Drilling columnar chart template generation method and system based on multi-modal visual large model

Abstract

The application provides a drilling columnar chart template generation method and system based on a multi-modal visual large model. By splitting the base map into a frame and a body, the multi-modal visual large model is used to perform semantic analysis on the body part to extract a drilling feature set, and a standardized code mapping is established based on a feature set code dictionary. Meanwhile, feature coordinate mapping is obtained through pre-analysis of a CAD entity cache, the standardized code is automatically replaced, and redundant graphics are cleaned up, thereby realizing full-process automatic generation of the template. This technical solution significantly reduces the cost of manual repetitive labor, realizes the standardization and unification of the template format, can dynamically adapt to different types of base maps, improves the feature positioning accuracy through entity cache coordinate mapping, and is directly based on CAD file processing, which is compatible with the existing geological engineering software ecology.

Description

Technical Field

[0001] This invention belongs to the field of digital management technology for geological engineering, and relates to a method and system for generating borehole columnar diagram templates. Background Technology

[0002] Borehole columnar sections are core deliverables of geological exploration work. They are used to visually display key information such as geological stratification, lithological characteristics, and physical and mechanical parameters of boreholes, and serve as important basic data for engineering design and construction.

[0003] Currently, the creation of borehole columnar sections mainly relies on manual data filling based on fixed templates, which presents the following problems: 1. High repetitive workload: Borehole columnar sections from different exploration projects have similar structures, but the template framework needs to be repeatedly drawn, consuming a lot of manpower; 2. Inconsistent templates: Different units and projects have different requirements for the format and feature annotation of borehole columnar sections, resulting in template chaos and high collaboration and handover costs; 3. Low feature extraction efficiency: Traditional template creation requires manual identification of features (such as project name, borehole number, etc.) in the base map and positioning of their locations, which is prone to omissions or errors; 4. Poor adaptability: When new types of borehole columnar section base maps appear, the template structure needs to be redesigned, and it cannot dynamically adapt to different format requirements. In existing technologies, although there are CAD-based drawing tools to assist in the creation of borehole columnar sections, they lack the ability to automatically identify, map, and dynamically replace base map features, and cannot achieve intelligent template generation. Summary of the Invention

[0004] To address the following problems mentioned in the background art: Manually filling data into borehole columnar section templates using fixed templates results in repetitive workload, inconsistent templates, low feature extraction efficiency, and poor adaptability. CAD-based methods for creating borehole columnar section templates lack the ability to automatically identify, map, and dynamically replace base map features, thus failing to achieve intelligent template generation. This invention provides a method and system for generating borehole columnar section templates based on a multimodal visual large model. Utilizing multimodal visual large model technology, it achieves automated and standardized generation of borehole columnar section templates, meeting the important needs of the field of digital geological engineering.

[0005] In a first aspect, the present invention provides a method for generating borehole columnar section templates based on a multimodal visual large model, comprising the following steps: Borehole column chart base image structure decomposition: Obtain the borehole column chart base image to be processed, and according to the frame-body decomposition logic, decompose the base image into a frame part containing fixed structural information and a body part containing variable content, and use the feature annotation area in the body part as the core processing object for borehole column chart template generation; Drilling feature set extraction based on multimodal visual large model: The bottom image of the drill bar chart to be processed is converted into image format and encoded into Base64 data. The encoded Base64 data, together with the preset prompt words, is input into the multimodal visual large model. For the core processing object, the multimodal visual large model performs semantic analysis on the image of the bottom image of the drill bar chart to be processed, identifies and outputs the key-value pair set of drill-related features as the drilling feature set. Feature set code mapping establishment: Load a preset feature set code dictionary, which defines the correspondence between each feature item and the standardized code; match each feature item in the borehole feature set with the feature set code dictionary to generate a feature set code mapping, wherein matched feature items are assigned the corresponding standardized code identifier, and unmatched feature items are marked with an empty identifier; Feature set coordinate mapping acquisition: Load the CAD format file of the borehole column chart base map to be processed, pre-parse the text entities in it and extract the text content, insertion point and coordinate range information of each text entity and store them in the entity cache; for each feature value in the obtained borehole feature set, perform text matching search in the entity cache to determine the coordinate position of each feature value in the borehole column chart base map to be processed, and record the spatial positioning information of each feature value in the form of a rectangular range to form the feature set coordinate mapping; Dynamic generation of borehole column chart templates: Load the base map file of the borehole column chart to be processed, locate the position of each feature to be replaced in the base map file of the borehole column chart to be processed based on feature set code mapping and feature set coordinate mapping, delete the original feature text entities and insert the corresponding standardized code in the corresponding position, and clean up redundant graphic entities in the map body area, and output a standardized borehole column chart template file.

