Fracturing design intelligent generation method and device

Abstract

The application relates to the field of oil and gas field development engineering, and provides a fracturing design intelligent generation method and device, wherein the method comprises the following steps: receiving a fracturing design generation request sent by a user through a client; using a large language model to generate a fracturing design report by operating a vector knowledge base, a structured database and a function module according to the fracturing design generation request and first prompt information; and sending the fracturing design report to the client; the first prompt information comprises first role information, skill information realized by the vector knowledge base, the structured database and the function module, outline requirement information, format requirement information, chapter requirement information and calling protocol information. The application can improve the fracturing design generation efficiency and the fracturing design generation quality.

Description

Technical Field

[0001] This application relates to the field of oil and gas field development engineering, and in particular to a method and apparatus for intelligent generation of fracturing design. Background Technology

[0002] Hydraulic fracturing is one of the key technologies for unconventional oil and gas resource development. Current fracturing design schemes mainly rely on engineers' manual analysis of geological data, construction data, and expert experience, which presents the following technical problems: (1) The design process is time-consuming, usually taking several days from data collection to report generation; (2) The design quality is limited by the engineer's personal experience, and the solutions developed by different engineers vary greatly; (3) Multi-source heterogeneous data (including geological design, fracturing design, construction summary, flowback report, production data and rock mechanics test report, etc.) are difficult to effectively integrate and utilize; (4) The parameter optimization process is complex and it is difficult to find the optimal fracturing design scheme quickly; (5) The preparation of design reports is tedious and prone to errors.

[0003] How to efficiently and accurately generate fracturing design schemes has become a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0004] This application provides a fracturing design generation method and apparatus to solve the problems of low utilization rate of relevant data resources, difficulty in data processing, long fracturing design cycle, low efficiency and unstable quality in existing fracturing design schemes.

[0005] To address the aforementioned technical problems, this application provides, in one aspect, a method for intelligent generation of fracturing design, comprising: Receives a fracturing design generation request sent by a user through a client; the fracturing design generation request includes at least design well information; Using a large language model, a fracturing design request and first prompt information are generated based on the fracturing design. Through an operation vector knowledge base, a structured database, and functional modules, a fracturing design report is generated. Send the fracturing design report to the client; The first prompt information includes first role information, skill information implemented by operating the vector knowledge base, the structured database and the functional modules, outline requirement information, format requirement information, chapter requirement information and calling protocol information; The outline requirements information is used to indicate the chapters in the fracturing design and the order in which the chapters are generated. The format requirements information is used to indicate the format requirements that must be followed when generating a fracturing design report; The chapter requirement information is used to indicate at least one of the following: chapter content requirement information, chapter content reference information, and function module call information; The invocation protocol information is used to indicate the invocation protocol information of the vector knowledge base, the structured database, and the functional modules; The vector knowledge base stores professional knowledge descriptions of historical fracturing designs, the structured database stores key parameter information of historical fracturing designs, and the functional modules include fracturing optimization logic and engineering calculation logic.

[0006] A second aspect of this application provides a fracturing design intelligent generation device, comprising: a fracturing design generation intelligent agent, wherein the fracturing design generation intelligent agent is configured with a large language model and connected to a vector knowledge base, a structured database, and functional modules, and the fracturing design generation intelligent agent is configured to perform the following operations: Receives a fracturing design generation request sent by a user through a client; the fracturing design generation request includes at least design well information; Using a large language model, a fracturing design request and first prompt information are generated based on the fracturing design. Through an operation vector knowledge base, a structured database, and functional modules, a fracturing design report is generated. Send the fracturing design report to the client; The first prompt information includes first role information, skill information implemented by operating the vector knowledge base, the structured database and the functional modules, outline requirement information, format requirement information, chapter requirement information and calling protocol information; The outline requirements information is used to indicate the chapters in the fracturing design and the order in which the chapters are generated. The format requirements information is used to indicate the format requirements that must be followed when generating a fracturing design report; The chapter requirement information is used to indicate at least one of the following: chapter content requirement information, chapter content reference information, and function module call information; The invocation protocol information is used to indicate the invocation protocol information of the vector knowledge base, the structured database, and the functional modules; The vector knowledge base stores professional knowledge descriptions of historical fracturing designs, the structured database stores key parameter information of historical fracturing designs, and the functional modules include fracturing optimization logic and engineering calculation logic.

[0007] A third aspect of this application provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the method described in any of the foregoing embodiments.

[0008] A fourth aspect of this application provides a computer-readable storage medium storing a computer program that, when executed by a processor of a computer device, implements the methods described in any of the foregoing embodiments.

[0009] The fifth aspect of this application provides a computer program product, the computer program product including a computer program that, when executed by a processor of a computer device, implements the method described in any of the foregoing embodiments.

[0010] The intelligent fracturing design generation method and apparatus provided in this application enable users to simply input a fracturing design generation request through a client. The fracturing design generation request includes at least design well information. Based on the fracturing design generation request and the first prompt information, a fracturing design report is generated using a large language model, through an operation vector knowledge base, a structured database, and functional modules. The fracturing design report is then sent to the client, enabling the effective use of historical fracturing design scheme data, achieving full automation and intelligence in fracturing design, and improving fracturing design efficiency, accuracy, and decision support capabilities.

[0011] To make the above and other objects, features and advantages of this application more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0012] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0013] Figure 1 A structural diagram of the intelligent generation system for fracturing design according to an embodiment of this application is shown; Figure 2 The interactive flowchart of the intelligent generation system for fracturing design according to an embodiment of this application is shown; Figure 3 This paper illustrates a schematic diagram of the service invocation mechanism in the intelligent generation system for fracturing design according to an embodiment of this application. Figure 4 A first flowchart of the intelligent generation method for fracturing design according to an embodiment of this application is shown; Figure 5 A second flowchart of the intelligent generation method for fracturing design according to an embodiment of this application is shown; Figure 6 A third flowchart of the intelligent generation method for fracturing design according to an embodiment of this application is shown; Figure 7A flowchart illustrating the standardized document determination process according to an embodiment of this application is shown; Figure 8 A first structural diagram of the intelligent fracturing design generation device according to an embodiment of this application is shown; Figure 9 A second structural diagram of the intelligent fracturing design generation device according to an embodiment of this application is shown; Figure 10 A structural diagram of a computer device according to an embodiment of this application is shown. Detailed Implementation

[0014] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0015] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, apparatus, product, or device that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices.

[0016] This specification provides the operational steps of the methods described in the embodiments or flowcharts, but based on conventional or non-inventive labor, more or fewer operational steps may be included. The order of steps listed in the embodiments is merely one possible execution order among many and does not represent the only possible execution order. In actual system or device products, the methods shown in the embodiments or drawings can be executed sequentially or in parallel.

[0017] It should be noted that the data involved in this application (including but not limited to data used for analysis, data stored, data displayed, etc.) are all information and data authorized by the user or fully authorized by all parties, and the acquisition, transmission, storage, use and processing of the relevant data comply with the relevant laws, regulations and standards of the relevant countries and regions.

[0018] It should be noted that in the embodiments of this application, certain software, components, models and other existing solutions in the industry may be mentioned. These should be regarded as exemplary and are only intended to illustrate the feasibility of implementing the technical solution of this application. However, it does not mean that the applicant has used or necessarily used the solution.

[0019] In some embodiments, a fracturing design intelligent generation system is provided to address the problems of low utilization of relevant data resources, difficult data processing, long fracturing design cycle, low efficiency, and unstable quality in existing fracturing design schemes. Specifically, such as... Figure 1 As shown, the intelligent generation system for fracturing design includes: client 101 and server 102.

[0020] Client 101 provides a user interface through which users can upload fracturing engineering documents and input fracturing design generation requests, fracturing design modification requests, and query requests (e.g., viewing design results). During implementation, guidance information can be displayed in the user interface to guide users in inputting fracturing design generation requests, fracturing design modification requests, and query requests. This improves the quality of fracturing design report generation and the quality of question responses.

[0021] Server 102 receives requests from clients. When the client sends a fracturing engineering document, it updates the vector knowledge base and structured database based on the document. When the client sends a fracturing design generation request, it automatically generates a fracturing design report and sends it to client 101. When the client sends a fracturing design modification request, it automatically modifies the fracturing design report and sends the modified report to client 101. When the client sends a query request, it generates a response and sends it to client 101.

[0022] In detail, client 101 can be a self-service terminal device, a mobile terminal (such as a smartphone), a monitor, a desktop computer, a tablet computer, a laptop computer, or other electronic devices with a certain physical form, or it can be a software application running on the aforementioned electronic devices.

[0023] The server 102 can be a server device with computing and network interaction functions, or a software system running on the server device that provides business logic for data processing and network interaction.

