Modular geological oil and gas intelligent investigation system and scheduling method

Abstract

The application provides a modularized intelligent geological oil and gas investigation system and a scheduling method.The investigation system comprises a core intelligent agent and a professional MCP Server cluster.The core intelligent agent is responsible for task analysis, module scheduling, result integration and exception handling, and its core components include a task analyzer, a module scheduler, a result integrator, an exception handler and a state monitor.The professional MCP Server cluster uses a standardized model calling protocol to realize communication and data exchange between the core intelligent agent and each professional Server.The professional MCP Server cluster comprises a regional information extraction Server, an oil and gas feature identification Server, a cross-document association reasoning Server, a data reliability evaluation Server, a space-time dimension classification Server, a reservoir forming element reasoning Server and a dynamic potential evaluation Server.The application has the capabilities of "modularized architecture, standardized interface, flexible scheduling and resource optimization", and can realize efficient construction, flexible expansion and intelligent scheduling of the modularized intelligent geological oil and gas investigation system.

Description

Technical Field

[0001] This invention relates to the field of interdisciplinary technology of geological oil and gas exploration technology and artificial intelligence, specifically to a modular intelligent geological oil and gas survey system and scheduling method. Background Technology

[0002] In the field of geological oil and gas resource surveys, traditional exploration data processing and analysis workflows suffer from the following core pain points:

[0003] (1) High system coupling

[0004] Existing systems are mostly monolithic architectures with tightly coupled functional modules, making them difficult to upgrade or replace independently;

[0005] (2) Difficulty in technological iteration

[0006] When new technologies (such as new AI models or data processing algorithms) need to be introduced, the entire system needs to be reconstructed, which is costly and time-consuming.

[0007] (3) Unreasonable allocation of resources

[0008] Different functional modules have vastly different requirements for hardware resources (CPU / GPU / memory), making it difficult for a monolithic architecture to achieve fine-grained resource allocation;

[0009] (4) Poor scalability

[0010] It is difficult to flexibly add or remove functional modules according to business needs, and new survey tasks often require the development of new dedicated systems;

[0011] (5) High maintenance costs

[0012] Problems in any part of the system can affect the overall operation, making maintenance difficult and costly.

[0013] To address the aforementioned industry pain points, there is an urgent need for a technical solution with "modular architecture, standardized interfaces, flexible scheduling, and resource optimization" capabilities to achieve efficient construction, flexible expansion, and intelligent scheduling of modular geological oil and gas intelligent survey systems. Summary of the Invention

[0014] To address the technical problems existing in the background technology mentioned above, this invention proposes a modular intelligent geological oil and gas survey system and scheduling method, which has the capabilities of "modular architecture, standardized interface, flexible scheduling, and resource optimization", realizing the efficient construction, flexible expansion and intelligent scheduling of the modular intelligent geological oil and gas survey system, and realizing the automation, intelligence and efficient scheduling of geological oil and gas surveys.

[0015] To address the aforementioned technical problems, this invention provides a modular intelligent geological oil and gas survey system, which includes a core intelligent agent, a specialized MCP Server cluster, and a data storage layer.

[0016] The core intelligent agent is responsible for task parsing, module scheduling, result integration and exception handling. Its core components include a task parser, a module scheduler, a result integrator, an exception handler and a status monitor.

[0017] The task parser is used to parse user requests into executable subtasks;

[0018] The module scheduler is used to intelligently schedule MCP Servers based on task type and system status;

[0019] The result integrator is used to integrate the outputs of each MCP Server into a final result;

[0020] The exception handler is used to handle abnormal situations during execution to ensure stable system operation;

[0021] The status monitor is used to monitor the operating status and resource usage of each module of the system;

[0022] The specialized MCP Server cluster utilizes a standardized model invocation protocol to enable communication and data exchange between the core intelligent agent and each specialized server. The specialized MCP Server cluster includes a regional information extraction server, an oil and gas feature recognition server, a cross-document association reasoning server, a data reliability assessment server, a spatiotemporal dimension classification server, an oil and gas accumulation element reasoning server, and a dynamic potential assessment server.

[0023] The regional information extraction server is used to accurately extract text or charts from a specified region;

[0024] The oil and gas feature recognition server is used to filter oil and gas related information and extract core features;

[0025] The cross-document association reasoning server is used to associate complementary data from multiple documents to fill in information gaps.

[0026] The data reliability assessment server is used to detect contradictions in the multi-document complementary data output by the cross-document association reasoning server and verify its credibility.

[0027] The spatiotemporal dimension classification server is used to integrate the associated data output by the regional information extraction server, the oil and gas feature recognition server, and the cross-document association reasoning server according to the three dimensions of "time-method-element".

[0028] The hydrocarbon accumulation element inference server is used to calculate the hydrocarbon accumulation matching degree based on a knowledge graph weighted calculation.

[0029] The dynamic potential assessment server is used to combine the credibility results output by the data reliability assessment server with the reservoir matching degree data of the reservoir element inference server to output the resource potential range.

[0030] The data storage layer is used to store various types of data during system operation, including a document database, a knowledge graph library, and a results database. The document database stores basic data such as original geological and oil and gas survey documents and charts. The knowledge graph library stores structured knowledge data such as knowledge related to oil and gas accumulation and association rules. The results database stores the intermediate execution results of each MCPServer and the final output results of the system.

[0031] A modular intelligent geological oil and gas scheduling method, based on the aforementioned modular intelligent geological oil and gas survey system, includes the following steps:

[0032] (1) Establish a standardized interface and data transmission format between the core intelligent agent and the professional MCP Server cluster based on the MCP protocol:

[0033] (1.1) Interface Definition

[0034] The MCP protocol defines a standardized interface between the core intelligent agent and the specialized MCP Server cluster; the specific process is as follows:

[0035] (1.2) Construction of data transmission format

[0036] The data transmission format is the transmission format of task request and response data between the core intelligent agent and the professional MCP Server cluster. The request and response format is JSON, the encoding method is UTF-8, and the compression algorithm is gzip, which is used for large data transmission.

[0037] Data type definition: Includes three categories: text data, image data, and structured data; among them, text data is a string array containing content, source, and confidence level; image data is an object array containing access address, description, and metadata; structured data is an object array containing entity, attribute, and confidence level.

[0038] Error code definitions: 200 indicates success, 400 indicates incorrect request parameters, 401 indicates unauthorized, 404 indicates resource does not exist, 408 indicates request timeout, 500 indicates internal server error, and 503 indicates service unavailable.