[0006] Furthermore, in the structural decomposition of the borehole columnar diagram base map, the fixed structural information of the frame portion includes: title and signature; the variable content of the body portion includes: geological stratification and feature annotation.

[0007] Furthermore, in the borehole feature set extraction based on the multimodal visual large model, the calling environment of the multimodal visual large model is initialized through a configuration file, which at least includes an API access key and a model identifier parameter.

[0008] Furthermore, in the borehole feature set extraction based on the multimodal visual large model, the encoded Base64 data along with preset prompt words are input into the multimodal visual large model. The prompt words require the multimodal visual large model to output feature values ​​item by item according to the predefined feature item list, and to identify and output newly added feature items in the base map that exceed the range of the predefined list.

[0009] Furthermore, in the borehole feature set extraction based on the multimodal visual large model, the borehole feature set includes at least one or more of the following: project name, borehole number, and borehole coordinates; the borehole feature set is compared with a preset borehole feature set dictionary, and new feature items that do not exist in the borehole feature set dictionary are automatically added to the borehole feature set dictionary, thereby realizing the dynamic expansion of the borehole feature set dictionary.

[0010] Furthermore, in the acquisition of feature set coordinate mapping, the CAD format file is in DWG or DXF format. When the input file is in DWG format, it is automatically converted to DXF format before entity parsing. The text-type entities include TEXT entities, MTEXT entities, and INSERT entities. The information stored in the entity cache includes the text content string of each text-type entity, the coordinates of the insertion point, and the rectangular coordinate range defined by the coordinates of the lower left corner and the upper right corner.

[0011] Furthermore, in the acquisition of the feature set coordinate mapping, each feature item in the borehole feature set is divided into key feature items and non-key feature items; for key feature items, a full match method is used for text search, and for non-key feature items, an inclusion match method is used for text search, wherein the key feature items include at least the borehole number.

[0012] Furthermore, in the dynamic generation of the borehole column chart template, the x-axis offset and y-axis offset parameters set in the existing borehole column chart base template file are read, and the insertion position is fine-tuned when inserting standardized code text to compensate for the possible positional deviation between deleting the original feature text entity and inserting the new entity; the output format of the standardized borehole column chart template file is DWG or DXF format, and a timestamp is automatically added to the file name during output to distinguish different versions.