[0024] During implementation, server 102 is equipped with a fracturing design generation agent, a fracturing design modification agent, and a knowledge extraction agent. The fracturing design generation agent automatically generates a fracturing design report based on the fracturing design generation request and the first prompt information. The fracturing design modification agent automatically modifies the fracturing design report based on the fracturing design modification request and the third prompt information. The knowledge extraction agent extracts professional knowledge descriptions and key parameter information from historical fracturing engineering documents based on the second prompt information.

[0025] The interactive process of the intelligent generation system for fracturing design is explained in detail below with reference to the accompanying drawings. For example... Figure 2 As shown, the interactive process of the intelligent generation system for fracturing design includes: Step 201: Client 101 receives the multi-source heterogeneous fracturing engineering document uploaded by the user and sends the multi-source heterogeneous fracturing engineering document to server 102.

[0026] The multi-source heterogeneous fracturing engineering documents consist of fracturing-related documents from developed wells in the target block, i.e., fracturing-related documents from adjacent wells of the wells to be developed. Specifically, these include fracturing engineering documents of several professional types: geological design, fracturing design, fracturing construction, flowback monitoring, fracturing production, and rock mechanics experiments. Each type of fracturing engineering document corresponds to one stage. Specifically, the above types of fracturing engineering documents correspond to the geological, design, construction, flowback, production, and experimental stages, respectively.

[0027] Geological design documents mainly include formation-related data, such as formation porosity, permeability, and oil saturation interpreted from well logging data.

[0028] Fracturing design documents mainly include well logging information (e.g., well number, reservoir section, well type), core reservoir parameters (e.g., geological parameters, mechanical and fluid parameters, interlayers, etc.), supporting basic data (e.g., test data, reference data from adjacent wells, etc.), fracturing target data (e.g., displacement, proppant dosage, fluid usage, fluid-to-spar ratio, target stimulation volume, target fracture parameters, target production, safety and environmental protection targets, etc.), core process data (e.g., fracturing process type, segmented and clustered design, fracture parameter design, etc.), fracturing fluid and proppant data (e.g., fracturing fluid, proppant, etc.), construction parameters and reservoir protection data, safety, environmental protection and emergency response plan data, implementation support and monitoring data, etc. The goal of fracturing design is to design the optimal pumping construction scheme based on the geological conditions of the formation to maximize the expected production.

[0029] Fracturing construction documents mainly include well logging basic information, construction design data (such as process parameters, fracturing fluid parameters, proppant parameters, etc.), material and equipment information, construction process information, personnel and safety information, etc.

[0030] The flowback monitoring documents mainly include well logging basic information, flowback fluid data (such as production data, fluid properties, etc.), anomaly monitoring data, and flowback effect evaluation data.

[0031] Fracturing production documentation mainly includes basic well logging information, dynamic production data (such as production data, pressure data, fluid property data, etc.), production measures records, effect analysis and optimization suggestions, etc.

[0032] Rock mechanics experimental documents mainly include various properties of strata rocks, such as fracture strength and Young's modulus. Experiments can also obtain data on porosity, permeability, and saturation.

[0033] Step 202: Server 102 preprocesses the multi-source heterogeneous fracturing engineering documents to obtain standardized documents.

[0034] In some embodiments, this step includes the following steps: a1. Preprocess the multi-source heterogeneous fracturing engineering documents to obtain the first document.

[0035] Preprocessing includes, but is not limited to, standardization of encoding and noise removal. Multi-source heterogeneous fracturing engineering documents come in various formats. For each format, an encoding tool is used to convert the document to a unified format. For example, text files are converted to docx, table files to xlsx, multimedia files to pptx, and images to jpg.

[0036] b1. Use a document classification model to identify the first document and obtain the professional type of the first document.

[0037] The professional categories include: geological design, fracturing design, fracturing construction, flowback monitoring, fracturing production, and rock mechanics experiments. The document classification model can be pre-trained using a neural network based on fracturing engineering documents with pre-labeled categories.

[0038] c1 determines the format type, well identifier, and block identifier of the first document.

[0039] The file types include: text, table, multimedia, and image. Hash tags and block tags can be obtained from the basic information of the first document. Text files include formats such as doc and docx. Table files include formats such as xlsx, xls, and csv. Multimedia files include formats such as ppt and pptx.

[0040] d1 uses a parsing algorithm related to the format type of the first document to parse the first document and obtain the second document.

[0041] The second document is obtained by splitting the first document, which can reduce the size of individual text blocks and avoid exceeding the upper limit of the model context.

[0042] For the first document, an adaptive text segmentation algorithm is used to divide it into appropriate paragraphs based on the principle of semantic integrity, resulting in the second document. Adaptive text segmentation algorithms typically use methods such as page-based segmentation, symbol-based segmentation, or semantic segmentation. During segmentation, it is also necessary to summarize and integrate the preceding and following text blocks to improve the semantic context of each individual text block.

[0043] For a first document of table type, extract table structure information and data content from a second document, and use the extracted information to form the second document.

[0044] For the first document, which is an image, optical character recognition (OCR) technology is used to recognize the text and numerical information in the first document (i.e., image data), and the recognized information constitutes the second document.

[0045] e1 converts the second document into a standardized document.

[0046] During this step, the second document is converted into a standardized document according to the organization's requirements. For example, the segmented text document is converted into a Markdown document according to certain formatting requirements (such as first-level headings, second-level headings, body text format, table format, etc.). Multimedia files such as images will be uniformly converted into file addresses embedded in the original text location. When users or models retrieve the corresponding text block, they can click the file address to jump to and read it.

[0047] f1 determines the index of standardized documents according to the preset naming rules, and stores the standardized documents to the document library according to the index.

[0048] The preset naming rules include at least a hash symbol, a block symbol, and a professional type. For example, standardized documents can be named according to the naming rule of "hash number-block-document type-timestamp". Naming standardized documents using preset naming rules enables the traceability of standardized documents.

[0049] Step 203: Server 102 invokes the knowledge extraction agent. The agent uses a large language model to extract key parameters, relationships between parameters, and professional knowledge descriptions related to fracturing design from standardized documents based on the second prompt information. A structured database is established based on the key parameters and their relationships; a vector knowledge base is established based on the professional knowledge descriptions related to fracturing design. During implementation, the key parameters and their relationships are quality-verified and then stored in the structured database; the professional knowledge descriptions related to fracturing design are vectorized and then stored in the vector knowledge base.

[0050] The professional knowledge descriptions related to fracturing design are presented as text data, such as conclusions, analytical statements, and descriptive statements, stored in a vector knowledge base. Each text entry also requires tags such as hash symbol, block, text source, text section, text type, and text content. Text source examples include fracturing design, construction summary, progress report, and measurement analysis. Text section examples include the chapter name within the document. Text type examples include conclusion, analysis, and description. Text content examples include surface description, surface transportation, geographical location, and surface climate.

[0051] Key parameters include tabular data and sequential data. Tabular data contains key parameter information for each well, such as geological and engineering parameters. Sequential data consists of real-time data recorded during construction, such as well logging sequence data, drilling sequence data, and pumping curves. The structured database uses the well number as the primary key to distinguish each data entry.

[0052] The second prompt information includes second role information, extraction task requirement information, fracturing domain knowledge dictionary and storage instruction information.

[0053] The second role information is used to indicate the role of the information extraction expert and the overall requirements information.

[0054] The task requirements information is used to indicate the format, storage, and other information for the extracted data.

[0055] The fracturing knowledge dictionary includes parameter information for geological parameters, engineering parameters, construction parameters, and production parameters. Parameter information includes parameter identifier, unit of measurement, value range, and semantic description. In one specific embodiment, the fracturing knowledge dictionary includes over 200 parameter items. The fracturing knowledge dictionary can be developed by those skilled in the art or established through analysis of existing fracturing knowledge.

[0056] Storage indication information is used to indicate the address and storage statement information of the structured database and vector knowledge base.

[0057] In some embodiments, the second prompt message is, for example: { Role Requirements: You are an information extraction expert in the petroleum industry. You need to retrieve the information specified in the task requirements {@task requirements} from the given text, strictly generate the corresponding structured information, and store it in the corresponding storage object.

[0058] Task Requirements: Retrieve knowledge points and key parameters from the knowledge dictionary, determine the task type, and complete the extraction. Organize the extracted results according to the JSON transmission format. For knowledge points, copy the corresponding text, retrieve the document containing the text, add hash symbols, blocks, text sources, and text chapters, and analyze the content of the knowledge points. Summarize the text type and text content tags, with the tag list including {@tag list}. Store the results in the vector knowledge base {@vector knowledge base} according to the transmission format. For key parameters, extract the corresponding values, adjust them according to the format requirements, and store them in a structured database according to the transmission format.