[0039] (2) Execute a modular scheduling process in the survey system:

[0040] (2.1) Geological oil and gas survey task parsing and decomposition, that is, the core intelligent agent receives the geological oil and gas survey requirements in natural language form input by the user, and decomposes them into executable sub-tasks through the task parser;

[0041] (2.2) Formulation and execution of intelligent scheduling strategy for geological oil and gas, that is, based on the dependency relationship between executable sub-tasks, the resource configuration and real-time load of each MCP Server, the execution order and target Server of each executable sub-task are determined to achieve optimized resource allocation and load balancing;

[0042] (2.3) The application of the geological oil and gas scheduling algorithm and the dynamic expansion mechanism, that is, the candidate task allocation scheme is iteratively optimized based on the genetic algorithm to obtain a task scheduling scheme that meets the preset fitness target;

[0043] (2.4) The application of dynamic expansion mechanism, that is, the number of MCPServer instances is automatically adjusted according to the workload and server running status, and the online access and exit of the server is supported to realize the dynamic expansion of the system.

[0044] As a preferred embodiment of the present invention: the standardized interfaces in step (1.1) include a service registration interface, a task submission interface, a task status query interface, a result acquisition interface, and a service status update interface;

[0045] The service registration interface is used by the specialized MCP Server to submit service information to the core intelligent agent, including server identifier, function description, input and output data format and resource requirements, to complete the service access registration;

[0046] The task submission interface is used by the core intelligent agent to submit subtasks to the registered specialized MCP Server, including task identifier, target server identifier, task parameters, priority and timeout threshold;

[0047] The task status query interface is used by the core intelligent agent to query the execution status of submitted subtasks, including four statuses: pending execution, processing, completed, and failed, as well as corresponding progress, result preview, or error information.

[0048] The result acquisition interface is used by the core intelligent agent to obtain the final execution result of the completed sub-task, including task identifier, execution server identifier, result data, timestamp and execution time;

[0049] The service status update interface is used by the specialized MCP Server to synchronize its own operating status and real-time resource utilization with the core intelligent agent.

[0050] As a preferred embodiment of the present invention, the specific process of task parsing and decomposition in step (2.1) is as follows:

[0051] (2.1.1) Receive natural language input from the user through the core intelligent agent;

[0052] (2.1.2) The natural language requirements are decomposed into a series of sub-tasks by the task parser. These sub-tasks include regional information extraction, oil and gas feature recognition, cross-document association reasoning, data reliability assessment, spatiotemporal dimension classification, hydrocarbon accumulation element reasoning, and dynamic potential assessment.

[0053] As a preferred embodiment of the present invention, the regional information extraction is based on the geographical range keywords in the user's requirements, and accurately locates and extracts relevant information such as geological structure and stratigraphic data of the region from the original survey documents and charts;

[0054] The oil and gas feature identification involves filtering the regional information extraction results, retaining effective information related to oil and gas, and extracting core feature parameters such as source rock thickness, reservoir porosity, and oil saturation.

[0055] The cross-document association reasoning involves retrieving multiple relevant survey documents, associating complementary data, filling information gaps in a single document, and forming a complete regional oil and gas related dataset.

[0056] The data reliability assessment involves comparing datasets obtained through cross-document association reasoning, detecting data inconsistencies, and verifying data validity through a credibility algorithm.

[0057] The spatiotemporal dimension classification is based on three dimensions: "time-method-element". It classifies and integrates valid data that has undergone reliability assessment and establishes a spatiotemporal correlation index.

[0058] The reasoning of the hydrocarbon accumulation elements involves calling the hydrocarbon accumulation rules in the knowledge graph library, performing weighted calculations on the spatiotemporally classified dataset, and obtaining the regional hydrocarbon accumulation matching degree.

[0059] The dynamic potential assessment combines data reliability assessment results with reservoir matching data to construct a potential assessment model and output oil and gas resource potential ranges.

[0060] In a preferred embodiment of the present invention, the formulation and execution of the intelligent geological oil and gas scheduling strategy in step (2.2) is achieved through a modular scheduler based on the following factors:

[0061] (2.2.1) Dependencies of Geological Oil and Gas Survey Subtasks

[0062] The module scheduler determines the execution order of each subtask based on the logical dependencies between subtasks obtained by the task resolver;

[0063] (2.2.2) Optimal allocation of resources

[0064] CPU-intensive tasks are assigned to CPU servers;

[0065] GPU-intensive tasks are assigned to GPU servers;

[0066] Memory-intensive tasks are assigned to servers with large memory.

[0067] (2.2.3) Load balancing

[0068] Monitor the load of each MCP Server and assign new tasks to servers with lower load.

[0069] (2.2.4) Priority scheduling

[0070] Supports setting task priority, with higher priority tasks being executed first;

[0071] (2.2.5) Fault Tolerance Mechanism

[0072] When a task fails, it will be automatically retried or reassigned to another available MCP Server.

[0073] As a preferred embodiment of the present invention, the logical dependencies between the subtasks include the output of the preceding subtask serving as the input of the following subtask, the outputs of multiple subtasks jointly supporting the following subtask, and the ability of non-dependent subtasks to be executed in parallel.

[0074] As a preferred embodiment of the present invention, the geological oil and gas scheduling algorithm in step (2.3) is implemented by using an improved genetic algorithm to achieve optimized scheduling, and the specific implementation method is as follows:

[0075] (2.3.1) Chromosome coding

[0076] The task allocation scheme is encoded as chromosomes, with each gene representing a task assigned to a specific MCPServer;

[0077] (2.3.2) Construction of fitness function

[0078] Fitness=α×(1-LoadBalance)+β×(1-ResourceWaste)+γ×(1-ExecutionTime);

[0079] In the above formula, LoadBalance is the load balance degree; ResourceWaste is the resource waste rate; ExecutionTime is the normalized execution time; α, β, and γ are all weight coefficients that satisfy α+β+γ=1, and are used to control the weights of the influence of load balance degree, resource waste rate, and execution time on fitness, respectively.

[0080] (2.3.3) Selection Operation

[0081] Selecting superior individuals based on Fitness scores;

[0082] (2.3.4) Crossover operation

[0083] By exchanging some genes between two superior individuals, a new allocation scheme can be generated;

[0084] (2.3.5) Mutation operation

[0085] Randomly altering some genes in superior individuals increases population diversity;

[0086] (2.3.6) Iterative optimization

[0087] The process follows a multi-generational selection-crossover-mutation cycle, which includes repeating steps (2.3.4) to (2.3.5). Specifically, each generation includes one selection, one crossover, and one mutation operation. The iteration stops when either of these conditions is met: ① the average fitness value change rate of the population over three consecutive generations is less than 0.01; or ② the number of iterations reaches a preset maximum value. The iteration stops when either condition is met, and the individual with the highest fitness value is output as the optimal task allocation scheme.

