AI (artificial intelligence)-based geological GIS (geographic information system) model quick retrieval method and system

Abstract

The invention discloses an AI-based geological GIS model rapid retrieval method and system, and relates to the technical field of geological information retrieval, and the method comprises the following steps: obtaining a geological GIS model library, extracting geological attribute characteristic parameters and unique identifiers in the geological GIS model library, forming a mapping relation, forming a relation mapping table, and obtaining geological data to be recognized. Geological attribute analysis is carried out on the to-be-recognized geological data to obtain to-be-recognized geological attribute parameters, first retrieval is carried out on the to-be-recognized geological attribute parameters based on the relation mapping table to obtain a candidate set, second retrieval is carried out on the candidate set to obtain a comprehensive matching degree, and the geological GIS models in the candidate set are subjected to priority ranking according to the comprehensive matching degree to obtain the geological GIS models. And outputting the retrieval results sorted according to the priorities. The problems that a traditional method is low in efficiency and insufficient in precision are effectively solved, and the efficiency and accuracy of geological GIS model retrieval are remarkably improved.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of geological information retrieval, and in particular to an AI-based geological GIS model rapid retrieval method and system. BACKGROUND

[0002] In recent years, with the development of geographic information science, GIS models, as the core tools for spatial analysis and decision support, have continuously increased in number and type, covering multiple fields such as land use, environmental assessment, and urban planning. Traditional model management relies on manual classification and experience-based retrieval, which is inefficient and difficult to meet the needs of complex scenarios. At the same time, the explosion of spatial big data has driven the demand for model reuse, and cross-domain model sharing has become a trend. However, the problems of model structural heterogeneity and parameter complexity have become prominent.

[0003] At present, the Chinese invention with the application number CN202210212659.8 discloses a method for automatically selecting GIS data retrieval entry to optimize retrieval. When a user performs spatial information retrieval, the invention can intelligently select a retrieval mode according to the query conditions, combine the retrieval advantages of Elasticsearch and iServer cloud platform in different aspects, and improve the retrieval efficiency. Since the monitoring module is completely automatic and efficient, the user cannot feel the lag caused by the strategy switching process when using it. In actual use, the cluster can be expanded according to the specific data volume, and the user's autonomy in use can be improved by configuring the entry selection strategy. However, the related technology does not establish a relationship mapping table for structured indexing, associate attributes and identifiers, which is not conducive to the fusion and rapid matching of multi-source data attributes, does not narrow down the model matching range through a hierarchical retrieval mechanism, which is not conducive to simplifying the retrieval process of geological GIS models, and is not conducive to the rapidness of retrieval response. SUMMARY

[0004] The technical problem solved by the application is that related technology does not establish a relationship mapping table for structured indexing, associate attributes and identifiers, which is not conducive to the fusion and rapid matching of multi-source data attributes, does not narrow down the model matching range through a hierarchical retrieval mechanism, which is not conducive to simplifying the retrieval process of geological GIS models, and is not conducive to the rapidness of retrieval response.

[0005] To solve the above technical problems, the application provides the following technical solutions: The AI-based geological GIS model rapid retrieval method comprises the following steps: Step S1, a geological GIS model library is obtained, geological attribute feature parameters and unique identifiers in the geological GIS model library are extracted, a mapping relationship between the geological attribute feature parameters and the unique identifiers is constructed, and a relationship mapping table is formed; In step S2, obtain to-be-identified geological data, perform geological attribute analysis on the to-be-identified geological data, and obtain to-be-identified geological attribute parameters; In step S3, based on the relationship mapping table, perform first retrieval on the to-be-identified geological attribute parameters to obtain a candidate set; In step S4, perform second retrieval on the candidate set to obtain a comprehensive matching degree, and perform priority ranking on the geological GIS models in the candidate set according to the comprehensive matching degree, and output a retrieval result.

[0006] As a preferred scheme of the AI-based geological GIS model rapid retrieval method, the relationship mapping table is constructed by: Obtain a geological GIS model library, extract attribute features of each geological GIS model in the geological GIS model library, and the attribute features include geological lithology, geological age, geological structure, mineral resource type, geochemical element content, and terrain slope; Divide each attribute feature into an attribute item, each attribute item corresponds to a specific feature parameter, and the specific feature parameter is a geological attribute feature parameter; Assign a unique identifier to each geological GIS model, and establish an association mapping relationship between the unique identifier and all geological attribute feature parameters of the corresponding geological GIS model; Store the unique identifier, the geological attribute feature parameters, and the weight value corresponding to each geological attribute feature parameter as a relationship mapping table.