[0059] JSON passing format: @ passing format.

[0060] Knowledge Dictionary: A dictionary consisting of knowledge points and key parameters, including a list of knowledge points and a list of key parameters. Each knowledge point includes an explanation and a content example, and each key parameter includes an explanation and a content example. The explanation section includes: definitions of technical terms, synonym suggestions, warnings of similar-looking words, and hints for common task requirements.

[0061] Vector Knowledge Base: @Vector Knowledge Base interface address and storage statement reference.

[0062] Structured database: @Structured database interface address and storage statement reference.

[0063] }

[0064] In some embodiments, the large language model in this step is a large language model pre-trained on texts in the fracturing domain, such as a pre-trained model based on the Transformer architecture. The large language model performs deep semantic analysis on the standardized document content based on the second cue information. Numerical parameters, such as formation depth, fracturing pressure, displacement, and sand ratio, are extracted from the standardized documents using named entity recognition technology. Relationships between parameters are established using relation extraction technology. Descriptive information about fracturing technology is extracted using text summarization technology.

[0065] During this step, data quality verification, outlier detection, unit standardization, and logical consistency checks are also performed on numerical parameters and the relationships between them. Then, a structured database is built based on the processed data.

[0066] The structured database stores key parameters, such as a horizontal section length of 2000m and a displacement of 18 L / min.

[0067] During this step, the professional knowledge description information related to fracturing design is also vectorized and converted into a high-dimensional vector representation before being stored in a vector knowledge base.

[0068] The vector knowledge base stores vector information of statements, such as: Well xx is located under the xx structure, is a sandstone stratum, and has xxx geological characteristics.

[0069] Steps 201 to 203 involve establishing the vector knowledge base and structured database. Based on the second instruction, standardized text blocks are extracted and stored using a large language model. After reading the standardized file, the large language model first determines the possible task content based on the document's content. Then, it compares and checks the knowledge point list and key parameter list in the fracturing domain knowledge dictionary to view the extracted content and specific requirements. The extraction task is then completed, and the extraction results are organized, transmitted, and stored through a data interface.

[0070] In practice, the stored results are also manually sampled to check indicators such as extraction accuracy, recall, and F1 score. The decision to re-extract the data is made based on the test results.

[0071] Step 204: Client 101 sends the fracturing design generation request input by the user to server 102.

[0072] In some embodiments, the fracturing design generation request includes at least design well information. The design well information includes the well number and the block in which it is located.

[0073] In some embodiments, the fracturing design generation request includes, in addition to design well information, at least one of the following design parameters: reference block information, fracturing method information, fracturing fluid information, proppant information, optimization target information, pumping requirements information, and fracturing parameter information. The fracturing parameter information may include, for example, whether temporary plugging is required, and the length of the initial stage. The optimization target information includes multiple optimization objectives that balance production, cost, and risk.

[0074] In some embodiments, users can input design parameters using natural language.

[0075] In some embodiments, users input design parameters through a client configuration interface. This configuration interface has multiple parameter controls for users to input design parameters. The parameter controls can be in an option mode. Users can select design parameters simply by manipulating the parameter controls through a drop-down list.

[0076] In some specific embodiments, the fracturing design generation request is, for example: I need you to complete the fracturing design report for Well B in Block A (the block it is located in) based on historical data. The report should use a segmented, clustered bridge plug perforation fracturing method (fracturing method), with the first-year production efficiency as the optimization objective (optimization objective), and employ a low-spar, batch-addition method for pumping (pumping requirements). Block D (reference block) is similar to Block A and can be considered together with Block A.

[0077] Step 205: Server 102 receives fracturing design generation request.

[0078] Step 206: Server 102 calls the fracturing design generation agent. The fracturing design generation agent uses a large language model to generate a fracturing design report based on the fracturing design generation request and the first prompt information, through the operation vector knowledge base, structured database and functional modules; and sends the fracturing design report to client 101.

[0079] The fracturing design report conforms to industry technical specifications and design requirements, and supports visual preview and export in multiple formats.

[0080] The first prompt information includes the first role information, the skill information implemented by the operation vector knowledge base, structured database and functional modules, the outline requirement information, the format requirement information, the chapter requirement information and the calling protocol information.

[0081] The first role information indicates the role of the fracturing design expert and the requirements to be followed. These requirements include adherence to the fracturing design generation request, outline requirements, formatting requirements, and chapter requirements when generating the fracturing design report.

[0082] The operation of the vector knowledge base, the structured database, and the skill information implemented by the functional modules includes: obtaining professional knowledge description information of the design well by querying the vector knowledge base; obtaining key parameter information of the design well by querying the structured database; obtaining geological description and construction suggestions of the design well by comparing data of neighboring wells related to the design well in the vector knowledge base; and obtaining optimization parameters and engineering calculation results by calling the functional modules.

[0083] The outline requirements information is used to indicate the chapters in the fracturing design and the order in which the chapters are generated.

[0084] The format requirements information indicates the format requirements that must be followed when generating a fracturing design report.

[0085] The chapter requirement information is used to indicate at least one of the following: chapter content requirement information, chapter content reference information, and function module call information.

[0086] The calling protocol information is used to indicate the calling protocol information of the vector knowledge base, structured database, and functional modules.

[0087] The vector knowledge base stores professional knowledge descriptions of historical fracturing designs, while the structured database stores key parameter information of historical fracturing designs.

[0088] The functional modules include fracturing optimization logic and engineering calculation logic. Specifically, the functional modules include a parameter optimization module and an engineering calculation module.

[0089] The parameter optimization module is used to determine the optimal production index and optimal parameter combination based on the fracturing design generation request, chapter requirement information, vector knowledge base and structured database.

[0090] The engineering calculation module is used to perform engineering calculations based on fracturing design requests, chapter requirements, optimal parameter combinations, vector knowledge bases, and structured databases.

[0091] In some embodiments, the engineering calculation module includes a segmentation and clustering module, a pumping design module, and a temporary plugging design module.

[0092] In some embodiments, the fracturing design generator and functional modules uniformly adopt the industry-standard Model Context Protocol (MCP) as a unified interaction basis.

[0093] The MCP protocol defines a complete set of specifications for service discovery, capability negotiation, task invocation, and result return. Each functional module (such as the parameter optimization module) provides services externally as an MCP server and declares the following information to a central registry (or directly informs the large model in the prompt) at startup: Endpoint address: The network location of the module; Capability description: What can be done (e.g., calculating the optimal cluster location, optimizing construction flow rate); Input Schema: What input parameters are required (e.g., geological parameter JSON object, engineering constraints); Output Schema: What results are returned (e.g., a JSON object listing segment cluster positions, or an optimized JSON object of parameters).

[0094] When a functional module is invoked, its input parameters are determined according to the module's invocation protocol. These parameters are then input into the module, which calculates and generates the output data. The fracturing design and generation agent then organizes the data according to the required format based on the output data from the functional module.

[0095] In some embodiments, the first prompt message is, for example: { Role Requirements: You are a fracturing design expert in the oil industry. You need to complete the fracturing design task for the designed well based on user instructions and generate a fracturing design report. You need to follow the outline requirements, format requirements, and chapter requirements, and improve the fracturing design report through methods such as querying, comparing, and calling.

[0096] Skills required: You need to obtain professional knowledge descriptions of the design wells from the text vector knowledge base {@vector knowledge base}, and the professional text knowledge summary list {@knowledge list} contains the professional knowledge points that may need to be queried; you need to provide geological descriptions and construction suggestions by comparing the design wells and adjacent wells in the historical corpus {@vector knowledge base}; you need to obtain numerical information of key parameters by querying the structured database {@structured database}; you need to obtain the results of the fracturing parameters that need to be optimized by calling the parameter optimization module {@parameter optimization}.

[0097] Outline Requirements: The outline for fracturing design includes: 1. Basic Drilling and Completion Information; 1.1 Reservoir Location and Environmental Conditions; 1.2 Well Location Deployment; 1.3 Basic Data; 1.4 Comprehensive Interpretation of Logging Data; 1.5 Cementing Quality; 2. Geological Engineering Design; 2.1 Reservoir Engineering Fundamentals; 2.2 Fracturing Stimulation Engineering Analysis; 3. Fracturing Stimulation Design Principles; 4. Fracturing Stimulation Engineering Design; 4.1 Selection of Segment Completion Technology; 4.2 Selection of Fracturing Engineering Materials; 4.3 Segment Cluster Selection and Scale Design; 4.4 Selection of Fracturing Parameters; 4.5 Temporary Fracturing Plugging Technology; 4.6 Fracturing Pumping Procedure; 4.7 Contingency Plan for Complex Issues; 4.8 Material Statistics; 5. Construction Preparation and Requirements; 5.1 Construction Method; 5.2 Data Acquisition; 5.3 Wellbore Preparation; 5.4 Well Site Requirements; 5.5 Equipment Requirements; 5.6 Material Preparation; 5.7 Construction Requirements; 6. Flowback System and Requirements; 7. 7.1 Quality, Health, Safety and Environmental Protection Requirements; 7.2 Health, Safety and Environmental Protection Requirements; 8 Well Control Requirements; 9 Risk Identification and Construction Emergency Measures; 10 Standards, Regulations and Procedures to be Implemented; Appendix 1 Wellbore Trajectory Data Table for Well B; Appendix 2 Casing Joint Data Table for Well B; Appendix 3 Segmented and Clustered Design Table for Well B; Appendix 4 Approval and Signature Form for Well B; Design Highlights.