[0088] As a preferred embodiment of the present invention, the dynamic expansion mechanism in step (2.4) includes:

[0089] (2.4.1) Automatic expansion and contraction

[0090] The number of MCP Server instances can be automatically increased or decreased based on the workload.

[0091] (2.4.2) Module hot-swappable

[0092] Supports dynamically adding or removing MCP Servers during the operation of the modular intelligent geological oil and gas survey system without restarting the system;

[0093] (2.4.3) Version Management

[0094] It supports the coexistence of multiple versions of the same MCP Server, enabling smooth upgrades.

[0095] As a preferred embodiment of the present invention, the specific process of step (2.4.2) is as follows:

[0096] ① Add Server: Automatic access is completed through the service registration interface of the MCP protocol, and the core intelligent agent updates the service registry in real time;

[0097] ② Remove Server: First, migrate any unfinished tasks on the server to other servers of the same type, then mark its status as "offline" and stop assigning new tasks.

[0098] By adopting the above technical solution, the present invention has the following beneficial effects:

[0099] The modular intelligent geological oil and gas survey system and scheduling method of this invention are reasonably conceived and have the capabilities of "modular architecture, standardized interface, flexible scheduling, and resource optimization". It realizes the efficient construction, flexible expansion and intelligent scheduling of the modular intelligent geological oil and gas survey system, and can realize the automation, intelligence and efficient scheduling of geological oil and gas surveys.

[0100] The technical effects of this invention are mainly reflected in the following aspects:

[0101] Architecture flexibility: It adopts a "1+7" modular architecture, and each module can be developed, deployed, upgraded and replaced independently, which greatly improves the system's flexibility and maintainability;

[0102] Resource optimization and allocation: Based on task type and resource requirements, fine-grained allocation of hardware resources is achieved, increasing resource utilization from 45-60% in traditional architecture to 75-85%;

[0103] Significant performance improvement: Processing time has been reduced from 15 days in the traditional monolithic architecture to 3 days, with an efficiency improvement of 80%;

[0104] Highly scalable: Supports horizontal scaling (increasing the number of servers) and vertical scaling (increasing the configuration of a single server), allowing for flexible expansion according to business needs;

[0105] Standardized Interface: The MCP protocol defines standardized interfaces and data formats, which facilitates system integration and access by third-party modules.

[0106] This invention can produce the following economic benefits:

[0107] Development cost savings: Module reusability results in development cost savings of ≥65%, and the development cycle of new systems is shortened by more than 65%;

[0108] Hardware cost optimization: Configure hardware resources on demand to avoid over-investment and reduce hardware costs by 30-40%;

[0109] Reduced maintenance costs: The modular architecture enables more precise problem localization and repair, reducing maintenance costs by more than 50%.

[0110] Labor cost savings: Automated processing reduces human intervention, resulting in labor cost savings of 40-60%;

[0111] Improved ROI: Shorter system development cycle, faster deployment, and a 30-50% increase in ROI.

[0112] This invention also has the following industry value:

[0113] Promoting intelligent exploration: It provides a standardized and modular intelligent oil and gas survey system architecture to promote the process of intelligent exploration;

[0114] Promoting technological innovation: The open MCP interface and modular architecture facilitate the introduction of new technologies (such as AI models and big data algorithms), thus promoting technological innovation;

[0115] Standardized industry applications: Defines standardized processes and interfaces for intelligent oil and gas surveys, improving the consistency and comparability of industry applications;

[0116] Knowledge Accumulation and Reuse: The professional MCP Server enables modular accumulation and reuse of domain knowledge, avoiding repetitive work;

[0117] Talent development and collaboration: Modular architecture facilitates team division of labor and collaboration, as well as the training of professional talents, thereby improving the overall technical level of the industry.

[0118] Compared with the prior art, the present invention has the following features or advantages:

[0119] A modular architecture of "1 core intelligent agent + 7 specialized MCP Servers" was proposed, which enabled the flexible construction and efficient operation of the modular geological oil and gas intelligent survey system.

[0120] A standardized MCP protocol was established, defining a unified interface and data format, which solved the communication and integration problems between different modules;

[0121] A genetic algorithm-based intelligent scheduling method was developed, which achieved optimized task allocation and efficient resource utilization.

[0122] Enables hot-swapping and dynamic expansion of modules, allowing modules to be added / removed during system runtime without requiring a system restart;

[0123] Build a reusable, specialized MCP Server library to support the rapid construction of different types of oil and gas survey systems, saving ≥65% in development costs.

[0124] This invention also has the following features:

[0125] (1) End-to-end intelligent chain for hydrocarbon accumulation mechanism

[0126] This invention is not a general computing scheduling platform, but rather a business link that connects "regional information extraction → oil and gas feature identification → cross-document association → data reliability assessment → spatiotemporal classification → reservoir element reasoning → dynamic potential assessment" into an end-to-end business chain around the entire process of oil and gas accumulation.

[0127] Compared with existing systems that only perform general computing power scheduling or single-point AI model calls, this invention can directly output "oil and gas resource potential range" and "key control elements", with a complete technology chain and stronger targeting.

[0128] (2) Credibility and traceability mechanism for complex geological data

[0129] This invention, through a combination of "cross-document association reasoning + data reliability assessment", not only integrates data from multiple sources, but also marks and calculates the credibility of contradictory data; finally, it explicitly introduces credibility weights in the potential assessment, so that the output results have the characteristics of "traceable source and quantifiable confidence".

[0130] Compared with traditional methods that rely solely on simple averaging or subjective manual screening, this invention reduces the risk of "false anomalies" and "false favorable regions," significantly improving the reliability and interpretability of the results.

[0131] (3) Spatiotemporal three-dimensional classification + knowledge graph reasoning, supporting dynamic updates and rolling evaluation.

[0132] This invention introduces a three-dimensional classification of "time-method-elements" and a knowledge graph reasoning mechanism, with all data and conclusions attached to specific time slices, exploration methods, and hydrocarbon accumulation elements.

[0133] When new seismic, well logging, or drilling data are added later, only the data of the corresponding time slices and elements need to be updated locally, which can automatically trigger the "rolling update" of hydrocarbon accumulation matching degree and resource potential. There is no need to manually redo the interpretation from scratch, which improves the efficiency and consistency of dynamic evaluation.

[0134] (4) Domain-specific scheduling algorithm for heterogeneous computing power clusters

[0135] Although genetic algorithm scheduling has been applied in other fields, this invention considers "load balancing, resource waste rate and execution time" in the fitness function, and combines the characteristics of oil and gas survey tasks (such as different computing power requirements of text processing, image recognition and knowledge graph reasoning) to differentiate the allocation of CPU-intensive, GPU-intensive and memory-intensive tasks.