[0007] As a preferred scheme of the AI-based geological GIS model rapid retrieval method, the geological attribute analysis on the to-be-identified geological data specifically includes: Obtain to-be-identified geological data, determine the data type of the to-be-identified geological data, the data type includes at least one of a geological text report, a remote sensing image, geophysical data, and geochemical data, and perform geological attribute analysis according to the data type, and the geological attribute analysis includes: If the to-be-identified geological data is a geological text report, identify and extract first geological attribute information, and the first geological attribute information includes lithology information, stratigraphic age information, geological structure information, and mineral resource information; If the to-be-identified geological data is geophysical data or geochemical data, identify second geological attribute information, and the second geological attribute information includes statistical feature information, abnormal value information, and change trend feature information; If the to-be-identified geological data is a remote sensing image, identify third geological attribute information, and the third geological attribute information includes lithology distribution area information, geological structure trace information, and terrain slope information; If the to-be-identified geological data includes at least two of geological text reports, remote sensing image maps, geophysical data and geochemical data, the corresponding geological attribute information of the to-be-identified geological data is associated to the same geological GIS model, different expressions of the corresponding geological attribute information are compared, when a conflict occurs, the corresponding geological attribute information that is complementary and not in conflict is replaced according to the reliability level, and the inferred geological attribute information is obtained by merging; The reliability level includes a first level, a second level and a third level; The first level is the highest reliability of the first geological attribute information; The second level is the second highest reliability of the second geological attribute information; The third level is the lowest reliability of the third geological attribute information; The first geological attribute information, the second geological attribute information, the third geological attribute information and the inferred geological attribute information are uniformly formatted to obtain the to-be-identified geological attribute parameter.

[0008] As a preferred scheme of the AI-based geological GIS model rapid retrieval method, the first retrieval specifically includes: The to-be-identified geological attribute parameter and the corresponding geological attribute feature parameter of the geological GIS model are compared one by one, the parameter type of the to-be-identified geological attribute parameter and the corresponding geological attribute feature parameter of the geological GIS model, and the parameter type includes discrete parameters and continuous parameters; If the parameter type is a discrete parameter, the to-be-identified geological attribute parameter and the corresponding geological attribute feature parameter of the geological GIS model are matched according to a preset geological knowledge graph, and the matching result includes: If the to-be-identified geological attribute parameter and the corresponding geological attribute feature parameter of the geological GIS model are the same, the matching degree is 1.0; If the to-be-identified geological attribute parameter and the corresponding geological attribute feature parameter of the geological GIS model are determined to be synonyms, the matching degree is matching degree K (0<K<1); If the to-be-identified geological attribute parameter and the corresponding geological attribute feature parameter of the geological GIS model are different, the matching degree is 0; If the parameter type is a continuous parameter, the absolute value difference between the to-be-identified geological attribute parameter and the corresponding geological attribute feature parameter of the geological GIS model is calculated , according to a preset ideal tolerance threshold and a maximum tolerance threshold , the matching degree is calculated, including: When , the matching degree=1.0; When , the matching degree ; When , the matching degree = 0; According to the weight value predefined by each geological attribute characteristic parameter, the matching degrees are weighted and linearly summed to calculate the total matching score of the geological GIS model and the to-be-identified geological data, and the calculation expression is: Total matching score ; Wherein, i 匹配 is the matching degree of the i-th attribute, i 权重 is the weight value of the i-th attribute, and the sum of all weight values is 1; Traverse the geological GIS model in the relationship mapping table, calculate the total matching score of each geological GIS model, and input all geological GIS models with a score higher than a preset score threshold into the candidate set.

[0009] As a preferred scheme of the AI-based geological GIS model fast retrieval method, the second retrieval specifically includes: Extract the spatial feature parameters of the to-be-identified geological data to generate a to-be-retrieved spatial feature vector, wherein the spatial feature parameters include position parameters, shape parameters and azimuth angle parameters; Locate the geological GIS model with a similar spatial position to the to-be-identified geological data by using a spatial indexing technology, and obtain a preliminary spatial matching subset from the candidate set; Obtain the spatial feature vector of each geological GIS model in the preliminary spatial matching subset, calculate the similarity between the to-be-retrieved spatial feature vector and the spatial feature vector of each geological GIS model in the preliminary spatial matching subset, and the similarity includes position similarity, shape similarity and azimuth similarity; The position similarity is calculated based on the overlap rate of the coordinate range and the deviation of the barycentric coordinates; The shape similarity is calculated based on the vector distance of the contour feature points after feature transformation; The azimuth similarity is calculated based on the included angle of the spatial orientation vector; According to a preset weight value, the position similarity, the shape similarity and the azimuth similarity scores are weighted and summed to obtain a comprehensive matching degree; According to the comprehensive matching degree, the geological GIS models in the preliminary spatial matching subset are prioritized, and the top N models are output as the final retrieval result, wherein N is a preset positive integer.

[0010] As a preferred scheme of the AI-based geological GIS model fast retrieval method, the extraction of the spatial feature parameters of the to-be-identified geological data includes: According to the geometric type of the geological data to be identified, the spatial feature parameters thereof are extracted, and the geometric type includes a point set type, a linear type and a planar type; If the geometric type is the planar type, the coordinate range of the minimum outer rectangle of the planar type is extracted as a position parameter, the area, the perimeter, the compactness and the contour polygon sequence of the planar type are extracted as a shape parameter, and the principal axis direction or the long side direction of the planar type is extracted as an azimuth angle parameter; If the geometric type is the linear type, the coordinate range of the overall envelope line of the linear type is extracted as a position parameter, the length and the tortuosity of the linear type are extracted as a shape parameter, and the direction of the line connecting the starting point and the ending point of the linear type or the overall trend direction is extracted as an azimuth angle parameter; If the geometric type is the point set type, the coordinate range of the distribution range of the point set of the point set type is extracted as a position parameter, the point density and the ratio of the major axis and the minor axis of the distribution ellipse of the point set type are extracted as a shape parameter, and the principal axis direction of the distribution ellipse of the point set type is extracted as an azimuth angle parameter; The position parameter, the shape parameter and the azimuth angle parameter corresponding to the geometric type are combined in a predefined order and structure according to the geometric type, and a spatial feature vector to be searched is generated.