[0098] Formatting Requirements: Use natural language paragraphs to describe content, closely resembling human language analysis and expression. Please primarily refer to the description format and expression logic in the chapter requirements {@Chapter Requirements}. Avoid using point-based summaries and numbered descriptions.

[0099] Chapter Requirements: Organize the corpus according to the content requirements of each chapter, with corresponding content references for each chapter.

[0100] Chapter 1: Basic Information After Well Completion; Content requirements: The description should include 1.1 reservoir location and environmental conditions, 1.2 well location deployment, 1.3 basic data, 1.4 comprehensive interpretation of logging, and 1.5 specific details of cementing quality.

[0101] }

[0102] Chapter 1.1: Reservoir Location and Environmental Conditions; Content requirements: Describe the geographical location, surface conditions, surface climate, and surface transportation conditions of the block where the design well is located.

[0103] Reference: {@Block} is located in {@Geographical Location}, see {@Geographical Location Map}; the surface is {@Surface Conditions} terrain, the surface temperature and wind force can reach {@Surface Climate} level, and the surface transportation in the work area is {@Surface Transportation}, which is relatively convenient.

[0104] } ...

[0106] Knowledge list: Knowledge point 1, Knowledge point 2, Knowledge point 3...; Vector Knowledge Base: @Vector Knowledge Base address and MCP calling protocol; Structured database: @structured database address and MCP calling protocol; Parameter optimization: @Parameter optimization module interface and MCP calling protocol; Output: Generate fracturing design documents to the workbench.

[0107] }

[0108] The fracturing design generator agent embeds a format description and content outline in the initial prompt message, requiring the model to organize the corpus according to a specific format and output according to a fixed chapter order and text requirements. The format requirements are embedded in the initial prompt message, declaring the usage requirements of various skills to be used later, such as querying, comparing, and calling, as well as the corresponding vector knowledge base, structured database, and the calling addresses / interfaces and calling protocols of functional modules.

[0109] Users do not need to perform queries, comparisons, or invocations through the fracturing design generation agent. They only need to input a fracturing design generation request and wait for the agent to generate the fracturing design report. The fracturing design generation agent determines the operations to be performed based on chapter requirements, skill information, and invocation protocol information. These operations include retrieving key parameters from a structured database, acquiring knowledge from a vector knowledge base, integrating and analyzing the key parameters and knowledge, and obtaining optimization and engineering parameters through interaction with functional modules. For example, in the outline example above, the basic information descriptions (Chapters 1 and 2) and supplementary content (Chapters 3 and 5-10) are derived from historical data and generated by the large model by pulling data from the vector knowledge base and structured database; the substantive design content (Chapter 4) requires the cooperation of functional modules to generate.

[0110] In some embodiments, during the generation of fracturing design reports, when the fracturing design generation agent encounters a chapter requiring specialized calculations (such as Chapter 4, "Fracturing Modification Engineering Design"), it will initiate a complete interaction based on pre-configured skill information and calling protocol information. The interaction process between the fracturing design generation agent and functional modules includes: Step 1: Task identification and breakdown.

[0111] The large language model parses the fracturing design generation request and identifies the steps that require the intervention of functional modules. For example, when writing "4.3 Segment Cluster Selection and Scale Design", the "Segmentation and Clustering Module" needs to be called.

[0112] Step 2, construct the MCP request.

[0113] (1) The large language model determines the input data of the functional modules based on the input schema of the functional modules.

[0114] For example, the input schema for the segmentation and clustering module is collected and organized from the following sources: User instructions: such as "Use segmented cluster bridge plug perforation fracturing method".

[0115] Vector knowledge base: static geological description of the target well and engineering experience of adjacent wells.

[0116] Structured database: Drilling and logging parameters of the target well (such as well depth, well inclination, GR, resistivity, sonic transit time curve data, etc.) are retrieved from the structured database. These are the core inputs of the segmentation and clustering module.

[0117] Context: The currently generated design content.

[0118] (2) The large language model packages the input data into a standard request instruction according to the format required by the MCP protocol (such as JSON). The request instruction will specify the following information: Skills employed: Fracturing design segmentation and clustering optimization; Target module: fracture_stage_cluster_optimizer; Input data: A structured JSON object containing all the necessary drilling and logging parameters.

[0119] Step 3: Send request command and receive result.

[0120] For example, the large language model sends request commands to the segmentation and clustering module through the MCP client. The segmentation and clustering module performs complex numerical calculations (involving rock mechanics analysis, geostress field simulation, fracture propagation simulation, etc.), and then packages the calculation results (such as the recommended number of segments, the start / end well depth of each segment, the number of clusters, the cluster spacing, etc.) into a standard response according to the output schema, and returns it to the large language model through the MCP protocol.

[0121] Step 4: Result analysis and content generation; The large language model receives and parses the MCP response to obtain structured calculation results. Then, combined with the format requirements in the chapter requirements (such as natural language paragraph descriptions), the MCP response is transformed into professional text and tables for the fracturing design report.

[0122] For example, if the large language model learns from the JSON returned by the functional module that "the first segment starts at a depth of 3200m and ends at a depth of 3225m, divided into 3 clusters", the large language model will generate the following paragraph based on the context: Based on well logging interpretation and geostress profile analysis, this well entered the target formation at 3200m. To effectively control fracture height and improve the targeting of the fracturing, the 3200m to 3225m section was designed as the first fracturing stage, with three clusters of perforations within this stage, spaced 8-10m apart, to form a complex fracture system with multiple interfering but fully fracturing lines. Simultaneously, this data will be automatically entered into "Appendix 3, Well B Segmentation and Clustering Design Table".

[0123] In some embodiments, the functional modules include a parameter optimization module, a segmentation and clustering module, a pumping design module, and a temporary plugging design module.

[0124] Continuing with the specific example above, when generating Chapter 4, the chapter requirements for Chapter 4 should include the following functional module call information: { ... Calling requirements: First, the parameter optimization module is called to obtain the average segment spacing (m), the actual average cluster spacing (m), and a set of construction liquid strength, construction sand addition strength, construction pre-construction liquid ratio (%), construction sand ratio (%), and maximum actual construction discharge as references, and the results are stored. Then, the segmentation and clustering module is called to obtain the segmentation and clustering results; Based on the segmentation results, the parameter optimization module was called again for each fracturing segment to calculate the construction fluid strength, construction sand addition strength, pre-construction fluid ratio (%), construction sand ratio (%), and maximum actual construction discharge rate for each fracturing segment. When temporary plugging is required, the temporary plugging design module is invoked to optimize the temporary plugging process and material scheme for each fracturing section. Finally, based on the parameter optimization results of each fracturing section and the design results of the temporary plugging scheme, a pumping procedure is formed.

[0125] @Parameter Optimization Module MCP Service Protocol; @Segmentation and Clustering Module MCP Service Protocol; @Temporary Blockage Design Module MCP Service Protocol; @Pump Design Module MCP Service Agreement. ...

[0127] }

[0128] In some embodiments, the process of the fracturing design generating agent calling the parameter optimization module includes: Step 1: The fracturing design generates input data for the intelligent agent to organize the parameter optimization module, and inputs the input data into the parameter optimization module.

[0129] The input data includes: Geological parameters: Actual length of the modified section (m), Class I oil layer (m), Class II oil layer (m), Class III oil layer (m), minimum porosity (%), maximum porosity (%), average porosity (%), minimum permeability (mD), maximum permeability (mD), average permeability (mD), minimum oil saturation (%), maximum oil saturation (%), average oil saturation (%), oil layer drilling rate (%), number of sections lost due to casing deformation, number of sections not fully modified due to casing deformation; Engineering parameters: average segment spacing (m), actual average cluster spacing (m), construction fluid intensity, construction sand addition intensity, pre-construction fluid ratio (%), construction sand ratio (%), maximum actual construction discharge, and information on the well number, block, and reference block of the current design well.