[0136] Compared to simple polling or scheduling strategies based on a single metric (such as CPU utilization), this invention can achieve higher cluster efficiency and more stable task completion time in oil and gas exploration scenarios.

[0137] (5) Cross-basin / cross-business reuse value at the project level

[0138] This invention abstracts various specialized servers into a reusable module library. Once deployed and verified in a basin (such as the Tarim Basin), new systems can be quickly built in other basins or at different exploration stages (general survey, preliminary exploration, detailed exploration, and development dynamic monitoring) simply by reusing the original modules and adding a small number of special modules.

[0139] Compared to the existing model of "building a system from scratch" for each project, this invention can create a pattern of "a unified, standardized intelligent survey platform + rapid reuse across multiple projects" at the company level, further amplifying the benefits of early-stage R&D investment. Attached Figure Description

[0140] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0141] Figure 1 This is an architecture diagram of the modular intelligent geological oil and gas survey system of the present invention. Detailed Implementation

[0142] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0143] The present invention will be further explained below with reference to specific embodiments.

[0144] like Figure 1 As shown in the figure, this embodiment provides a modular intelligent geological oil and gas survey system, which includes a core intelligent agent, a professional MCP (Model Call Protocol) server cluster, and a data storage layer.

[0145] The core intelligent agent is responsible for task parsing, module scheduling, result integration, and exception handling. Its core components include a task parser, a module scheduler, a result integrator, an exception handler, and a status monitor.

[0146] This task parser is used to parse user requirements into executable subtasks.

[0147] This module scheduler is used to intelligently schedule MCP Servers based on task type and system status.

[0148] This results integrator is used to combine the outputs of each MCP Server into a final result.

[0149] This exception handler is used to handle abnormal situations during execution, ensuring stable system operation.

[0150] This status monitor is used to monitor the operating status and resource usage of each module in the system.

[0151] This specialized MCP Server cluster utilizes a standardized model invocation protocol to enable communication and data exchange between the core agent and each specialized server.

[0152] The regional information extraction server is used to accurately extract text or charts from a specified region; specifically, it is used to locate the specified region and extract relevant information.

[0153] This oil and gas feature recognition server is used to filter oil and gas related information and extract core features.

[0154] This cross-document association reasoning server is used to associate complementary data from multiple documents to fill in information gaps.

[0155] This data reliability assessment server is used to detect inconsistencies in the "multi-document complementary data" output by the cross-document association inference server (such as conflicts in reservoir thickness data between different documents in the same region), and to verify the validity of the data through a credibility algorithm (such as a weighted score based on the authority of the data source and the sample size).

[0156] This spatiotemporal dimension classification server is used to integrate the regional basic data extracted by the regional information extraction server, the core feature data output by the oil and gas feature identification server, and the complementary data of the cross-document association reasoning server according to the three dimensions of "time-method-element" to establish a spatiotemporal index (such as "2020-seismic exploration-reservoir porosity", "2022-well logging-oil saturation").

[0157] This hydrocarbon accumulation element inference server is used to calculate hydrocarbon accumulation matching degree based on knowledge graph weighting.

[0158] This dynamic potential assessment server combines the credibility results output by the data reliability assessment server with the hydrocarbon accumulation matching data calculated by the hydrocarbon accumulation factor inference server, and outputs the oil and gas resource potential range through potential assessment models (such as multiple linear regression and random forest).

[0159] This data storage layer is used to store various types of data during system operation, including a document database, a knowledge graph database, and a results database;

[0160] Among them, the document database stores unstructured / semi-structured basic data such as original geological oil and gas survey documents (such as exploration reports and well logging data text) and charts (such as geological profile maps and reservoir distribution maps);

[0161] Knowledge graph database: Stores structured knowledge in the field of hydrocarbon accumulation (such as source rock-reservoir-caprock association rules, geological age and tectonic evolution relationship), entity attributes and association data;

[0162] Results Database: Stores intermediate execution results of each specialized MCP Server (such as regional information extraction results and oil and gas characteristic parameters) as well as the final output of the system, including survey reports and potential assessment results.

[0163] The modular intelligent geological oil and gas scheduling method of the present invention includes the following steps:

[0164] S100. Establish a standardized interface and data transmission format between the core intelligent agent and the professional MCP Server cluster based on the MCP protocol:

[0165] S110, Interface Definition

[0166] The MCP protocol defines a standardized interface between the core intelligent agent and the specialized MCP Server cluster, comprising five types of interfaces. The functions, request methods, and core roles of each interface are as follows, achieving a closed-loop communication process of "service registration - task allocation - status tracking - result collection - status synchronization":

[0167] ① Service registration interface

[0168] Core function: Enables professional MCP Servers to "access and register" with core intelligent agents, allowing core intelligent agents to identify the Server's functions, resource requirements, and data format compatibility;

[0169] Request method: POST (actively submit registration information to the core agent);

[0170] Key parameters include: Server unique identifier (server_id), functionality, input and output data format (input_schema / output_schema), hardware resource requirements (such as number of CPU cores, memory size, whether GPU is required), and initial running status (online / offline).

[0171] Practical function: After receiving the registration information, the core intelligent agent updates the "service registry". During subsequent scheduling, it can accurately match "task requirements" with "server capabilities" (e.g., GPU-intensive tasks are only assigned to servers that are marked "requires GPU" during registration).

[0172] ②Task Submission Interface

[0173] Core function: The core intelligent agent assigns subtasks to registered specialized MCP Servers and transmits task parameters and execution requirements;

[0174] Request method: POST (The core agent pushes the task to the target server);

[0175] Key parameters include: unique task identifier (task_id), target server identifier (server_id), specific task parameters (e.g., for the "Regional Information Extraction" task, "target region name + list of document IDs to be analyzed" is required), task priority (high / medium / low), and execution timeout threshold (e.g., 300 seconds, if the timeout occurs, the task is considered to have failed).

[0176] Practical function: Ensures that tasks are accurately distributed to the corresponding server, and controls the execution priority and fault-tolerant triggering conditions of tasks through the "priority" and "timeout threshold" parameters (such as the high-priority "cause accumulation inference" task occupying resources first).

[0177] ③Task status query interface

[0178] Core functionality: The core intelligent agent obtains the execution progress and current status of submitted subtasks in real time to avoid "task disconnection";

[0179] Request method: GET (the core agent actively queries the target task status);

[0180] Key return information includes: task identifier (task_id), execution status (pending / processing / completed / failed), progress percentage (0-100, such as "processing 60%" indicating that more than half has been processed), and error reason when it fails (such as "data format error").