[0011] As a preferred scheme of the AI-based geological GIS model rapid retrieval method, the comprehensive matching degree calculation method is: Comprehensive matching degree = position similarity score × W 位置 + shape similarity score × W 形态 + azimuth similarity score × W 方位 ; Wherein, W 位置 is a position weight value, W 形态 is a shape weight value, and W 方位 is an azimuth weight value, and W 位置 +W 形态 +W 方位 =1. The weight value is set as a dynamic weight, and the basic range thereof includes: W 位置 ∈[A, B], W 形态 ∈[C, D], and W 方位 ∈[E, F], wherein A, B, C, D, E and F are pre-set weight boundary values, and the weight value is adjusted according to the type of the geological data to be identified, including: If the geological data to be identified is tectonic geological data, the position weight value is increased. If the to-be-identified geological data is sedimentary basin data, the shape weight value is increased.

[0012] As a preferred scheme of the AI-based geological GIS model rapid retrieval method, the value range of N is 1-10, and when the difference between the highest value and the second highest value of the comprehensive matching degree is less than a preset threshold, the value of N is automatically increased.

[0013] As a preferred scheme of the AI-based geological GIS model rapid retrieval method, the relationship mapping table further includes: Receiving user feedback data on the retrieval result, and adjusting the weight value corresponding to each geological attribute feature parameter in the relationship mapping table based on the feedback data.

[0014] The AI-based geological GIS model rapid retrieval system includes an analysis module and a retrieval module. The analysis module is used to construct a relationship mapping table, which is used to store the mapping relationship between the geological attribute feature parameters of the geological GIS model and the unique identifier of the geological GIS model, obtain to-be-identified geological data, and perform geological attribute analysis on the to-be-identified geological data to obtain to-be-identified geological attribute parameters. The retrieval module is used to perform first retrieval on the to-be-identified geological attribute parameters by using the relationship mapping table to obtain a candidate set, perform second retrieval on the candidate set, obtain a comprehensive matching degree, sort the geological GIS models in the candidate set according to the comprehensive matching degree, and output a retrieval result.

[0015] The present application has the following advantages: The two-stage retrieval shortens the query time of the massive model library, the first stage establishes a relationship mapping table, constitutes an efficient association between the model unique identifier and the multi-dimensional features, and fuses multiple types of data attributes for attribute matching to quickly filter out most irrelevant models. The dynamic weight mechanism in the spatial similarity calculation can highlight key features, the adaptive output of the retrieval result provides more options when the results are similar, avoids missing potential targets, and effectively solves the problems of low efficiency and insufficient precision of the traditional method. BRIEF DESCRIPTION OF DRAWINGS

[0016] Figure 1 The basic flowchart of the AI-based geological GIS model rapid retrieval method provided by an embodiment of the present application is shown.

[0017] Figure 2 The signal transmission flowchart of the AI-based geological GIS model rapid retrieval method provided by an embodiment of the present application is shown.

[0018] Figure 3A first retrieval process schematic diagram of an AI-based geological GIS model rapid retrieval method is provided for an embodiment of the present application. DETAILED DESCRIPTION

[0019] To make the above objectives, characteristics and advantages of the present application more obvious and easy to understand, the specific embodiments of the present application will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present application, not all embodiments.

[0020] Embodiments, with reference to Figures 1-3 For an embodiment of the present application, an AI-based geological GIS model rapid retrieval method is provided, including the following steps: Step S1, obtain a geological GIS model library, extract geological attribute feature parameters and unique identification in the geological GIS model library, construct a mapping relationship of the geological attribute feature parameters and the unique identification, and form a relationship mapping table; Step S2, obtain the to-be-identified geological data, perform geological attribute analysis on the to-be-identified geological data, and obtain to-be-identified geological attribute parameters; Step S3, based on the relationship mapping table, performing first retrieval on the to-be-identified geological attribute parameters to obtain a candidate set; Step S4, performing second retrieval on the candidate set to obtain a comprehensive matching degree, and performing priority sorting on the geological GIS models in the candidate set according to the comprehensive matching degree, and outputting a retrieval result.

[0021] The relationship mapping table includes: Obtain a geological GIS model library, extract the attribute features of each geological GIS model in the geological GIS model library, and the attribute features include geological lithology, geological age, geological structure, mineral resource type, geochemical element content and topographic slope; Divide each attribute feature into an attribute item, each attribute item corresponds to a specific feature parameter, and the specific feature parameter is a geological attribute feature parameter; Assign a unique identification to each geological GIS model, and establish an associated mapping relationship of the unique identification and all geological attribute feature parameters of the corresponding geological GIS model; Store the unique identification, the geological attribute feature parameters and the weight value corresponding to each geological attribute feature parameter as the relationship mapping table.