[0013] Secondly, based on the above method, the present invention provides a borehole columnar section template generation system based on a multimodal visual large model, comprising: The base map structure splitting module is used to obtain the base map of the borehole columnar diagram to be processed. According to the decomposition logic of the frame and the body, the base map is split into a frame part containing fixed structural information and a body part containing variable content. The feature annotation area in the body part is used as the core processing object for generating the borehole columnar diagram template. The feature extraction module is used to convert the borehole bar chart background image to be processed into image format and encode it into Base64 data. The encoded Base64 data, along with preset prompt words, is input into the multimodal vision big model. For the core processing object mentioned in the background image structure decomposition module, the multimodal vision big model performs semantic analysis on the image of the borehole bar chart background image to be processed, identifies and outputs a set of key-value pairs of borehole-related features as the borehole feature set. The code mapping module is used to load a preset feature set code dictionary, which defines the correspondence between each feature item and a standardized code; the feature items in the borehole feature set extracted by the feature extraction module are matched with the feature set code dictionary to generate a feature set code mapping, wherein matched feature items are assigned corresponding standardized code identifiers, and unmatched feature items are marked with an empty identifier; The coordinate positioning module is used to load the CAD format file of the borehole column base map to be processed, pre-parse the text entities in it and extract the text content, insertion point and coordinate range information of each text entity and store them in the entity cache; for each feature value in the borehole feature set obtained by the code mapping module, text matching search is performed in the entity cache to determine the coordinate position of each feature value in the borehole column base map to be processed, and the spatial positioning information of each feature value is recorded in the form of a rectangular range to form a feature set coordinate mapping; The template generation module loads the borehole column base map file to be processed. Based on the feature set code mapping in the code mapping module and the feature set coordinate mapping in the coordinate positioning module, it locates the position of each feature to be replaced in the borehole column base map file, deletes the original feature text entities and inserts the corresponding standardized code in the corresponding position, and cleans up redundant graphic entities in the map body area, and outputs a standardized borehole column template file.

[0014] Thirdly, the present invention provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the above-described method for generating borehole columnar diagram templates based on a multimodal visual large model.

[0015] Compared with the prior art, the present invention has the following advantages: (1) Achieve full-process automation: Use multimodal visual large model to perform semantic analysis on the borehole column chart base map and automatically extract borehole feature set. Combine feature set code dictionary matching and CAD entity cache coordinate mapping, the program automatically completes the entire process from base map splitting, feature recognition to template generation, without the need for manual filling and positioning, which significantly reduces repetitive labor costs. (2) High degree of template standardization: The correspondence between feature items and standardized codes is established through the preset feature set code dictionary, and the same type of features in different source base maps are uniformly coded and identified, which solves the problem of template format confusion between different units and different projects, and facilitates subsequent collaboration and data management; (3) Strong dynamic adaptation capability: Based on the semantic understanding capability of the multimodal visual big model, combined with the decomposition logic of the frame and the body, it can handle different types and structures of drill bar chart base maps without redesigning the template structure for new types of base maps, thus improving the adaptability to diverse format requirements. (4) Reliable feature positioning accuracy: By pre-parsed text-type entities in CAD format files and establishing entity cache, the coordinate position of each feature value is determined by text content matching search, and spatial positioning information is recorded in a rectangular range, replacing the traditional manual visual recognition and positioning method, reducing feature omission and position error; (5) Compatible with the existing software ecosystem: Entity analysis, feature replacement and template generation are performed directly based on the CAD format file of the borehole columnar diagram base map. Standardization processing is completed on the basis of existing CAD files without having to leave the existing CAD software environment in the field of geological engineering, which facilitates the connection with the existing workflow.

[0016] In summary, this invention automates the entire template generation process by splitting the base map into a frame and a body, utilizing a multimodal visual large model to perform semantic analysis on the body to extract borehole feature sets, establishing standardized code mappings based on a feature set code dictionary, and obtaining feature coordinate mappings through pre-parsed CAD entity cache. This automatically replaces standardized codes and cleans up redundant graphics. This significantly reduces the cost of repetitive manual labor, achieves standardized template formats, dynamically adapts to different types of base maps, improves feature positioning accuracy through entity cache coordinate mapping, and is directly based on CAD file processing, ensuring compatibility with the existing geological engineering software ecosystem. Attached Figure Description

[0017] Figure 1 This is a flowchart of the method of the present invention.

[0018] Figure 2 This is a system architecture diagram of the present invention.

[0019] Figure 3 This is a schematic diagram of the borehole columnar section underlined in Example 4.

[0020] Figure 4 This is a schematic diagram of the feature set code mapping in Example 4.