[0130] Step 2: The parameter optimization module checks the model library for trained production prediction models based on the design well block and reference block information. If a model exists, it calls it; otherwise, it retrieves the structured data for the design well block and reference block from the structured database and transmits it to the model training module for production prediction model training. If the model training process begins, both the parameter optimization module and the front-end design and model generation module remain inactive. After the production prediction model is trained, its performance and name are confirmed, and it is ready to be called.

[0131] The production prediction model has the following functions: inputting corresponding geological parameters and engineering parameters (input data from step 1), it predicts production indicators. The parameters predicted by the production prediction model are related to the indicator selection during training. The attributes of the production prediction model include model name, training hyperparameter combination, training data source, training data characteristics, model indicators, and training time.

[0132] Step 3: After confirming the production prediction model, the optimization constraint range for parameter optimization is determined. This constraint range is generally determined based on the characteristics of the model training data (such as the maximum and minimum values ​​of parameters, the 95th and 5th quantiles). The parameter optimization module automatically sets the hyperparameters for the optimization process and performs optimization autonomously. The parameter optimization module has a built-in logic to analyze the rationality of the optimization process and results. Through repeated experiments and hyperparameter adjustments, a set of optimized fracturing design parameters is finally selected, including the average segment spacing (m), the actual average cluster spacing (m), the strength of the fluid used in construction, the strength of the sand added during construction, the proportion of fluid placed before construction (%), the sand ratio during construction (%), the maximum actual construction discharge rate, and feedback is provided to the fracturing design generation agent.

[0133] The optimization process uses Newton's gradient descent method. The optimizer randomly selects multiple sets of design parameter schemes and inputs them into the production prediction model to predict the corresponding production indicators. Then, in the next iteration step, based on the changes in production indicators among different schemes, the design parameter combination is directionally changed along the gradient descent (or ascent) direction, and the production indicators are calculated again. Then, the changes in production indicators are calculated again, and the design parameter combination is directionally changed again along the gradient descent (or ascent) direction until the changes in production indicators meet the convergence condition. At this point, the optimization of the design parameter combination is completed.

[0134] In some embodiments, the segmentation and clustering module invocation process includes: Step 1: After the average segment spacing (m) and the actual average cluster spacing (m) are calculated, obtain the input data of the segmentation module and input the input data into the segmentation and clustering module.

[0135] The input data for the segmentation and clustering module includes: well number, block, reference block, average segment spacing (m), actual average cluster spacing (m), first segment length, construction fluid intensity, construction sand addition intensity, maximum actual construction discharge, etc.

[0136] Step 2: The segmentation and clustering module retrieves the following parameters based on the hash number: Drilling parameters: weight on bit, torque, rate of drilling, rotational speed, inlet flow rate, inlet density, mechanical energy, bit diameter, screw speed; Well logging parameters: starting depth, ending depth, depth, inclination, formation pressure, minimum horizontal principal stress, reservoir type; Basic information: well type, artificial well bottom, coupling location.

[0137] Step 3: The segmentation and clustering module calculates the formation strength evaluation index based on the input data and built-in formulas.

[0138] The formation evaluation index uses drilling pressure, torque, drilling speed, rotational speed, inlet flow rate, inlet density, mechanical specific energy, drill bit diameter, screw speed, well depth, well inclination, formation pressure, and minimum horizontal principal stress at each depth as inputs. A value is calculated to represent the formation strength at the corresponding depth. Based on the formation strength evaluation index and using the average segment spacing (m) as a segmentation constraint, a dynamic programming algorithm is used to segment the formation. The goal is to minimize the sum of variances (or sum of absolute ranges, or sum of standard deviations) of the formation strength evaluation indexes within all fracturing segments, thus generating the segmentation locations for each fracturing segment.

[0139] Step 4: Within each fracturing section, based on the formation strength evaluation index and the actual average cluster spacing (m) as the clustering constraint, the formation is clustered using a dynamic programming algorithm. The goal is to minimize the variance (or absolute range, or standard deviation) of each cluster within the section and avoid the joint position (more than 1m away). This generates the number of holes, hole diameter, hole density, and cluster location for each cluster.

[0140] Step 5: Organize the calculation results, and send the table containing the segment positions, cluster positions and perforation parameters of each cluster to the fracturing design generator.

[0141] In some embodiments, the call flow for the temporary blocking design module includes: Step 1: Obtain the input data from the temporary blockage design module and send the input data to the temporary blockage design module.

[0142] The input data includes: segmentation and clustering results, parameter optimization results of the fracturing section, well number, block, reference block, etc. The segmentation and clustering results include cluster location, cluster aperture, number of cluster holes, cluster hole density, and maximum actual operational discharge rate.

[0143] Step 2: The temporary plugging design module performs temporary plugging simulations (e.g., 3D / 2D simulation, finite element / discrete element simulation) based on cluster location, cluster aperture, number of cluster holes, cluster hole density, and maximum actual construction discharge. Through multiple temporary plugging simulations, the size, weight, quantity, and temporary plugging pressure of the temporary plugging material are adjusted, and different temporary plugging methods are optimized to finally determine the temporary plugging method and scheme for the fracturing section.

[0144] Step 3: Organize the temporary plugging methods and schemes, and feed them back to the fracturing design to generate the intelligent agent.

[0145] In some embodiments, the pump design module invocation process includes: The pumping design module includes a pumping program. This program serves as a pumping construction guide, detailing the appropriate flow rate and duration of delivery, as well as how proppant (sand addition) should be added during delivery and the required sand-liquid concentration (construction sand ratio (%)).

[0146] The pumping procedure is divided into three stages: the pre-pumping stage, the sand-carrying stage, and the displacement stage. The pre-pumping stage delivers only fluid without adding sand, used to build up pressure and increase discharge rate downhole. The sand-carrying stage adds sand, generally operating at maximum discharge rate, with the sand addition rate and the sand-to-water ratio (%) increasing progressively. The displacement stage delivers only fluid without adding sand, used to displace the sand-carrying fluid out of the wellbore and clean the wellbore.

[0147] The pumping procedure has the following parameters: stage number, stage sand addition amount, cumulative sand addition amount, stage liquid consumption amount, cumulative liquid consumption amount, stage sand ratio, stage discharge rate, and stage time. Knowing only the stage sand addition amount, stage discharge rate, and stage time is sufficient to calculate the remaining parameters, making them the main control parameters for pumping design.

[0148] Step 1: Obtain input data from the pump design module.

[0149] The input data for the pump design module includes: Segmentation and clustering results, parameter optimization results of fracturing sections, temporary plugging design results, well number, block, reference block, etc.; Engineering parameters: average segment spacing (m), actual average cluster spacing (m), construction liquid strength, construction sand addition strength, pre-construction liquid ratio (%), construction sand ratio (%), maximum actual construction discharge.

[0150] Step 2: Design the pre-fluidization stage based on the input data.

[0151] During implementation, the pumping capacity will be increased to the maximum capacity in 2-3 stages according to a progressively increasing sequence, with the stage time reasonably arranged to ensure that the pumped liquid volume is consistent with the pre-flush liquid volume. For example, if the maximum pumping capacity is 8 L / min, then it will be divided into 3 stages: 3 L / min, 6 L / min, and 8 L / min. The first two stages will pump for 5 minutes, and the third stage will pump until the cumulative liquid consumption reaches the pre-flush liquid volume.

[0152] Step 3: Design of the sand-carrying fluid stage.

[0153] The design of the proppant-carrying fluid stages is relatively complex. This example only uses a stepped continuous proppant addition system, focusing on controlling the proppant addition amount and stage proppant ratio. There are typically 5-10 proppant-carrying fluid stages. Here, we assume the proppant addition amount in each stage is an arithmetic sequence (or a geometric sequence, or a sequence that satisfies logarithmic fitting, or a sequence that satisfies square function fitting). With 8 proppant-carrying fluid stages, the total proppant addition amount in each stage is approximately equal to (with an error less than a threshold) the total proppant addition amount in the fracturing section. The corresponding stage proppant ratio sequence also satisfies certain fitting conditions, with the average stage proppant ratio approximately equal to the average proppant ratio during construction (%), and the maximum stage proppant ratio less than a preset value. Simultaneously, the cumulative fluid consumption at the end of each proppant-carrying fluid stage must be approximately equal to the total fluid consumption in the fracturing section. Further assuming a minimum stage proppant addition amount, a maximum stage proppant addition amount, or a minimum stage proppant ratio, linear programming can be used to solve for the proppant addition amount and corresponding stage proppant ratio for the 8 proppant-carrying fluid stages.

[0154] Step 4: Displacement fluid stage, the results of the pumping procedure are fed back to the fracturing design to generate an intelligent agent.

[0155] During the displacement fluid stage, pump the fluid from one to two wellbores at maximum capacity. This typically involves one to two stages.