[0022] In a specific embodiment, constructing the relationship mapping table is to create an efficient and structured indexing mechanism to convert the massive attribute information of the geological GIS model into a data structure that can be quickly queried and matched by the AI. The process first extracts the core attribute features of each geological GIS model and divides each attribute into specific attribute items. Then, the system assigns a unique identifier to each model and establishes a direct mapping from the unique identifier to the model's entire set of attribute feature parameters, allowing quick access to all attributes of the model through the unique identifier. The system also constructs an inverted index for each attribute value, which is an index structure with attribute values as keys and a list of model unique identifiers containing the attribute as values. For example, in the lithology field, the key "sandstone" is associated with a list of model unique identifiers containing all lithology as "sandstone".

[0023] The geological attribute analysis of the to-be-identified geological data specifically includes: Obtaining the to-be-identified geological data, determining the data type of the to-be-identified geological data, the data type including at least one of geological text report, remote sensing image, geophysical data and geochemical data, performing geological attribute analysis according to the data type, the geological attribute analysis including: If the to-be-identified geological data is a geological text report, identifying and extracting first geological attribute information, the first geological attribute information including lithology information, stratigraphic age information, geological structure information and mineral resources information; If the to-be-identified geological data is geophysical data or geochemical data, identifying second geological attribute information, the second geological attribute information including statistical characteristic information, abnormal value information and variation trend characteristic information; If the to-be-identified geological data is a remote sensing image, identifying third geological attribute information, the third geological attribute information including lithology distribution area information, geological structure trace information and terrain slope information; If the to-be-identified geological data includes at least two of geological text report, remote sensing image, geophysical data and geochemical data, associating the corresponding geological attribute information of the to-be-identified geological data to the same geological GIS model, comparing different expressions of the corresponding geological attribute information, replacing according to reliability level when there is a conflict, merging the corresponding geological attribute information that is complementary and non-conflicting to obtain inferred geological attribute information; In specific embodiments, a recognition model is established, different types of input data are processed by multiple parallel encoders, including encoding geological text reports into text feature vectors using a pre-trained language model, processing geophysical data and geochemical data by combining one-dimensional convolutional neural networks with recurrent neural networks to generate geophysical and geochemical feature vectors, and encoding remote sensing images into image feature vectors by convolutional neural networks. All these feature vectors are spliced and integrated into a unified joint feature representation by a fully connected layer. The joint feature is then input into a multi-task decoder, which simultaneously predicts all target geological attribute information through multiple parallel output layers. The recognition model is trained in an end-to-end manner, and the total loss function is the weighted sum of the loss of each attribute task. All parameters are optimized simultaneously through backpropagation, so that the recognition model can automatically learn the optimal feature representation from any available input data to cooperatively complete all geological attribute recognition tasks. This recognition model realizes the unified processing of multi-source data and the parallel output of attributes by a single architecture, avoiding complex multi-model fusion steps.

[0024] The reliability level includes a first level, a second level, and a third level; The first level has the highest reliability of the first geological attribute information; The second level has the second highest reliability of the second geological attribute information; The third level has the lowest reliability of the third geological attribute information; The first geological attribute information, the second geological attribute information, the third geological attribute information, and the inferred geological attribute information are formatted and combined to obtain the to-be-recognized geological attribute parameters.

[0025] In specific embodiments, the geological attribute analysis on the to-be-identified geological data is to convert the multi-source and heterogeneous original geological data into structured geological attribute parameters, so as to provide comparable and unified feature indicators for subsequent model retrieval. For example, if a geological text report describing a certain region is input, the system can identify and extract first geological attribute information such as sandstone, anticline structure and copper mineralization point by using natural language processing technology; if a remote sensing image of the region is input, the system can identify the distribution range of the granite body and other second geological attribute information by using computer vision algorithm. For geophysical data (such as gravity anomaly data) or geochemical data (such as soil element content), the system can extract the average value and variance and other third geological attribute information by using data analysis. When multiple types of to-be-identified geological data are received, information fusion can be performed. For example, when the system simultaneously obtains a geological text report (indicating "lithology: sandstone") and geochemical data (indicating "lithology: granite") of a certain region, the system first correlates the lithology information extracted from the two types of data to the same target geological GIS model; after comparing and finding that the lithology descriptions conflict with each other (sandstone ≠ granite), the system automatically adopts the "sandstone" conclusion of the text report according to the pre-defined reliability level (the text report is the first level with the highest reliability, and the geochemical data is the second level with the second highest reliability), while retaining the complementary information (such as "gold element anomaly") in the geochemical data that has no conflict; finally, the system fuses the information to form inferred geological attribute information "lithology: sandstone, gold element anomaly", and outputs the information as the to-be-identified geological attribute parameter after standardization.

[0026] The first retrieval specifically includes: The to-be-identified geological attribute parameter and the corresponding geological attribute feature parameter of the geological GIS model are compared one by one, and the parameter types of the to-be-identified geological attribute parameter and the corresponding geological attribute feature parameter of the geological GIS model include discrete parameters and continuous parameters; If the parameter type is a discrete parameter, the to-be-identified geological attribute parameter and the corresponding geological attribute feature parameter of the geological GIS model are matched according to the pre-defined geological knowledge graph, and the matching result includes: If the to-be-identified geological attribute parameter and the corresponding geological attribute feature parameter of the geological GIS model are the same, the matching degree is 1.0; If the to-be-identified geological attribute parameter and the corresponding geological attribute feature parameter of the geological GIS model are determined to be synonyms, the matching degree is matching degree K (0 < K < 1); If the to-be-identified geological attribute parameter and the corresponding geological attribute feature parameter of the geological GIS model are different, the matching degree is 0; If the parameter type is a continuous parameter, the absolute value difference between the to-be-identified geological attribute parameter and the corresponding geological attribute feature parameter of the geological GIS model is calculated , and the ideal tolerance threshold and a maximum tolerance threshold , calculating the matching degree, comprising: When , the matching degree = 1.0; When , the matching degree ; When , the matching degree = 0; According to the weight value predefined by each geological attribute characteristic parameter, the matching degrees are weighted linearly summed to calculate the total matching score of the geological GIS model and the to-be-identified geological data, and the calculation expression is: The total matching score ; Wherein, i 匹配 is the matching degree of the i-th attribute, i 权重 is the weight value of the i-th attribute, and the sum of all weight values is 1; Traverse the geological GIS model in the relationship mapping table, calculate the total matching score of each geological GIS model, and input all the geological GIS models with a score higher than the preset score threshold into the candidate set.