[0021] Figure 5 This is a schematic diagram of the feature set coordinate mapping in Example 4.

[0022] Figure 6 This is a schematic diagram of the feature value replacement code in Example 4.

[0023] Figure 7 This is a schematic diagram of the standardized borehole columnar section template file generated in Example 4. Detailed Implementation

[0024] To make the technical problems, technical solutions, and beneficial effects to be solved by this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and are not intended to limit the scope of this application.

[0025] Example 1

[0026] A method for generating borehole columnar section templates based on a multimodal visual large model, the flowchart of which is as follows: Figure 1 As shown, the specific steps are as follows.

[0027] S1. Borehole column chart base image structure decomposition: Obtain the borehole column chart base image to be processed, and according to the frame-body decomposition logic, decompose the base image into a frame part containing fixed structural information and a body part containing variable content, and use the feature annotation area in the body part as the core processing object for borehole column chart template generation.

[0028] Specifically, the fixed structural information in the frame includes: title and signature; the variable content in the body includes: geological stratification and feature annotations.

[0029] S2. Drilling feature set extraction based on multimodal visual large model: The bottom image of the drill bar chart to be processed is converted into image format and encoded as Base64 data. The encoded Base64 data, together with the preset prompt words, is input into the multimodal visual large model. For the core processing object mentioned in S1, the multimodal visual large model performs semantic analysis on the image of the bottom image of the drill bar chart to be processed, identifies and outputs the key-value pair set of drill-related features as the drilling feature set.

[0030] Specifically, the calling environment of the multimodal vision large model is initialized through a configuration file, which at least contains API access keys and model identification parameters.

[0031] More specifically, the encoded Base64 data, along with preset prompts, is input into the multimodal visual big model. The prompts require the multimodal visual big model to output feature values ​​item by item according to a predefined list of feature items, and to identify and output newly added feature items in the base image that exceed the range of the predefined list.

[0032] More specifically, the borehole feature set includes at least one or more of the following: project name, borehole number, and borehole coordinates; the borehole feature set is compared with a preset borehole feature set dictionary, and new feature items that do not exist in the borehole feature set dictionary are automatically added to the borehole feature set dictionary, thereby realizing the dynamic expansion of the borehole feature set dictionary.

[0033] S3. Feature set code mapping establishment: Load the preset feature set code dictionary, which defines the correspondence between each feature item and the standardized code; match each feature item in the borehole feature set extracted in S2 with the feature set code dictionary to generate a feature set code mapping, wherein matched feature items are assigned the corresponding standardized code identifier, and unmatched feature items are marked with an empty identifier.

[0034] S4. Feature Set Coordinate Mapping Acquisition: Load the CAD format file of the borehole column chart base map to be processed, pre-parse the text entities in it and extract the text content, insertion point and coordinate range information of each text entity and store them in the entity cache; for each feature value in the borehole feature set obtained in S2, perform text matching search in the entity cache to determine the coordinate position of each feature value in the borehole column chart base map to be processed, record the spatial positioning information of each feature value in the form of a rectangular range, and form the feature set coordinate mapping.

[0035] Specifically, CAD format files are either DWG or DXF format. When the input file is in DWG format, it is automatically converted to DXF format before entity parsing. Text entities include TEXT, MTEXT, and INSERT entities. The information stored in the entity cache includes the text content string of each text entity, the coordinates of the insertion point, and the rectangular coordinate range defined by the coordinates of the lower left and upper right corners.

[0036] More specifically, the features in the borehole feature set are divided into key features and non-key features; for key features, a full match method is used for text search, and for non-key features, an inclusion match method is used for text search, wherein the key features include at least the borehole number.