[0181] Practical function: Supports the "status monitoring" function of the core intelligent agent. If a task is found to be "pending" or "failed" for a long time, it can trigger a reassignment (such as transferring the failed task to another server of the same type).

[0182] ④ Result Acquisition Interface

[0183] Core function: The core intelligent agent extracts the final execution result from the server that has completed the task for subsequent "result integration";

[0184] Request method: GET (The core agent actively retrieves the result after the task status is "completed");

[0185] Key returned information includes: task identifier (task_id), execution server identifier (server_id), structured result data (e.g., "regional information extraction" returns "geological data fragments of the target area + confidence level"), execution timestamp (task completion time), and actual execution time (e.g., 25 seconds).

[0186] Practical function: To ensure that the output of each server can be collected in a unified manner, and to provide standardized input for the "results integrator" (such as associating the feature parameters of the "oil and gas feature identification server" with the potential range of the "dynamic potential assessment server").

[0187] ⑤ Service status update interface

[0188] Core Functions: A professional MCP Server synchronizes its own operating status and resource usage with the core intelligent agent in real time, supporting "load balancing" and "fault tolerance";

[0189] Request method: PUT (Server actively reports periodically, or reports are triggered when the status changes);

[0190] Key parameters include Server identifier (server_id), current running status (online normal / offline / maintenance), and real-time resource utilization (CPU / memory / GPU utilization percentage, such as "CPU: 75%, memory: 60%").

[0191] Practical function: The core intelligent agent judges the server load based on the reported "resource utilization" (e.g., if a server's CPU utilization reaches 90%, it is judged as high load), and subsequent tasks are preferentially assigned to low-load servers; if a server reports "maintenance", new tasks are temporarily stopped from being assigned to it to avoid task failure.

[0192] S120, Data transmission format construction

[0193] The data transmission format is the transmission format of task request and response data between the core intelligent agent and the professional MCP Server cluster. The request and response format is JSON, the encoding method is UTF-8, and the compression algorithm is gzip (automatically enabled when the size of a single data item is >10MB), which is used for large data transmission.

[0194] Data type definition: Includes three categories: text data, image data, and structured data;

[0195] The text data is stored in array form, with each element being an object containing "content-source-confidence level". The "content" is the extracted geological text fragment (e.g., "thickness of source rock in Kuqa Depression is 200m"), the "source" is the document number / name to which the data belongs, and the "confidence level" is the accuracy of the text extraction (0-1 range, e.g., 0.95 indicates 95% confidence).

[0196] Image data: stored in array form, with each element being an object containing "access address-description-metadata"; where "access address" is the storage path / URL of the chart in the document database, "description" is a summary of the chart content (e.g., "geological profile of Kuqa Depression in Tarim Basin"), and "metadata" is additional information such as chart resolution and acquisition time.

[0197] Structured data: stored in array form, each element is an object containing "entity-attribute-confidence"; where "entity" is a geological object (e.g., "Kuqa Depression reservoir"), "attribute" is the characteristic parameter of the entity (e.g., "porosity: 15%", "permeability: 10mD"), and "confidence" is the confidence level of the attribute value (0-1 range).

[0198] Error code definitions: 200 (request / response successful), 400 (request parameters missing / incorrect format, such as missing task ID), 401 (authentication failed, such as server not registered), 404 (requested resource does not exist, such as querying a non-existent task ID), 408 (request timed out, such as server not responding within 30 seconds), 500 (internal server error, such as algorithm execution error), 503 (server unavailable, such as under maintenance).

[0199] S200, Execute a modular scheduling process in the survey system.

[0200] S210 Geological Oil and Gas Survey Task Analysis and Breakdown

[0201] The core intelligent agent receives natural language input requests from users (such as "analyze the oil and gas resource potential of the Kuqa Depression in the Tarim Basin"); then the task parser decomposes the requests into a series of sub-tasks, including regional information extraction, oil and gas feature identification, cross-document association reasoning, data reliability assessment, spatiotemporal dimension classification, hydrocarbon accumulation element reasoning, and dynamic potential assessment.

[0202] Regional Information Extraction: Input the target region keywords (e.g., "Bohai Bay Basin") from the user's requirements, the original geological documents (exploration reports, well logs) and charts (geological profiles, regional structural maps) in the document database of the data storage layer; then, using a "keyword matching + geographic entity recognition algorithm", text fragments containing the target region (e.g., "thickness of reservoirs in the Dongying Depression of the Bohai Bay Basin") and chart annotation information (e.g., "depth annotation of a certain block in the Bohai Bay Basin" in the profile) are filtered from the original data, and irrelevant regional data is removed; finally, the structured data specific to the target region (including text fragments, chart metadata and extraction confidence scores, with confidence scores ≥0.8 considered as valid data) is output and temporarily stored in the results database.

[0203] Oil and gas feature identification: The results of regional information extraction are filtered to retain effective information related to oil and gas, and core feature parameters such as source rock thickness, reservoir porosity, and oil saturation are extracted. Specifically: the effective data of the target area output by the regional information extraction subtask are input; through "feature keyword extraction + parameter quantification", information directly related to oil and gas (such as "organic carbon content of source rock in Dongying Depression is 2.5%" and "reservoir porosity is 18%)" is filtered out, and non-oil and gas data (such as regional hydrological data) are removed; finally, the core feature parameter set of oil and gas in the target area (including feature type, value, and data source) is output and pushed to the cross-document association reasoning subtask.

[0204] Cross-document association reasoning: This involves retrieving multiple relevant survey documents, associating complementary data, filling information gaps in single documents, and forming a complete regional oil and gas related dataset. Specifically: Inputting the feature parameter set output from the oil and gas feature recognition subtask, and multiple complementary documents related to the target region from the document database (e.g., document A contains reservoir data, document B contains source rock data); employing a "semantic association algorithm + geological logic rules" (e.g., "spatial distribution correlation between reservoirs and source rocks"), associating complementary oil and gas data from multiple documents (e.g., using source rock maturity data from document B to supplement reservoir formation conditions in document A), and filling information gaps in single documents (e.g., missing "caprock thickness" data in a document); finally, outputting a complete oil and gas association dataset for the target region (including association descriptions, such as "reservoir porosity data comes from document A, source rock maturity data comes from document B"), and temporarily storing it in the results database.

[0205] Data reliability assessment: By comparing the datasets obtained from cross-document association inference, data inconsistencies are detected, and data validity is verified through a credibility algorithm. Specifically: Input the oil and gas association dataset output by the cross-document association inference subtask; then perform inconsistency detection, i.e., compare different source values ​​of the same parameter in the dataset (e.g., document A states "reservoir porosity 18%", document C states "25%)", and mark inconsistencies; then perform credibility verification, i.e., calculate credibility based on the authority of the data source (e.g., national-level exploration reports have a weight of 0.9, internal enterprise reports have a weight of 0.6) and sample size (data sampling points ≥50 have a weight of 0.8, <10 have a weight of 0.4), and a credibility score ≥0.7 is considered reliable data; finally, output the contradictory-free reliable dataset (including the credibility score of each parameter), and synchronize it to the spatiotemporal dimension classification subtask and results database.