[0156] To ensure fracturing design reports meet client requirements, the system supports visual previews and multiple output formats, such as Word, PDF, and HTML. Clients can preview, edit, and export online, ensuring compliance with industry technical specifications and on-site construction requirements.

[0157] Steps 204 to 206 constitute the process of automatically generating fracturing design reports.

[0158] Step 207: Client 101 sends a fracturing design modification request to server 102. This request includes, but is not limited to, format modifications, chapter modifications, and content updates. Users can input the fracturing design modification request using natural language.

[0159] Step 208: Server 102 receives the fracturing design modification request sent by the user through the client; calls the fracturing design modification agent, which uses the large language model to modify the fracturing design report according to the fracturing design modification request and the third prompt information.

[0160] The third prompt information includes third-party role information, related context information, chapter requirements information, and format requirements information.

[0161] The third role information is used to indicate the role of the expert writing fracturing designs in the oil industry.

[0162] The chapter requirements and formatting requirements are the same as those in the first prompt. The third prompt is more concise than the first and can improve the efficiency of revising fracturing design reports.

[0163] In some embodiments, the third prompt message is as follows: { Role Requirements: You are an expert in the petroleum industry responsible for writing fracturing designs. You need to make targeted and rigorous revisions to fracturing design documents based on the user's feedback. Workbench: @Currently generated fracturing design document; Context: @context system; Formatting requirements: @Formatting requirements; Chapter Requirements: @Chapter Requirements; Output: The revised fracturing design document. Please edit the modified parts in the workbench.

[0164] }

[0165] Steps 207 and 208 constitute the process of automating the modification of the fracturing design report. During implementation, to improve the accuracy of the fracturing design report, users can interact with the large language model multiple times and perform local modifications based on the context system.

[0166] Step 209: Client 101 sends the query request entered by the user to server 102.

[0167] In this step, the query request is a question related to fracturing design. During implementation, users can enter query information using natural language or text. There are two types of queries: quantitative queries and qualitative queries.

[0168] Quantitative queries include numerical queries, range statistics, and maximum / minimum value queries. For these types of queries, an SQL statement is generated to access a structured vector knowledge base to obtain the query results. Examples of quantitative query questions include: What is parameter b for well a? What is the list of well numbers from 12 to 15 for parameter d in block c? What is parameter e? What is the maximum value for the year?

[0169] Qualitative queries require analysis and answers based on textual descriptions. For this type of query, vector similarity retrieval technology is used to access a vector knowledge base to obtain the query results. A typical qualitative query question is: What are the stratigraphic properties of well A, and what are its characteristics?

[0170] Step 210: Server 102 calls the knowledge retrieval and intelligent question answering service to obtain the response information related to the query request from the vector knowledge base and structured database, and sends the response information to client 101.

[0171] Among them, knowledge retrieval and intelligent question answering services refer to obtaining response information by calling the fracturing vector knowledge base.

[0172] Steps 209 to 210 enable users to interact with the server in real time, improving the user experience.

[0173] When implementing on the server side, such as Figure 3 As shown, the server can achieve intelligent service coordination based on a multimodal intelligent agent architecture. The intelligent agent 300 includes the following components: a natural language understanding unit 301, a task scheduling unit 302, a dialogue management unit 303, and a result integration unit 304. The natural language understanding unit 301 includes an intent recognition model and an entity extraction model. Simultaneously, the intelligent agent 300 also integrates a parameter optimization service 305, an engineering calculation service 306, a knowledge retrieval and intelligent question answering service 307, a structured database 308, and a vector knowledge base 309. The parameter optimization service 305 includes a parameter optimization module. The engineering calculation service 306 includes an engineering calculation module. The knowledge retrieval and intelligent question answering service 307 can call the structured database 308 and the vector knowledge base 309.

[0174] The intelligent agent 300 receives fracturing design generation requests and query requests input by the user in natural language. It uses an intent recognition model to identify the user's input information, such as the purpose of the query (query, statistics, or performing a computational task). The entity extraction model extracts key entities from the user's input. For example, if the query request is for the amount of proppant added to wells b1 and b2 in block a, the intent recognition model first determines that the user's need is a query and breaks the question down into querying the amount of proppant added to well b1 in block a and querying the amount of proppant added to well b2 in block a. Then, the entity extraction module identifies three key entities in the question: block a, wells b1 / b2, and the amount of proppant added, and sends the intent information and key entities to the task scheduling unit 302.

[0175] The task scheduling unit 302 invokes relevant services to respond to user requests based on intent information.

[0176] When the intent information indicates a query, the task scheduling unit 302 invokes the knowledge retrieval and intelligent question-answering service 307. The knowledge retrieval and intelligent question-answering service 307 obtains the response information by calling the structured database 308 and the vector knowledge base 309. The task scheduling unit 302 receives the response information from the knowledge retrieval and intelligent question-answering service 307 and sends it back to the client through the dialogue management unit 303. The dialogue management unit 303 supports multi-turn dialogues, maintains the dialogue context state, understands referencing and ellipsis, and provides a continuous and coherent interactive experience.

[0177] When the intent information indicates that fracturing design is generated, the task scheduling unit 302 calls the parameter optimization service 305, the engineering calculation service 306, the structured database 308, and the vector knowledge base 309 to obtain fracturing optimization parameters and engineering calculation results; then it calls the result integration unit 304 to integrate the fracturing optimization parameters and engineering calculation results to obtain a fracturing design report, and sends the fracturing design report to the client.

[0178] In some embodiments, a method for intelligently generating fracturing designs is provided, such as Figure 4 As shown, it includes: Step 401: Receive a fracturing design generation request sent by the user through the client. The fracturing design generation request includes at least the design well information.

[0179] During implementation, the fracturing design generation request also includes at least one of the following design parameters: reference block information, fracturing method information, fracturing fluid information, proppant information, optimization target information, pumping requirements information, and fracturing parameter information.

[0180] During this step, the design parameters input by the user through the client configuration interface are received. The configuration interface has multiple parameter controls for users to input design parameters.

[0181] Step 402: Using a large language model, generate a fracturing design request and first prompt information based on the fracturing design, and generate a fracturing design report through an operation vector knowledge base, a structured database, and functional modules.

[0182] The first prompt information includes the first role information, the skill information implemented by the operation vector knowledge base, structured database and functional modules, the outline requirement information, the format requirement information, the chapter requirement information and the calling protocol information.

[0183] The outline requirements information is used to indicate the chapters in the fracturing design and the order in which the chapters are generated.

[0184] The format requirements information indicates the format requirements that must be followed when generating a fracturing design report.

[0185] The chapter requirement information is used to indicate at least one of the following: chapter content requirement information, chapter content reference information, and function module call information.

[0186] The calling protocol information is used to indicate the calling protocol information of the vector knowledge base, the structured database, and the functional modules.

[0187] The vector knowledge base stores professional knowledge descriptions of historical fracturing designs, the structured database stores key parameter information of historical fracturing designs, and the functional modules include fracturing optimization logic and engineering calculation logic.

[0188] In some embodiments, the functions include a parameter optimization module and an engineering calculation module.

[0189] The parameter optimization module is used to determine the optimal production index and optimal parameter combination based on the fracturing design generation request, chapter requirement information, vector knowledge base and structured database.

[0190] The engineering calculation module is used to perform engineering calculations based on fracturing design requests, chapter requirements, optimal parameter combinations, vector knowledge bases, and structured databases.

[0191] Step 403: Send the fracturing design report to the client.

[0192] This embodiment enables fully automated and intelligent fracturing design, significantly improving fracturing design efficiency, accuracy, and decision support capabilities.

[0193] In some embodiments, such as Figure 5 As shown, the intelligent generation method for fracturing design also includes: Step 501: Receive the fracturing design modification request sent by the user through the client.

[0194] Step 502: Modify the fracturing design report based on the fracturing design modification request and the third prompt information using the large language model.

[0195] The third prompt information includes third-party role information, related context information, chapter requirements information, and format requirements information.

[0196] In some embodiments, such as Figure 6 As shown, the intelligent generation method for fracturing design also includes: Step 601: Receive the query request sent by the user through the client.

[0197] Step 602: Invoke the knowledge retrieval and intelligent question answering module to obtain the response information related to the query request. The knowledge retrieval and intelligent question answering module includes interaction logic.

[0198] Step 603: Send a reply message to the client.

[0199] In some embodiments, the process of establishing a fracturing vector knowledge base includes preprocessing multi-source heterogeneous fracturing engineering documents of the target block to obtain standardized documents; using a large language model based on second prompt information to extract key parameters, relationships between parameters, and professional knowledge descriptions related to fracturing design from the standardized documents; establishing a structured database based on key parameters and relationships between parameters; and establishing a vector knowledge base based on professional knowledge descriptions related to fracturing design.