[0037] S5. Dynamic generation of borehole column chart template: Load the base map file of the borehole column chart to be processed. Based on the feature set code mapping in S3 and the feature set coordinate mapping in S4, locate the position of each feature to be replaced in the base map file of the borehole column chart to be processed, delete the original feature text entities and insert the corresponding standardized code in the corresponding position. At the same time, clean up the redundant graphic entities in the map body area and output a standardized borehole column chart template file.

[0038] Specifically, by reading the x-axis and y-axis offset parameters set in the existing borehole column chart template file, the insertion position is fine-tuned when inserting standardized code text to compensate for possible positional deviations between deleting the original feature text entity and inserting the new entity; the output format of the standardized borehole column chart template file is DWG or DXF format, and a timestamp is automatically added to the file name during output to distinguish different versions.

[0039] Example 2

[0040] A borehole columnar section template generation system based on a multimodal visual large model, the architecture diagram of which is shown below. Figure 2 As shown, it consists of a base map structure splitting module, a feature extraction module, a code mapping module, a coordinate positioning module, and a template generation module.

[0041] The base map structure splitting module is used to obtain the base map of the borehole columnar diagram to be processed. According to the decomposition logic of the map frame and the body, the base map is split into a map frame part containing fixed structural information and a map body part containing variable content. The feature annotation area in the map body part is used as the core processing object for generating the borehole columnar diagram template.

[0042] The feature extraction module is used to convert the borehole bar chart background image to be processed into image format and encode it into Base64 data. The encoded Base64 data, along with preset prompt words, is input into the multimodal visual big model. For the core processing object described in the background image structure decomposition module, the multimodal visual big model performs semantic analysis on the image of the borehole bar chart background image to be processed, identifies and outputs a set of key-value pairs of borehole-related features as the borehole feature set.

[0043] The code mapping module is used to load a preset feature set code dictionary, which defines the correspondence between each feature item and a standardized code. The module matches each feature item in the borehole feature set extracted by the feature extraction module with the feature set code dictionary to generate a feature set code mapping, wherein matched feature items are assigned corresponding standardized code identifiers, and unmatched feature items are marked with an empty identifier.

[0044] The coordinate positioning module is used to load the CAD format file of the borehole columnar diagram base map to be processed, pre-parse the text-type entities in it and extract the text content, insertion point and coordinate range information of each text-type entity and store them in the entity cache; for each feature value in the borehole feature set obtained by the code mapping module, text matching search is performed in the entity cache to determine the coordinate position of each feature value in the borehole columnar diagram base map to be processed, and the spatial positioning information of each feature value is recorded in the form of a rectangular range to form a feature set coordinate mapping.

[0045] The template generation module loads the borehole column base map file to be processed. Based on the feature set code mapping in the code mapping module and the feature set coordinate mapping in the coordinate positioning module, it locates the position of each feature to be replaced in the borehole column base map file, deletes the original feature text entities and inserts the corresponding standardized code in the corresponding position, and cleans up redundant graphic entities in the map body area, and outputs a standardized borehole column template file.

[0046] The specific implementation methods of each module in this system are the same as those described in Example 1, and will not be repeated here.

[0047] Example 3

[0048] A computer-readable storage medium storing a computer program that, when executed by a processor, implements the borehole columnar section template generation method based on a multimodal visual large model as described in Embodiment 1 above, and the borehole columnar section template generation system based on a multimodal visual large model as described in Embodiment 2.

[0049] Example 4

[0050] Using the method described in Embodiment 1, combined with the system described in Embodiment 2 and the computer-readable storage medium described in Embodiment 3, a template is generated for the borehole columnar diagram base image to be processed. The specific steps of the actual operation are as follows.

[0051] Environmental preparation includes the following: Install dependencies: ezdxf (CAD file processing), Visual Multimodal Large Model API (large model calling), etc.

[0052] Configuration files: Ensure that config.json contains valid api_key and model_id, and that drill_info_dict.txt and drill_code_dict.txt have initialized basic features and codes.

