one target exploration point to obtain at least one first model difference degree, wherein the first exploration point is any one of the other exploration points; Divide at least one first mode difference by the difference between the maximum and minimum values ​​of the geological response mode values ​​of all exploration points in the set of exploration points, and normalize it to obtain the corresponding normalized difference. Calculate the reciprocal of each normalized dissimilarity to obtain at least one initial similarity; The three-dimensional Euclidean distance between the first exploration point and the at least one target exploration point is obtained, and the at least one initial similarity is corrected based on each three-dimensional Euclidean distance to obtain the geological response pattern similarity between the first exploration point and the at least one target exploration point. From the at least one target exploration point, select M target exploration points that have the highest similarity to the final geological response pattern of the first exploration point as reference target exploration points; Based on the geological body type corresponding to the reference target exploration point in the geological body type identification result of the final target based on the multi-scale geological association network, and the similarity of the geological response patterns between the first exploration point and each reference target exploration point, a certain propagation probability of the first exploration point belonging to a certain geological body type is calculated. The geological body type identification result based on a multi-scale geological association network is used as the geological body type corresponding to the highest propagation probability as the first exploration point.

7. A geological body type identification system based on a multi-scale geological association network, characterized in that, include: The acquisition module is configured to acquire geological exploration data at at least one exploration point within a preset exploration depth range, and sort the geological exploration data at the same exploration point to obtain at least one geological exploration data sequence. The generation module is configured to perform multi-dimensional geological feature fusion analysis based on the at least one geological exploration data sequence to generate a comprehensive geological feature description for each exploration point. The module is configured to construct a multi-scale geological association network between exploration points based on the comprehensive geological feature descriptions and three-dimensional spatial coordinates of each exploration point. The partitioning module is configured to determine the geological response pattern of each exploration point based on the multi-scale geological correlation network, and partition the exploration points according to the geological response pattern to obtain at least one set of exploration points. The selection module is configured to obtain the three-dimensional spatial coordinates of each exploration point in a certain set of exploration points, and select at least one target exploration point in the certain set of exploration points according to the three-dimensional spatial coordinates and a preset sequence selection rule. The identification module is configured to input the geological exploration data sequence corresponding to the at least one target exploration point into a preset geological body type identification model based on a multi-scale geological association network, so as to obtain the preliminary geological body type identification result based on the multi-scale geological association network corresponding to the at least one target exploration point. The fusion module is configured to generate the final target geological body type identification result based on the multi-scale geological association network, using a multi-source decision fusion mechanism, based on the preliminary geological body type identification results from various preliminary multi-scale geological association networks. The process of generating the geological body type identification result includes: Obtain preliminary geological body type identification results based on a multi-scale geological association network corresponding to each target exploration point. A preliminary geological body type identification result based on a multi-scale geological association network includes the probability distribution of a target exploration point belonging to various predefined geological body types. Obtain the weight information corresponding to each target exploration point, wherein the weight information includes the comprehensive representative score of the first target exploration point and the supplementary selection score of the other target exploration points; The weight information of all target exploration points is normalized to obtain the normalized weight of each target exploration point. The probability of a certain geological body type in the preliminary geological body type identification results based on a multi-scale geological association network for a certain target exploration point is multiplied by the normalized weight of the target exploration point to obtain the weighted probability of a certain geological body type. The total weighted probability of a certain geological body type is obtained by summing the weighted probabilities of all target exploration points corresponding to a certain geological body type. Compare the total weighted probabilities of all predefined geological body types, and determine the geological body type with the highest total weighted probability as the geological body type identification result based on the multi-scale geological association network as the final target. The output module is configured to spatially propagate the geological body type identification results of the final target based on the geological response pattern similarity between other exploration points and the at least one target exploration point, thereby obtaining other geological body type identification results based on the multi-scale geological association network corresponding to other exploration points.

8. An electronic device, characterized in that, include: At least one processor, and a memory communicatively connected to the at least one processor, wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 6.

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

Images