[0206] Spatiotemporal Dimension Classification: Based on the three dimensions of "time, method, and element," valid data that has undergone reliability assessment is classified and integrated to establish a spatiotemporal correlation index. Specifically: The input consists of the reliable dataset output from the data reliability assessment subtask and the time range specified in the user's requirements (e.g., "2015-2025"). Then, a classification index is constructed according to the three dimensions of "time, method, and element": ① Time dimension: Data is divided by year (e.g., 2015-2020, 2021-2025); ② Method dimension: Classified by data acquisition method (e.g., seismic exploration, well logging, drilling); ③ Element dimension: Classified by hydrocarbon accumulation elements (e.g., source rock, reservoir, caprock, trap). Finally, a spatiotemporal classified structured dataset (containing three-dimensional index labels, e.g., "2021-2025-Seismic Exploration-Reservoir") is output and pushed to the hydrocarbon accumulation element inference subtask.

[0207] Reservoir Accumulation Factor Inference: The reservoir accumulation rules from the knowledge graph database are invoked to perform weighted calculations on the spatiotemporally classified dataset to obtain the regional reservoir accumulation matching degree. Specifically: The input consists of the structured dataset output from the spatiotemporal dimension classification subtask and the hydrocarbon accumulation rules from the knowledge graph database in the data storage layer (e.g., "Source rock organic carbon content ≥2% + reservoir porosity ≥15% → reservoir accumulation matching degree increases by 0.3"). Then, the "knowledge graph weighted calculation method" is used to assign weights to each reservoir accumulation factor (source rock weight 0.3, reservoir weight 0.25, caprock weight 0.2, trap weight 0.25), and the reservoir accumulation matching degree is calculated based on the dataset parameters (matching degree range 0-1, above 0.8 is considered a high match). Finally, the target region reservoir accumulation matching degree result (including the contribution value of each factor, such as "reservoir contribution 0.22") is output and temporarily stored in the results database.

[0208] Dynamic Potential Assessment: Combining data reliability assessment results with reservoir matching data, a potential assessment model is constructed to output oil and gas resource potential ranges. Specifically, the inputs are the reservoir matching results output by the reservoir element inference subtask and the credibility score output by the data reliability assessment subtask; then, a "potential assessment model" is constructed, which calculates the comprehensive potential value by weighting the reservoir matching degree (weight 0.6) and the credibility score (weight 0.4), and then, combined with the region's proven reserves data, divides the resource potential ranges (e.g., comprehensive potential value ≥ 0.8 → high potential area: resources 5-8 million tons; 0.5-0.8 → medium potential area: 2-5 million tons; < 0.5 → low potential area: < 2 million tons); finally, an oil and gas resource potential report for the target area (including potential ranges and key influencing factors) is output as the core part of the final result.

[0209] S220, Formulation and Execution of Intelligent Dispatch Strategies for Geological Oil and Gas

[0210] Based on the dependencies between executable subtasks, the resource configuration and real-time load of each MCP Server, the execution order and target server of each executable subtask are determined to achieve optimized resource allocation and load balancing.

[0211] Specifically, intelligent scheduling is achieved through a module scheduler based on the following factors:

[0212] S221, Geological and Oil Gas Survey Subtask Dependencies

[0213] The module scheduler determines the execution order of each subtask based on the logical dependencies between subtasks obtained by the task parser (e.g., "data reliability assessment" must be executed after "regional information extraction"). The specific dependency rules are as follows:

[0214] ① The output of the preceding subtask becomes the input of the following subtask: such as "Regional Information Extraction Subtask" → "Oil and Gas Feature Recognition Subtask" (regional data must be extracted first before the oil and gas features of the region can be identified); "Cross-Document Association Reasoning Subtask" → "Data Reliability Assessment Subtask" (multi-document data must be associated first before the data inconsistencies and credibility can be assessed);

[0215] ② Multiple subtask outputs jointly support subsequent subtasks: such as "data reliability assessment subtask" + "oil accumulation element inference subtask" → "dynamic potential assessment subtask" (reliability score and oil accumulation matching degree need to be input at the same time to calculate the potential range);

[0216] ③ Dependency-free subtasks can be executed in parallel: For example, after the input data conditions are met, some statistical or classification subtasks can be executed in parallel on different servers to improve overall efficiency.

[0217] S222, Optimized Resource Allocation

[0218] CPU-intensive tasks (such as text processing) are assigned to CPU servers;

[0219] GPU-intensive tasks (such as image processing and deep learning inference) are assigned to GPU servers;

[0220] Memory-intensive tasks (such as knowledge graph processing) are assigned to servers with large memory.

[0221] S223, Load Balancing

[0222] Monitor the load of each MCP Server and assign new tasks to servers with lower load.

[0223] S224, Priority Scheduling

[0224] Supports setting task priority, with higher priority tasks being executed first;

[0225] S225, Fault Tolerance Mechanism

[0226] When a task fails, it will be automatically retried or reassigned to another available MCP Server.

[0227] S230, geological oil and gas scheduling algorithm implementation

[0228] An improved genetic algorithm is used to achieve optimized scheduling, and the specific implementation is as follows:

[0229] S231, Chromosome Encoding

[0230] The task allocation scheme is encoded as chromosomes, with each gene representing a task assigned to a specific MCP Server (e.g., the gene "T1-S3" means "subtask T1 is assigned to MCP Server S3").

[0231] S232, Construction of Fitness Function

[0232] Fitness=α×(1-LoadBalance)+β×(1-ResourceWaste)+γ×(1-ExecutionTime);

[0233] In the above formula, LoadBalance represents the load balancing degree (0-1, the smaller the value, the more balanced the load of each server); ResourceWaste represents the resource waste rate (0-1, the smaller the value, the higher the resource utilization); ExecutionTime represents the normalized execution time (0-1, the smaller the value, the faster the execution speed); α, β, and γ are weighting coefficients that satisfy α+β+γ=1, and are used to control the weight of the impact of load balancing degree, resource waste rate, and execution time on fitness, respectively; currently, α=0.3, β=0.3, and γ=0.4 are set to make a trade-off between load balancing, resource utilization, and execution time;

[0234] S233, Selection Operation

[0235] Excellent individuals are selected based on their fitness value; a "roulette wheel selection method" is used, where individuals with higher fitness values ​​have a greater probability of being selected (e.g., an individual with Fitness=0.9 has twice the probability of being selected as an individual with Fitness=0.45), ensuring that high-quality allocation schemes are retained;

[0236] S234, Cross Operation

[0237] Exchange some genes of two superior individuals to generate a new allocation scheme; use "single-point crossover": randomly select a crossover position, exchange all genes after the crossover position of the two individuals, and set the crossover probability to 0.7-0.8.