[0053] S1. Borehole columnar chart base structure breakdown: A schematic diagram of the borehole columnar section base image to be processed is shown below. Figure 3 As shown, based on the decomposition logic of the borehole columnar plot frame and body, the input borehole columnar plot base map is split into the frame (including fixed structures such as title and signature) and the body (including variable content such as geological stratification and feature annotations). It is clear that the core area for template creation is the feature annotation part in the body.

[0054] S2. Drilling feature set extraction based on multimodal visual large model: Input base map and configuration initialization: Provide the borehole column base map to be processed (supports DWG and DXF formats), initialize the multimodal large model calling environment through the DrillHoleInfoExtractor class, and load parameters such as API key and model ID from the configuration file (config.json).

[0055] Feature recognition and extraction: The base map image is analyzed using a visual multimodal large model to extract borehole-related features (such as project name, borehole number, borehole coordinates, etc.). Specifically, this is achieved through the extract_information method: (1) Encode the base map image into Base64 format as the model input; (2) Combine the prompt word (prompt.txt) to require the model to output the key-value pairs of specified feature items and newly added features outside the table; (3) Example output feature set: {"Project Name": "Detailed Geotechnical Investigation of Phase I Project of XXX", "Brehole Number": "XBK1", ...}.

[0056] Dynamic update of feature set dictionary: The extracted feature set is compared with the preset [drill feature set dictionary] (drill_info_dict.txt), and the newly identified features (such as custom terms for special units) are automatically added through the append_new_keys method to realize the dynamic expansion of the dictionary.

[0057] S3. Establishment of Feature Set Code Mapping: Load the feature set code dictionary: The feature set code dictionary (drill_code_dict.txt) is read through the load_code_mapping function. This dictionary defines the correspondence between features and standardized codes (e.g., "drill hole number" corresponds to "[ZKBH]").

[0058] Feature-code matching: The map_info_to_codes function is used to map the borehole feature set extracted in step 2 to standardized codes, resulting in the [feature set code mapping]. Example: {"Borehole number": "[ZKBH]", "Project name": "[GCMC]", ...}. Unmatched features are marked with "[]".

[0059] The final feature set code mapping diagram is as follows: Figure 4 As shown.

[0060] S4. Feature set coordinate mapping acquisition: CAD base map loading and entity caching: The CADFeatureLocator class loads the CAD base map of the borehole column (automatically converting DWG to DXF format), and the preprocess_entities method pre-parses entities such as TEXT, MTEXT, and INSERT in the base map, extracts the text content, insertion point and coordinate range, and stores them in the entity_cache cache.

[0061] Feature location positioning: For each feature extracted in step 2, the feature value is matched in the entity cache using the find_text_in_cad function, and its coordinate range is calculated: (1) For key features such as borehole number, strict full matching is used; for other features, inclusion matching is used; (2) The feature value position is marked by a rectangular range, and the coordinates of the lower left corner and the upper right corner are taken as the replacement range. Example: "Borehole number: [ZKBH] :[2527.78,1762.58]-[2588.68,1769.58]", forming the

feature set coordinate mapping

[0062] The final feature set coordinate mapping diagram is shown below. Figure 5 As shown.

[0063] S5. Dynamic generation of borehole column chart template: Base map template loading: The drill column base map template is loaded through the DrillCADGenerator class, and the coordinate offsets (x_offset, y_offset) in the configuration file are read for fine-tuning of the text position.

[0064] Feature value replacement: Based on the feature set code mapping in step S3 and the coordinate mapping in step S4, the feature values ​​in the base image are replaced with standardized codes using the replace_features method. (1) Analyze the rectangular range in the coordinate mapping to locate the position of the feature to be replaced; (2) Delete the original feature text entity, insert the corresponding standardized code in the same position (such as replacing “XBK1” with “[ZKBH]”), and apply the offset to adjust the position.

[0065] Feature value replacement code diagram as shown below Figure 6 As shown.