[0200] The second prompt information includes second role information, extraction task requirement information, fracturing domain knowledge dictionary and storage instruction information.

[0201] Among them, the multi-source heterogeneous fracturing engineering documents include fracturing engineering documents of the following professional types: geological design, fracturing design, fracturing construction, flowback monitoring, fracturing production, and rock mechanics experiments.

[0202] The fracturing knowledge dictionary includes parameter information for geological parameters, engineering parameters, construction parameters, and production parameters. Parameter information includes parameter identifier, unit of measurement, value range, and semantic description.

[0203] like Figure 7 As shown, the multi-source heterogeneous fracturing engineering documents for the target block are preprocessed to obtain standardized documents, including: Step 701: Preprocess the multi-source heterogeneous fracturing engineering documents of the target block to obtain the first document.

[0204] Step 702: Determine the format type of the first document.

[0205] Step 703: Parse the first document using a parsing algorithm related to the format type of the first document to obtain the second document.

[0206] Step 704: Convert the second document into a standardized document.

[0207] In some embodiments, the multi-source heterogeneous fracturing engineering documents for the target block are preprocessed to obtain standardized documents, including: Determine the subject type, well identifier, and block identifier of the first document; determine the index of standardized documents according to the preset naming rules; the preset naming rules include at least the well identifier, block identifier, and subject type; store the standardized documents in the document library according to the index.

[0208] Based on the same inventive concept, this application also provides a fracturing design intelligent generation device, as described in the following embodiments. Since the principle of the fracturing design intelligent generation device in solving the problem is similar to that of the fracturing design intelligent generation method, the implementation of the fracturing design intelligent generation device can refer to the fracturing design intelligent generation method, and repeated details will not be elaborated further.

[0209] Specifically, such as Figure 8 As shown, the intelligent fracturing design generation device includes a fracturing design generation intelligent agent 800, which is equipped with a large language model 804 and connected to a vector knowledge base 801, a structured database 802, and a functional module 803.

[0210] The fracturing design generator agent 800 is configured to perform the following operations: Receive fracturing design generation requests sent by users through the client; the fracturing design generation request includes at least design well information; The fracturing design report is generated by using the large language model 804 based on the fracturing design request and the first prompt information, and by using the operation vector knowledge base 801, the structured database 802 and the functional module 803. Send the fracturing design report to the client.

[0211] The first prompt information includes the first role information, the skill information implemented by the operation vector knowledge base, structured database and functional modules, the outline requirement information, the format requirement information, the chapter requirement information and the calling protocol information.

[0212] The outline requirements information is used to indicate the chapters in the fracturing design and the order in which the chapters are generated.

[0213] The format requirements information indicates the format requirements that must be followed when generating a fracturing design report.

[0214] The chapter requirement information is used to indicate at least one of the following: chapter content requirement information, chapter content reference information, and function module call information.

[0215] The calling protocol information is used to indicate the calling protocol information of the vector knowledge base, structured database, and functional modules.

[0216] The vector knowledge base stores professional knowledge descriptions of historical fracturing designs, the structured database stores key parameter information of historical fracturing designs, and the functional modules include fracturing optimization logic and engineering calculation logic.

[0217] The fracturing design generation agent first parses the fracturing design generation request and the initial prompt information to obtain the generation approach for each chapter. This approach includes: determining the format requirements for each chapter's content according to the chapter generation order, the operation methods of the vector knowledge base, structured database, and functional modules, and the data to be acquired. Then, the fracturing design generation agent automatically generates the content for each chapter according to the given generation approach.

[0218] This embodiment allows users to simply input a fracturing design generation request through the client; the server uses a large language model to generate a fracturing design report based on the fracturing design generation request and the first prompt information, through an operation vector knowledge base, a structured database, and functional modules; the fracturing design report is then sent to the client, enabling fully automated and intelligent fracturing design, significantly improving fracturing design efficiency, accuracy, and decision support capabilities.

[0219] In some embodiments, such as Figure 9As shown, the intelligent fracturing design generation device also includes a fracturing design modification agent 900. The fracturing design modification agent 900 is equipped with a large language model 804 and is connected to a vector knowledge base 801 and a structured database 802. The fracturing design modification agent 900 is configured to perform the following operations: Receive fracturing design modification requests sent by users through the client; Using the large language model 804, the fracturing design report is modified based on the fracturing design modification request and the third prompt information.

[0220] The third prompt information includes third-party role information, related context information, chapter requirements information, and format requirements information.

[0221] The fracturing design modification agent first determines the modification approach based on the fracturing design modification request and the third-party prompt information. The modification approach includes: identifying the chapter to be modified and the user's modification requirements; determining the modification method based on the user's modification requirements, which includes improving the chapter to be modified based on the context, and improving the chapter to be modified through an operation vector knowledge base and a structured database, etc. Then, the fracturing design modification agent improves the chapter to be modified according to the modification approach.

[0222] In one specific embodiment, in a fracturing design project for a shale gas field, the server received 15 geological design reports, 8 fracturing design schemes, 22 construction summary reports, 35 sets of flowback monitoring data, more than 500 production data records, and 5 rock mechanics test reports.

[0223] The server automatically classifies these multi-source heterogeneous documents and extracts 186 parameter items, including formation parameters (such as formation pressure coefficient, Young's modulus, Poisson's ratio, etc.), construction parameters (such as displacement, sand ratio, liquid volume, etc.), and production parameters (such as daily gas production, cumulative gas production, etc.). A structured database containing more than 20,000 records and a semantic vector knowledge base containing more than 5,000 vectors are established.

[0224] In some embodiments, the client provides a web-based user interface supporting concurrent access by multiple users. The server-side system is deployed using Docker containers and orchestrated and managed using Kubernetes. The server adopts a microservice architecture, with services communicating through RESTful APIs and message queues. The system provides standardized data interfaces, enabling seamless integration with existing geological modeling software, production management systems, and real-time monitoring platforms in the oilfield.

[0225] In practice, access control permissions can be set based on roles, with different users having different access rights. For example, role 1 may only have query permissions, while role 2 may have both fracturing design and generation permissions and query permissions. By setting roles, system security and data confidentiality can be ensured.

[0226] As can be seen from the technical solutions provided in the above embodiments of this application, this application can achieve the following technical effects: 1. Improved fracturing design efficiency: By automating the processing of multi-source heterogeneous data and intelligently generating design schemes, the design work that traditionally takes several days to complete is shortened to the hour level, greatly improving design efficiency.

[0227] 2. Improved design quality: Parameter optimization and engineering calculation based on parameter optimization and engineering calculation services avoid design deviations caused by human factors, thereby improving the accuracy and reliability of the design scheme.

[0228] 3. Achieved efficient utilization of knowledge: By constructing a vector knowledge base and a structured database, the effective management and utilization of numerical data and descriptive knowledge were realized.

[0229] 4. Provides an intelligent interactive experience: Natural language interaction is achieved through intelligent agent technology, enabling non-professionals to quickly obtain professional fracturing design services.

[0230] In some embodiments of this application, a computer device is also provided, such as... Figure 10 As shown, computer device 1002 may include one or more processors 1004, such as one or more central processing units (CPUs), each of which may implement one or more hardware threads. Computer device 1002 may also include any memory 1006 for storing information of any kind, such as code, settings, data, etc. Non-limitingly, for example, memory 1006 may include any type of RAM, any type of ROM, flash memory, hard disk, optical disk, etc. More generally, any memory can use any technology to store information. Furthermore, any memory may provide volatile or non-volatile retention of information. Furthermore, any memory may represent a fixed or removable component of computer device 1002. In one case, when processor 1004 executes associated instructions stored in any memory or combination of memories, computer device 1002 may perform any operation of the associated instructions. Computer device 1002 also includes one or more drive mechanisms 1008 for interacting with any memory, such as hard disk drive mechanisms, optical disk drive mechanisms, etc.

[0231] Computer device 1002 may further include an input / output module 1010 (I / O) for receiving various inputs (via input device 1012) and providing various outputs (via output device 1014). A specific output mechanism may include a presentation device 1016 and an associated graphical user interface 1018 (GUI). In other embodiments, the input / output module 1010 (I / O), input device 1012, and output device 1014 may be omitted, and the device may function solely as a computer device within a network. Computer device 1002 may also include one or more network interfaces 1020 for exchanging data with other devices via one or more communication links 1022. One or more communication buses 1024 couple the components described above together.

[0232] The communication link 1022 can be implemented in any way, such as via a local area network, a wide area network (e.g., the Internet), a point-to-point connection, or any combination thereof. The communication link 1022 may include any combination of hardwired links, wireless links, routers, gateway functions, name servers, etc., governed by any protocol or combination of protocols.