[0238] S235, Mutation Operation

[0239] Randomly alter some genes of superior individuals to increase population diversity; Mutation rules: ① Only select 1%-3% of genes for mutation; ② The mutated genes must match the task type and server function.

[0240] S236, Iterative Optimization

[0241] The process is executed according to the multi-generation selection-crossover-mutation cycle, which includes repeating steps S234 to S235. Specifically, each generation includes one selection, one crossover, and one mutation operation. The iteration stops when: the average fitness value change rate of the population for three consecutive generations is <0.01 (i.e., the fitness value converges without significant improvement); or the number of iterations reaches a preset maximum value (e.g., 100 generations). If either condition is met, the iteration stops, and the individual with the largest fitness value is output as the optimal task allocation scheme.

[0242] S240, Application of Dynamic Expansion Mechanism

[0243] The system automatically adjusts the number of MCP Server instances based on workload and server running status, and supports online server access and exit to achieve dynamic system expansion. The dynamic expansion mechanism specifically includes:

[0244] S241, Automatic expansion and contraction capacity

[0245] The number of MCP Server instances can be automatically increased or decreased based on the workload.

[0246] S242, Module Hot-Swapping

[0247] Supports dynamic addition or removal of MCP Servers during the operation of the modular intelligent geological oil and gas survey system without restarting the system; the specific process is as follows: ① Add Server: Automatic access is completed through the service registration interface of the MCP protocol, and the core intelligent agent updates the service registry in real time; ② Remove Server: First, the unfinished tasks on the server are migrated to other servers of the same type, and then its status is marked as "offline" to stop the assignment of new tasks.

[0248] S243, Version Management

[0249] It supports the coexistence of multiple versions of the same MCP Server, enabling smooth upgrades.

[0250] The complete implementation process using the survey system of this invention is as follows:

[0251] Step 1: System Initialization and Service Registration

[0252] ① Operation:

[0253] Start the core intelligent agent and each MCP Server;

[0254] The MCP Server registers services with the core intelligent agent, providing functional descriptions, interface definitions, and resource requirements.

[0255] The core intelligent agent builds a service registry to record the status and capabilities of each MCP Server;

[0256] ② Output: System ready status, service registry.

[0257] Step 2: Task Reception and Parsing

[0258] ① Input: User requirements (e.g., "Analyze the oil and gas resource potential of the Kuqa Depression in the Tarim Basin");

[0259] ② Operation:

[0260] The core intelligent agent receives user requests;

[0261] The task parser breaks down the requirements into 7 sub-tasks;

[0262] Determine the dependencies and execution order between subtasks;

[0263] ③ Output: Task execution plan (including subtask list, dependencies, and execution order).

[0264] Step 3: Intelligent Scheduling and Task Execution

[0265] ① Input: Task execution plan;

[0266] ② Operation:

[0267] The module scheduler assigns a suitable MCP Server to each subtask based on the scheduling policy and system status.

[0268] Submit the task to the target MCP Server via the MCP protocol;

[0269] Monitor task execution status and progress;

[0270] Handling exceptional situations during task execution;

[0271] ③ Output: The execution results of each subtask.

[0272] Step 4: Results Integration and Output

[0273] ① Input: The execution results of each subtask;

[0274] ② Operation:

[0275] The results integrator combines the outputs of each subtask into a unified investigation report;

[0276] Generate visualization results (such as hydrocarbon accumulation potential maps and data correlation diagrams);

[0277] Output the final result;

[0278] ③ Outputs: Oil and gas resource survey report, visualization results.

[0279] Step 5: System Monitoring and Optimization

[0280] ① Operation:

[0281] The status monitor continuously monitors the operating status and resource usage of each module in the system;

[0282] Analyze system performance bottlenecks and propose optimization suggestions;

[0283] The resource allocation strategy is dynamically adjusted based on the task execution status.

[0284] ② Output: System operation report, performance optimization suggestions.

[0285] The invention will be further illustrated below with specific deployment examples and performance data:

[0286] Performance comparison under different hardware configurations:

[0287] When handling the same oil and gas survey task (oil and gas resource survey in the Kuqa Depression of the Tarim Basin), the performance data of different hardware configurations are as follows:

[0288]

[0289] Key findings:

[0290] ①The modular architecture in this embodiment reduces the processing time from 15 days to 3 days for the traditional monolithic architecture at the same hardware cost, improving efficiency by 80%;

[0291] ②In this embodiment, the resource utilization rate is significantly improved, with CPU utilization increasing from 45% to over 80% and GPU utilization increasing from 60% to over 85%;

[0292] ③ Deployment options can be flexibly selected based on task requirements and budget to achieve a balance between performance and cost.

[0293] Module reusability and development cost savings:

[0294] The modular architecture of this embodiment supports high module reuse, resulting in the following cost savings when developing different types of oil and gas survey systems:

[0295]

[0296] Statistical results:

[0297] Average development cost savings: 65.05%;

[0298] Average development cycle shortened by 65.05%;

[0299] Module reuse rate: 70-100% (depending on system type);

[0300] Key findings:

[0301] The modular architecture of this invention enables high module reuse, saving an average development cost of over 65%.

[0302] The development cycle has been significantly shortened, from the traditional 100-150 person-days to 30-55 person-days;

[0303] The modules have a high reuse rate, especially in similar systems, where most core modules can be reused.

[0304] System scalability testing:

[0305] The system's scalability test results under different task loads are as follows:

[0306]

[0307] Key findings:

[0308] ① The system supports automatic scaling based on task load. When the number of tasks increases by 20 times, reasonable response time can still be maintained by expanding the number of servers.

[0309] ② Throughput increases with the number of tasks, but the growth rate gradually slows down, which is consistent with the system expansion pattern;

[0310] ③ Resource utilization rate remains within a reasonable range (40%-85%) under different loads, avoiding resource waste.

[0311] System stability test:

[0312] The system performed as follows during a 30-day stability test:

[0313]

[0314] Key findings:

[0315] ① The system has high stability and 99.9% availability, meeting the requirements of the production environment;

[0316] ② Strong fault recovery capability, with an average recovery time of only 15 minutes;

[0317] ③ The task has a high success rate, data consistency is guaranteed, and the reliability is high.