[0066] Area cleanup and optimization: The clear_rectangle_area method is used to clean up redundant horizontal and diagonal line segments and other entities in the image body, while retaining the template framework.

[0067] Template Output: The `save_new_cad` method saves the replaced template as a DWG or DXF file, automatically adding timestamps to distinguish different versions. The final generated standardized borehole columnar section template file is shown in the diagram below. Figure 7 As shown.

[0068] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of this application can be implemented in various computer languages, such as object-oriented programming languages ​​like Java, C++, Python, and interpreted scripting languages ​​like JavaScript.

[0069] This application is described with reference to flowchart illustrations and / or block diagrams of methods, electronic devices (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing electronic device to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing electronic device, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0070] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing electronic device to operate in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0071] These computer program instructions can also be loaded onto a computer or other programmable data processing electronic device to cause a series of operational steps to be performed on the computer or other programmable electronic device to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable electronic device for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0072] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0073] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A method for generating borehole columnar section templates based on a multimodal visual large model, characterized in that, Includes the following steps: Borehole column chart base image structure decomposition: Obtain the borehole column chart base image to be processed, and according to the frame-body decomposition logic, decompose the base image into a frame part containing fixed structural information and a body part containing variable content, and use the feature annotation area in the body part as the core processing object for borehole column chart template generation; Drilling feature set extraction based on multimodal visual large model: The bottom image of the drill bar chart to be processed is converted into image format and encoded into Base64 data. The encoded Base64 data, together with the preset prompt words, is input into the multimodal visual large model. For the core processing object, the multimodal visual large model performs semantic analysis on the image of the bottom image of the drill bar chart to be processed, identifies and outputs the key-value pair set of drill-related features as the drilling feature set. Feature set code mapping establishment: Load a preset feature set code dictionary, which defines the correspondence between each feature item and the standardized code; match each feature item in the borehole feature set with the feature set code dictionary to generate a feature set code mapping, wherein matched feature items are assigned the corresponding standardized code identifier, and unmatched feature items are marked with an empty identifier; Feature set coordinate mapping acquisition: Load the CAD format file of the borehole column chart base map to be processed, pre-parse the text entities in it and extract the text content, insertion point and coordinate range information of each text entity and store them in the entity cache; for each feature value in the obtained borehole feature set, perform text matching search in the entity cache to determine the coordinate position of each feature value in the borehole column chart base map to be processed, and record the spatial positioning information of each feature value in the form of a rectangular range to form the feature set coordinate mapping; Dynamic generation of borehole column chart templates: Load the base map file of the borehole column chart to be processed, locate the position of each feature to be replaced in the base map file of the borehole column chart to be processed based on feature set code mapping and feature set coordinate mapping, delete the original feature text entities and insert the corresponding standardized code in the corresponding position, and clean up redundant graphic entities in the map body area, and output a standardized borehole column chart template file.

2. The method for generating borehole columnar section templates based on a multimodal visual large model according to claim 1, characterized in that: In the breakdown of the borehole columnar diagram base map structure, the fixed structural information of the frame portion includes: title and signature; the variable content of the body portion includes: geological stratification and feature annotation.

3. The method for generating borehole columnar section templates based on a multimodal visual large model according to claim 1, characterized in that: In the borehole feature set extraction based on the multimodal visual large model, the calling environment of the multimodal visual large model is initialized through a configuration file, which at least includes an API access key and a model identifier parameter.

4. The method for generating borehole columnar section templates based on a multimodal visual large model as described in claim 1, characterized in that: In the borehole feature set extraction based on the multimodal visual large model, the encoded Base64 data along with preset prompt words are input into the multimodal visual large model. The prompt words require the multimodal visual large model to output feature values ​​item by item according to the predefined feature item list, and to identify and output newly added feature items in the base map that exceed the range of the predefined list.