[0027] In specific embodiments, the first retrieval quickly filters out a candidate model set highly similar to the data to be identified in terms of geological attribute characteristics from a large geological GIS model library through quantitative comparison. The core is to convert the abstract geological attribute similarity into a calculable and sortable numerical score, laying the foundation for the second retrieval. Specifically, the system compares the attribute parameters extracted from the data to be identified with the corresponding feature parameters of each geological GIS model in the geological GIS model library one by one, and first distinguishes the parameter types: for discrete parameters (such as lithology, structure type), the system relies on the pre-set geological knowledge graph for semantic matching. The pre-set geological knowledge graph is composed by integrating the terminology system in the geological discipline dictionary and industry standards. For example, if the lithology of the data to be identified is sandstone and the model parameter is sandy gravel, the knowledge graph determines that they are synonymous, and the matching degree is K value (such as 0.8). For continuous parameters (such as element content, slope), quantitative calculation is performed, for example, the average slope of the data to be identified is 20°, and the model parameter is 25°, Tideal=5°, Tmax=15°, then the absolute value difference δ=5°, because δ=Tideal, the matching degree=1.0; if the model parameter is 32°, δ=12°, between Tideal and Tmax, the matching degree=1-(12-5) / (15-5)=0.3. Subsequently, the system sums the weighted matching degrees according to the pre-defined weights (such as lithology weight 0.4, age weight 0.3, slope weight 0.3), for example, lithology matching degree 0.8×0.4+age matching degree 1.0×0.3+slope matching degree 0.3×0.3=0.71, to get the total matching score of the model. Finally, after traversing all the models, the models with total scores higher than the set threshold (such as 0.6) are included in the candidate set, completing efficient and accurate preliminary screening.

[0028] The second retrieval specifically includes: extracting the spatial feature parameters of the geological data to be identified, generating a spatial feature vector to be retrieved, the spatial feature parameters including position parameters, shape parameters and azimuth angle parameters; locating the geological GIS models with similar spatial positions to the geological data to be identified by using spatial indexing technology, and obtaining a preliminary spatial matching subset from the candidate set; obtaining the spatial feature vector of each geological GIS model in the preliminary spatial matching subset, and calculating the similarity between the spatial feature vector to be retrieved and the spatial feature vector of each geological GIS model in the preliminary spatial matching subset, the similarity including position similarity, shape similarity and azimuth similarity; the position similarity is calculated based on the overlap rate of the coordinate range and the deviation of the center of gravity coordinates; the shape similarity is calculated based on the vector distance of the contour feature points after feature transformation; the azimuth similarity is calculated based on the included angle of the spatial orientation vector; According to the preset weight value, the position similarity, the shape similarity and the orientation similarity scores are weighted and summed to obtain a comprehensive matching degree; According to the comprehensive matching degree, the geological GIS models in the preliminary spatial matching subset are prioritized, and the first N models are output as the final search results, where N is a preset positive integer.

[0029] In specific embodiments, the second search performs spatial similarity screening on the attribute candidate set generated by the first search, quantitatively evaluates the matching degree of the to-be-identified target and the candidate geological GIS model in spatial position, geometric shape and orientation, and finally outputs the first N models with the highest comprehensive spatial matching degree as the optimal search results, thereby ensuring that the searched models are not only similar in attribute characteristics but also highly consistent in spatial distribution characteristics, and improving the spatial relevance and practicality of the search results.

[0030] The position similarity is obtained by calculating the overlap rate of the intersection area and the union area of the outer rectangles of the two geometric bodies, and normalizing the Euclidean distance between the centers of the two geometric bodies. Finally, the position similarity score is obtained by comprehensively considering the two indicators. For example, if the overlap rate of the to-be-identified target and the candidate model is 70%, and the center deviation is within the tolerance range, the position similarity score is high.

[0031] The shape similarity converts the contour feature points into feature vectors, and evaluates the shape similarity by calculating the cosine similarity or Euclidean distance of the two vectors. For example, if both of the two are elliptical sedimentary basins, even if the sizes are different, the distance of the vectors after feature transformation is small, and the shape similarity score is high.

[0032] The orientation similarity generates a direction vector by extracting main orientation features (such as structural strike and topographic trend), and calculates the cosine value of the included angle between the two vectors as the similarity score. For example, if the strike vectors of the two linear structures have a small included angle, the cosine value is close to 1, and the orientation similarity score is high.