[0233] This application also provides a computer-readable storage medium, such as a non-transient computer-readable storage medium, on which a computer program is stored, and which, when run by a processor, performs the steps of the above-described method.

[0234] This application also provides a computer-readable instruction, wherein when a processor executes the instruction, the program therein causes the processor to perform the method shown in the above embodiments.

[0235] It should be understood that in the various embodiments of this application, the order of the above-mentioned processes does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0236] It should also be understood that, in the embodiments of this application, the term "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Additionally, the character " / " in this application generally indicates that the preceding and following related objects have an "or" relationship.

[0237] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed in this application can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0238] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0239] In the embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the couplings or direct couplings or communication connections shown or discussed may be indirect couplings or communication connections through some interfaces, apparatuses, or units, or they may be electrical, mechanical, or other forms of connection.

[0240] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of the embodiments of this application, depending on actual needs.

[0241] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0242] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0243] This application uses specific embodiments to illustrate the principles and implementation methods of this application. The description of the above embodiments is only for the purpose of helping to understand the method and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for intelligently generating fracturing designs, characterized in that, include: Receive fracturing design generation requests sent by users through the client; The fracturing design generation request includes at least the design well information; Using a large language model, a fracturing design request and first prompt information are generated based on the fracturing design. Through an operation vector knowledge base, a structured database, and functional modules, a fracturing design report is generated. Send the fracturing design report to the client; The first prompt information includes first role information, skill information implemented by operating the vector knowledge base, the structured database and the functional modules, outline requirement information, format requirement information, chapter requirement information and calling protocol information; The outline requirements information is used to indicate the chapters in the fracturing design and the order in which the chapters are generated. The format requirements information is used to indicate the format requirements that must be followed when generating a fracturing design report; The chapter requirement information is used to indicate at least one of the following: chapter content requirement information, chapter content reference information, and function module call information; The invocation protocol information is used to indicate the invocation protocol information of the vector knowledge base, the structured database, and the functional modules; The vector knowledge base stores professional knowledge description information of historical fracturing designs, the structured database stores key parameter information of historical fracturing designs, and the functional modules include fracturing optimization logic and engineering calculation logic. The process of establishing the vector knowledge base and the structured database includes: The multi-source heterogeneous fracturing engineering documents for the target block are preprocessed to obtain standardized documents; Using the large language model and based on the second prompt information, key parameters, the relationships between parameters, and professional knowledge descriptions related to fracturing design are extracted from the standardized document. A structured database is established based on the key parameters and the relationships between them; Based on the professional knowledge description information related to fracturing design, a vector knowledge base is established; The second prompt information includes second role information, extraction task requirement information, fracturing domain knowledge dictionary and storage instruction information; The preprocessing of multi-source heterogeneous fracturing engineering documents for the target block yields standardized documents, including: The multi-source heterogeneous fracturing engineering documents for the target block are preprocessed to obtain the first document; Determine the format type of the first document; The first document is parsed using a parsing algorithm related to its format type to obtain the second document; Convert the second document into a standardized document; Determine the professional type, well identifier, and block identifier of the first document; The index of the standardized document is determined according to a preset naming rule; the preset naming rule includes at least a well identifier, a block identifier, and a professional type. The standardized documents are stored in the document library according to the index.

2. The method as described in claim 1, characterized in that, The functional modules include a parameter optimization module and an engineering calculation module; The parameter optimization module is used to determine the optimal production index and the optimal parameter combination based on the fracturing design generation request, the chapter requirement information, the vector knowledge base, and the structured database. The engineering calculation module is used to perform engineering calculations based on the fracturing design generation request, the chapter requirement information, the optimal parameter combination, the vector knowledge base, and the structured database.

3. The method as described in claim 1, characterized in that, The multi-source heterogeneous fracturing engineering documents include fracturing engineering documents of the following professional types: Geological design, fracturing design, fracturing construction, flowback monitoring, fracturing production, and rock mechanics experiments.

4. The method as described in claim 1, characterized in that, The fracturing knowledge dictionary includes parameter information for geological parameters, engineering parameters, construction parameters, and production parameters. The parameter information includes: parameter identifier, unit of measurement, value range, and semantic description information.

5. The method as described in claim 1, characterized in that, The skill information implemented by operating the vector knowledge base, the structured database, and the functional modules includes: By querying the vector knowledge base, professional knowledge description information of the design well can be obtained; The key parameter information of the design well is obtained by querying the structured database; By comparing the data of neighboring wells related to the design well in the vector knowledge base, the geological description and construction suggestions of the design well are obtained; By calling the aforementioned functional modules, optimization parameters and engineering calculation results can be obtained.

6. The method as described in claim 1, characterized in that, The fracturing design generation request also includes at least one of the following design parameters: Reference block information, fracturing method information, fracturing fluid information, proppant information, optimization target information, pumping requirements information, and fracturing parameter information.

7. The method as described in claim 6, characterized in that, Receive fracturing design generation requests sent by users through the client, including: Receive design parameters input by the user through the client configuration interface; The configuration interface has multiple parameter controls, which are used for users to input design parameters.

8. The method as described in claim 1, characterized in that, After generating the fracturing design report, it also includes: Receive fracturing design modification requests sent by users through the client; The fracturing design report is modified using a large language model based on the fracturing design modification request and the third prompt information; The third prompt information includes third role information, associated context information, chapter requirement information, and format requirement information.

9. The method as described in claim 1, characterized in that, Also includes: Receive query requests sent by users through the client; The knowledge retrieval and intelligent question answering modules are invoked to obtain the response information related to the query request; The knowledge retrieval and intelligent question answering module includes interactive logic; The reply information is sent to the client.

10. A smart fracturing design generation device, characterized in that, include: A fracturing design and generation agent is configured with a large language model and connected to a vector knowledge base, a structured database, and functional modules. The fracturing design and generation agent is configured to perform the following operations: Receives a fracturing design generation request sent by a user through a client; the fracturing design generation request includes at least design well information; Using a large language model, a fracturing design request and first prompt information are generated based on the fracturing design. Through an operation vector knowledge base, a structured database, and functional modules, a fracturing design report is generated. Send the fracturing design report to the client; The first prompt information includes first role information, skill information implemented by operating the vector knowledge base, the structured database and the functional modules, outline requirement information, format requirement information, chapter requirement information and calling protocol information; The outline requirements information is used to indicate the chapters in the fracturing design and the order in which the chapters are generated. The format requirements information is used to indicate the format requirements that must be followed when generating a fracturing design report; The chapter requirement information is used to indicate at least one of the following: chapter content requirement information, chapter content reference information, and function module call information; The invocation protocol information is used to indicate the invocation protocol information of the vector knowledge base, the structured database, and the functional modules; The vector knowledge base stores professional knowledge description information of historical fracturing designs, the structured database stores key parameter information of historical fracturing designs, and the functional modules include fracturing optimization logic and engineering calculation logic. The process of establishing the vector knowledge base and the structured database includes: The multi-source heterogeneous fracturing engineering documents for the target block are preprocessed to obtain standardized documents; Using the large language model and based on the second prompt information, key parameters, the relationships between parameters, and professional knowledge descriptions related to fracturing design are extracted from the standardized document. A structured database is established based on the key parameters and the relationships between them; Based on the professional knowledge description information related to fracturing design, a vector knowledge base is established; The second prompt information includes second role information, extraction task requirement information, fracturing domain knowledge dictionary and storage instruction information; The preprocessing of multi-source heterogeneous fracturing engineering documents for the target block yields standardized documents, including: The multi-source heterogeneous fracturing engineering documents for the target block are preprocessed to obtain the first document; Determine the format type of the first document; The first document is parsed using a parsing algorithm related to its format type to obtain the second document; Convert the second document into a standardized document; Determine the professional type, well identifier, and block identifier of the first document; The index of the standardized document is determined according to a preset naming rule; the preset naming rule includes at least a well identifier, a block identifier, and a professional type. The standardized documents are stored in the document library according to the index.

11. The apparatus as claimed in claim 10, characterized in that, Also includes: A fracturing design modification agent, configured with a large language model and connected to a vector knowledge base and a structured database, is configured to perform the following operations: Receive fracturing design modification requests sent by users through the client; The fracturing design report is modified using a large language model based on the fracturing design modification request and the third prompt information; The third prompt information includes third role information, associated context information, chapter requirement information, and format requirement information.

12. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method according to any one of claims 1 to 9.

13. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor of the computer device, it implements the method of any one of claims 1 to 9.

14. A computer program product, the computer program product comprising a computer program, characterized in that, When the computer program is executed by the processor of the computer device, it implements the method of any one of claims 1 to 9.

Images