5. The method for generating borehole columnar section templates based on a multimodal visual large model according to claim 4, characterized in that: In the borehole feature set extraction based on the multimodal visual large model, the borehole feature set includes at least one or more of the following: project name, borehole number, and borehole coordinates. The borehole feature set is compared with a preset borehole feature set dictionary. New feature items that are not present in the borehole feature set dictionary are automatically added to the borehole feature set dictionary, thereby realizing the dynamic expansion of the borehole feature set dictionary.

6. The method for generating borehole columnar section templates based on a multimodal visual large model according to claim 1, characterized in that: In the acquisition of feature set coordinate mapping, the CAD format file is either DWG or DXF format. When the input file is in DWG format, it is automatically converted to DXF format before entity parsing. The text-type entities include TEXT entities, MTEXT entities, and INSERT entities. The information stored in the entity cache includes the text content string of each text-type entity, the coordinates of the insertion point, and the rectangular coordinate range defined by the coordinates of the lower left corner and the upper right corner.

7. The method for generating borehole columnar section templates based on a multimodal visual large model according to claim 6, characterized in that: In the acquisition of the feature set coordinate mapping, each feature item in the borehole feature set is divided into key feature items and non-key feature items; for key feature items, a full matching method is used for text search, and for non-key feature items, an inclusion matching method is used for text search, wherein the key feature items include at least the borehole number.

8. The method for generating borehole columnar section templates based on a multimodal visual large model according to claim 1, characterized in that: In the dynamic generation of the borehole column chart template, the x-axis offset and y-axis offset parameters set in the existing borehole column chart base template file are read, and the insertion position is fine-tuned when inserting standardized code text to compensate for the possible positional deviation between deleting the original feature text entity and inserting the new entity. The output format of the standardized borehole column chart template file is DWG or DXF format, and a timestamp is automatically added to the file name to distinguish different versions during output.

9. A borehole columnar section template generation system based on a multimodal visual large model, implementing the method of any one of claims 1-8, characterized in that, include: The base map structure splitting module is used to obtain the base map of the borehole columnar diagram to be processed. According to the decomposition logic of the frame and the body, the base map is split into a frame part containing fixed structural information and a body part containing variable content. The feature annotation area in the body part is used as the core processing object for generating the borehole columnar diagram template. The feature extraction module is used to convert the borehole bar chart background image to be processed into image format and encode it into Base64 data. The encoded Base64 data, along with preset prompt words, is input into the multimodal vision big model. For the core processing object mentioned in the background image structure decomposition module, the multimodal vision big model performs semantic analysis on the image of the borehole bar chart background image to be processed, identifies and outputs a set of key-value pairs of borehole-related features as the borehole feature set. The code mapping module is used to load a preset feature set code dictionary, which defines the correspondence between each feature item and a standardized code; the feature items in the borehole feature set extracted by the feature extraction module are matched with the feature set code dictionary to generate a feature set code mapping, wherein matched feature items are assigned corresponding standardized code identifiers, and unmatched feature items are marked with an empty identifier; The coordinate positioning module is used to load the CAD format file of the borehole column base map to be processed, pre-parse the text entities in it and extract the text content, insertion point and coordinate range information of each text entity and store them in the entity cache; for each feature value in the borehole feature set obtained by the code mapping module, text matching search is performed in the entity cache to determine the coordinate position of each feature value in the borehole column base map to be processed, and the spatial positioning information of each feature value is recorded in the form of a rectangular range to form a feature set coordinate mapping; The template generation module loads the borehole column base map file to be processed. Based on the feature set code mapping in the code mapping module and the feature set coordinate mapping in the coordinate positioning module, it locates the position of each feature to be replaced in the borehole column base map file, deletes the original feature text entities and inserts the corresponding standardized code in the corresponding position, and cleans up redundant graphic entities in the map body area, and outputs a standardized borehole column template file.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by the processor, it implements the borehole columnar section template generation method based on a multimodal visual large model as described in any one of claims 1-8.

Images