[0033] The spatial feature parameters of the to-be-identified geological data include: According to the geometric type of the to-be-identified geological data, the spatial feature parameters are extracted, and the geometric type includes point set type, linear type and planar type; If the geometric type is the planar type, the coordinate range of the minimum outer rectangle of the planar type is extracted as the position parameter, the area, perimeter, compactness and contour polygon sequence of the planar type are extracted as the shape parameter, and the principal axis direction or long side direction of the planar type is extracted as the azimuth angle parameter; If the geometric type is the line type, the coordinate range of the overall envelope of the line type is extracted as the position parameter, the length and tortuosity of the line type are extracted as the shape parameter, and the direction of the line connecting the start point and the end point of the line type or the overall trend direction is extracted as the azimuth angle parameter; If the geometric type is the point set type, the coordinate range of the point set distribution range of the point set type is extracted as the position parameter, the point density and the ratio of the major and minor axes of the distribution ellipse of the point set type are extracted as the shape parameter, and the major axis direction of the distribution ellipse of the point set type is extracted as the azimuth angle parameter; The position parameter, the shape parameter, and the azimuth angle parameter corresponding to the geometric type are combined in a predefined order and structure to generate a spatial feature vector to be searched.

[0034] In specific embodiments, extracting the spatial feature parameters of the geological data to be identified converts the spatial geometric characteristics of the geological object into standardized and quantifiable feature vectors, providing a unified basis for subsequent spatial similarity calculation. The above parameters are normalized in a predefined order, which is position parameter, shape scalar parameter, azimuth parameter, and shape vector parameter, to obtain the spatial feature vector to be searched. For example, a fixed-dimensional numerical vector [longitude min, latitude min, longitude max, latitude max, 50, 30, 0.75,..., 60] is generated, which fully represents the spatial properties of the geological object and lays a foundation for subsequent spatial retrieval based on vector similarity.

[0035] The comprehensive matching degree calculation method is: Comprehensive matching degree = position similarity score × W 位置 + shape similarity score × W 形态 + azimuth similarity score × W 方位 ; Wherein, W 位置 is the position weight value, W 形态 is the shape weight value, and W 方位 is the azimuth weight value, and W 位置 +W 形态 +W 方位 =1; The weight value is set as a dynamic weight, and its basic range includes: W 位置 ∈[A, B], W 形态 ∈[C, D], W 方位 ∈[E, F], where A, B, C, D, E, F are pre-set weight boundary values, and the weight value is adjusted according to the type of the geological data to be identified, including: If the geological data to be identified is tectonic geological data, the position weight value is increased; If the geological data to be identified is sedimentary basin data, the shape weight value is increased.

[0036] In specific embodiments, A, B, C, D, E, and F are pre-set weight boundary values, and preferably, A = 0.3, B = 0.5, C = 0.2, D = 0.4, E = 0.1, and F = 0.3. When processing tectonic geological data, W 位置 tends to be close to 0.5; and when processing sedimentary basin data, W 形态 tends to be close to 0.4.

[0037] The value of N ranges from 1 to 10, and when the difference between the highest value and the second highest value of the comprehensive matching degree is less than a pre-set threshold, the value of N is automatically increased.

[0038] In specific embodiments, the calculation of the comprehensive matching degree is through a dynamic weight strategy and adaptive result quantity adjustment, which ensures that the search results can reflect the key feature differences of different types of geological data and can flexibly handle ambiguous scenarios, and finally realizes the continuous optimization of system performance through user feedback. The system calculates the comprehensive matching degree of all candidate models, for example, a model position score of 0.9 x 0.5 + shape score of 0.6 x 0.3 + orientation score of 0.8 x 0.2 = 0.45 + 0.18 + 0.16 = 0.79, and sorts them. The system will intelligently judge the matching degree of the results. If the difference between the highest score and the second highest score is less than a pre-set threshold, the pre-set threshold is a configurable similarity difference threshold, which is used to measure the discrimination of the matching results, and is usually set to 0.05 to 0.1. It indicates that the discrimination of the results is low and there are multiple similar options. At this time, the system automatically increases the value of N to provide more reference schemes for the user and avoid missing potential matching models.

[0039] The relationship mapping table further includes: receiving feedback data of the user on the search results, and adjusting the weight value corresponding to each geological attribute feature parameter in the relationship mapping table based on the feedback data.

[0040] The application establishes the correlation between the geological attribute characteristic parameters and the model unique identifier by constructing a relationship mapping table, and gives each characteristic parameter a weight value; the to-be-identified geological data is subjected to multi-source data fusion analysis according to its data type to obtain the to-be-identified geological attribute parameters; in the first retrieval stage, the to-be-identified geological attribute parameters are classified, the matching degree K is calculated for the discrete parameters by judging the synonymous relationship through the knowledge graph, the matching degree is calculated for the continuous parameters by using the tolerance threshold method, and then the total matching score is obtained by weighted summation to generate a candidate set; in the second retrieval stage, the spatial characteristic parameters of the to-be-identified geological data are extracted to generate a characteristic vector, the similarity is calculated, and finally the comprehensive matching degree is obtained by weighting; the difference threshold of the matching degree is set to realize the result quantity adjustment, and the attribute characteristic weight is continuously optimized through the user feedback mechanism, so that the system has a continuous learning ability. The application effectively solves the problems of low efficiency, insufficient precision and lack of semantic understanding of the traditional retrieval method, and significantly improves the accuracy and practicability of the geological GIS model retrieval.

[0041] Those skilled in the art will appreciate that embodiments of the application can be provided as methods, systems or computer program products. Accordingly, the application can be embodied in the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the application can be embodied in the form of a computer program product embodied on one or more computer-usable storage media having computer-usable program code embodied thereon. The storage media can be any type of volatile or non-volatile storage device, or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk or optical disk. These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific way, so that the instructions stored in the computer-readable memory produce a manufactured product including instruction devices, which realize the processes specified in the flowcharts Figure 1 one or more flows and / or blocks Figure 1 one or more flows and / or blocks

[0042] It should be noted that the above examples are only used to illustrate the technical solutions of the present application but not limit the present application. Although the present application is described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present application can be modified or equivalently replaced, without departing from the spirit and scope of the technical solutions of the present application, which should be covered in the scope of the claims of the present application.

Claims

1. A method for rapid retrieval of geological GIS models based on AI, characterized in that, Includes the following steps: Step S1: Obtain the geological GIS model library, extract the geological attribute feature parameters and unique identifiers from the geological GIS model library, construct the mapping relationship between the geological attribute feature parameters and unique identifiers, and form a relationship mapping table; Step S2: Obtain the geological data to be identified, perform geological attribute analysis on the geological data to be identified, and obtain the geological attribute parameters to be identified; Step S3: Based on the relation mapping table, perform a first search on the geological attribute parameters to be identified to obtain a candidate set; Step S4: Perform a second search on the candidate set to obtain the comprehensive matching degree, and sort the geological GIS models in the candidate set according to the comprehensive matching degree, and output the search results.

2. The AI-based rapid retrieval method for geological GIS models as described in claim 1, characterized in that, Constructing the relation mapping table includes: Obtain a geological GIS model library and extract the attribute features of each geological GIS model in the library. The attribute features include geological lithology, geological age, geological structure, mineral resource type, geochemical element content, and topographic slope. Each attribute feature is divided into attribute items, and each attribute item corresponds to specific feature parameters, which are geological attribute feature parameters. Assign a unique identifier to each geological GIS model, and establish an association mapping relationship between the unique identifier and all the geological attribute feature parameters of the corresponding geological GIS model; The unique identifier, the geological attribute feature parameters, and the weight value corresponding to each geological attribute feature parameter are stored as a relational mapping table.

3. The AI-based rapid retrieval method for geological GIS models as described in claim 2, characterized in that, The geological attribute analysis of the geological data to be identified specifically includes: Obtain geological data to be identified, determine the data type of the geological data to be identified, the data type includes geological text reports, remote sensing image maps, geophysical data, and geochemical data, and perform geological attribute analysis based on the data type, the geological attribute analysis including: If the geological data to be identified is a geological text report, then the first geological attribute information is identified and extracted. The first geological attribute information includes lithological information, stratigraphic age information, geological structure information, and mineral resource information. If the geological data to be identified is geophysical data or geochemical data, then the second geological attribute information is identified, which includes statistical characteristic information, outlier information and trend characteristic information. If the geological data to be identified is a remote sensing image, then the third geological attribute information is identified, which includes lithological distribution area information, geological structural trace information, and topographic slope information. If the geological data to be identified includes at least two of the following: geological text reports, remote sensing images, geophysical data, and geochemical data, then the geological attribute information corresponding to the geological data to be identified is associated with the same geological GIS model. Different descriptions of the same corresponding geological attribute information are compared. When a conflict occurs, the corresponding geological attribute information is replaced according to the reliability level. The complementary and non-conflicting corresponding geological attribute information is merged to obtain inferred geological attribute information. The reliability levels include Level 1, Level 2, and Level 3; The first level represents the highest reliability of the first geological attribute information; The second level represents the second highest reliability of the second geological attribute information; The third level represents the lowest reliability of the third geological attribute information. The first geological attribute information, the second geological attribute information, the third geological attribute information, and the inferred geological attribute information are formatted in a unified manner and combined to obtain the geological attribute parameters to be identified.

4. The AI-based rapid retrieval method for geological GIS models as described in claim 3, characterized in that, The first search specifically includes: The geological attribute parameters to be identified are compared one by one with the corresponding geological attribute feature parameters of the geological GIS model. The parameter types of the geological attribute parameters to be identified and the corresponding geological attribute feature parameters of the geological GIS model include discrete parameters and continuous parameters. If the parameter type is discrete, the geological attribute parameter to be identified and the corresponding geological attribute feature parameter of the geological GIS model are matched according to a preset geological knowledge map. The matching results include: If the geological attribute parameter to be identified is the same as the corresponding geological attribute feature parameter of the geological GIS model, then the matching degree is 1.

0. If the geological attribute parameter to be identified is determined to be a synonym of the corresponding geological attribute feature parameter of the geological GIS model, then the matching degree of this item is the matching degree K(0). <K<1); If the geological attribute parameter to be identified is different from the corresponding geological attribute feature parameter of the geological GIS model, the matching degree is 0. If the parameter type is a continuous parameter, then calculate the absolute difference between the geological attribute parameter to be identified and the corresponding geological attribute feature parameter of the geological GIS model. Based on the preset ideal tolerance threshold and maximum tolerance threshold Calculate the matching degree, including: when At that time, the matching degree = 1.0; when At that time, matching degree ; when At that time, the matching degree = 0; Based on the predefined weight values ​​of each geological attribute feature parameter, the matching degrees of each item are weighted and linearly summed to calculate the total matching score between the geological GIS model and the geological data to be identified. The calculation expression is as follows: Total Match Score ; in, Let i be the matching degree of the i-th attribute. Let be the weight value of the i-th attribute, and the sum of all weight values ​​is 1; Traverse the geological GIS models in the relation mapping table, calculate the total matching score for each geological GIS model, and input all geological GIS models with scores higher than a preset score threshold into the candidate set.

5. The AI-based rapid retrieval method for geological GIS models as described in claim 4, characterized in that, The second search specifically includes: Spatial feature parameters of the geological data to be identified are extracted to generate a spatial feature vector to be retrieved. The spatial feature parameters include location parameters, morphological parameters and azimuth parameters. Spatial indexing technology is used to locate geological GIS models that are spatially similar to the geological data to be identified, and a preliminary spatial matching subset is obtained from the candidate set; Obtain the spatial feature vector of each geological GIS model in the preliminary spatial matching subset, and calculate the similarity between the spatial feature vector to be retrieved and the spatial feature vector of each geological GIS model in the preliminary spatial matching subset. The similarity includes location similarity, morphological similarity and orientation similarity. The positional similarity is calculated based on the overlap rate of the coordinate range and the centroid coordinate deviation; The morphological similarity is calculated based on the vector distance between contour feature points after feature transformation. The orientation similarity is calculated based on the angle between spatial orientation vectors; Based on preset weight values, the scores of positional similarity, morphological similarity, and orientational similarity are weighted and summed to obtain the comprehensive matching degree. The geological GIS models in the preliminary spatial matching subset are prioritized based on the comprehensive matching degree, and the top N models are output as the final search results, where N is a preset positive integer.

6. The AI-based rapid retrieval method for geological GIS models as described in claim 5, characterized in that, The spatial feature parameters extracted from the geological data to be identified include: Based on the geometric type of the geological data to be identified, its spatial feature parameters are extracted. The geometric type includes point set type, line type and area type. If the geometric type is the planar type, then the coordinate range of the minimum bounding rectangle of the planar type is extracted as the position parameter, the area, perimeter, compactness and outline polygon sequence of the planar type are extracted as the shape parameter, and the principal axis direction or long side direction of the planar type is extracted as the azimuth parameter. If the geometric type is the linear type, then the coordinate range of the overall envelope of the linear type is extracted as the position parameter, the length and tortuosity of the linear type are extracted as the shape parameter, and the direction of the line connecting the start and end points of the linear type or the overall trend direction is extracted as the azimuth parameter. If the geometric type is the point set type, then the coordinate range of the point set distribution range of the point set type is extracted as the position parameter, the point density of the point set type and the ratio of the major and minor axes of the distribution ellipse are extracted as the shape parameter, and the direction of the principal axis of the distribution ellipse of the point set type is extracted as the azimuth parameter. The position parameters, morphological parameters, and azimuth parameters corresponding to the geometric type extracted according to the geometric type are combined in a predefined order and structure to generate a spatial feature vector to be retrieved.

7. The AI-based rapid retrieval method for geological GIS models as described in claim 6, characterized in that, The comprehensive matching degree is calculated as follows: Overall matching degree = positional similarity score × W 位置 + Morphological similarity score × W 形态 +Location similarity score × W 方位 ; Among them, W 位置 For position weight values, W 形态 For shape weight value, W 方位 Here is the azimuth weight value, and W 位置 +W 形态 +W 方位 =1; The weight values ​​are set as dynamic weights, and their basic range includes: W 位置 ∈[A, B],W 形态 ∈[C, D],W 方位 ∈[E, F], where A, B, C, D, E, and F are pre-defined weight boundary values, and the weight values ​​are adjusted according to the type of geological data to be identified, including: If the geological data to be identified is structural geological data, the location weight value is increased; If the geological data to be identified is sedimentary basin data, the morphological weight value is increased.

8. The AI-based rapid retrieval method for geological GIS models as described in claim 7, characterized in that: The value of N ranges from 1 to 10. When the difference between the highest and second-highest comprehensive matching degree is less than a preset threshold, the number of possible values ​​for N is automatically increased.

9. The AI-based rapid retrieval method for geological GIS models as described in claim 8, characterized in that, The relation mapping table also includes: Receive user feedback data on search results, and adjust the weight value corresponding to each geological attribute feature parameter in the relationship mapping table based on the feedback data.

10. An AI-based rapid retrieval system for geological GIS models, characterized in that, Includes an analysis module and a retrieval module; The analysis module is used to construct a relational mapping table, which stores the mapping relationship between the geological attribute feature parameters of the geological GIS model and the unique identifier of the geological GIS model, obtains the geological data to be identified, performs geological attribute analysis on the geological data to be identified, and obtains the geological attribute parameters to be identified. The retrieval module is used to perform a first retrieval on the geological attribute parameters to be identified using the relation mapping table to obtain a candidate set, perform a second retrieval on the candidate set to obtain a comprehensive matching degree, sort the geological GIS models in the candidate set according to the comprehensive matching degree, and output the retrieval results.

Images