[0318] This invention has a reasonable concept and possesses the capabilities of "modular architecture, standardized interface, flexible scheduling, and resource optimization," enabling the efficient construction, flexible expansion, and intelligent scheduling of a modular intelligent geological oil and gas survey system. It can achieve automated, intelligent, and efficient scheduling of geological oil and gas surveys.

[0319] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A modular intelligent geological oil and gas survey system, characterized in that: The survey system includes a core intelligent agent, a specialized MCP Server cluster, and a data storage layer; The core intelligent agent is responsible for task parsing, module scheduling, result integration and exception handling. Its core components include a task parser, a module scheduler, a result integrator, an exception handler and a status monitor. The task parser is used to parse user requests into executable subtasks; The module scheduler is used to intelligently schedule MCP Servers based on task type and system status; The result integrator is used to integrate the outputs of each MCP Server into a final result; The exception handler is used to handle abnormal situations during execution to ensure stable system operation; The status monitor is used to monitor the operating status and resource usage of each module of the system; The specialized MCP Server cluster utilizes a standardized model invocation protocol to enable communication and data exchange between the core intelligent agent and each specialized server. The specialized MCP Server cluster includes a regional information extraction server, an oil and gas feature recognition server, a cross-document association reasoning server, a data reliability assessment server, a spatiotemporal dimension classification server, an oil and gas accumulation element reasoning server, and a dynamic potential assessment server. The regional information extraction server is used to accurately extract text or charts from a specified region; The oil and gas feature recognition server is used to filter oil and gas related information and extract core features; The cross-document association reasoning server is used to associate complementary data from multiple documents to fill in information gaps. The data reliability assessment server is used to detect contradictions in the multi-document complementary data output by the cross-document association reasoning server and verify its credibility. The spatiotemporal dimension classification server is used to integrate the associated data output by the regional information extraction server, the oil and gas feature recognition server, and the cross-document association reasoning server according to the three dimensions of "time-method-element". The hydrocarbon accumulation element inference server is used to calculate the hydrocarbon accumulation matching degree based on a knowledge graph weighted calculation. The dynamic potential assessment server is used to combine the credibility results output by the data reliability assessment server with the reservoir matching degree data of the reservoir element inference server to output the resource potential range. The data storage layer is used to store various types of data during system operation, including a document database, a knowledge graph library, and a results database. The document database stores original geological and oil and gas survey documents and charts; the knowledge graph library stores knowledge and association rules related to oil and gas accumulation; and the results database stores the intermediate execution results of each MCP Server and the final output results of the system.

2. A modular intelligent geological oil and gas scheduling method, based on the modular intelligent geological oil and gas survey system described in claim 1, characterized in that, Includes the following steps: (1) Establish a standardized interface and data transmission format between the core intelligent agent and the professional MCP Server cluster based on the MCP protocol: (1.1) Interface Definition The MCP protocol defines a standardized interface between the core intelligent agent and the specialized MCP Server cluster; the specific process is as follows: (1.2) Construction of data transmission format The data transmission format is the transmission format of task request and response data between the core intelligent agent and the professional MCP Server cluster. The request and response format is JSON, the encoding method is UTF-8, and the compression algorithm is gzip, which is used for large data transmission. Data type definition: Includes three categories: text data, image data, and structured data; among them, text data is a string array containing content, source, and confidence level; image data is an object array containing access address, description, and metadata; structured data is an object array containing entity, attribute, and confidence level. Error code definitions: 200 indicates success, 400 indicates incorrect request parameters, 401 indicates unauthorized, 404 indicates resource does not exist, 408 indicates request timeout, 500 indicates internal server error, and 503 indicates service unavailable. (2) Execute a modular scheduling process in the survey system: (2.1) Geological oil and gas survey task parsing and decomposition, that is, the core intelligent agent receives the geological oil and gas survey requirements in natural language form input by the user, and decomposes them into executable sub-tasks through the task parser; (2.2) Formulation and execution of intelligent scheduling strategy for geological oil and gas, that is, based on the dependency relationship between executable sub-tasks, the resource configuration and real-time load of each MCPServer, the execution order and target Server of each executable sub-task are determined to achieve optimized resource allocation and load balancing. (2.3) The application of the geological oil and gas scheduling algorithm and the dynamic expansion mechanism, that is, the candidate task allocation scheme is iteratively optimized based on the genetic algorithm to obtain a task scheduling scheme that meets the preset fitness target; (2.4) The application of dynamic expansion mechanism, that is, the number of MCP Server instances is automatically adjusted according to the workload and Server running status, and the online access and exit of the Server is supported to realize the dynamic expansion of the system.

3. The modular intelligent geological oil and gas scheduling method according to claim 2, characterized in that: The standardized interfaces in step (1.1) include service registration interface, task submission interface, task status query interface, result retrieval interface and service status update interface; The service registration interface is used by the specialized MCP Server to submit service information to the core intelligent agent, including server identifier, function description, input and output data format and resource requirements, to complete the service access registration; The task submission interface is used by the core intelligent agent to submit subtasks to the registered specialized MCP Server, including task identifier, target server identifier, task parameters, priority and timeout threshold; The task status query interface is used by the core intelligent agent to query the execution status of submitted subtasks, including four statuses: pending execution, processing, completed, and failed, as well as corresponding progress, result preview, or error information. The result acquisition interface is used by the core intelligent agent to obtain the final execution result of the completed sub-task, including task identifier, execution server identifier, result data, timestamp and execution time; The service status update interface is used by the specialized MCP Server to synchronize its own operating status and real-time resource utilization with the core intelligent agent.

4. The modular intelligent geological oil and gas scheduling method according to claim 2, characterized in that, The specific process of task parsing and decomposition in step (2.1) is as follows: (2.1.1) Receive natural language input from the user through the core intelligent agent; (2.1.2) The natural language requirements are decomposed into a series of sub-tasks by the task parser. These sub-tasks include regional information extraction, oil and gas feature recognition, cross-document association reasoning, data reliability assessment, spatiotemporal dimension classification, hydrocarbon accumulation element reasoning, and dynamic potential assessment.

5. The modular intelligent geological oil and gas scheduling method according to claim 4, characterized in that, The regional information extraction is based on the geographical scope keywords in the user's requirements, and accurately locates and extracts the geological structure and stratigraphic data of the region from the original survey documents and charts; The oil and gas feature identification involves filtering the regional information extraction results, retaining effective information related to oil and gas, and extracting source rock thickness, reservoir porosity, and oil saturation. The cross-document association reasoning involves retrieving multiple relevant survey documents, associating complementary data, filling information gaps in a single document, and forming a complete regional oil and gas related dataset. The data reliability assessment involves comparing datasets obtained through cross-document association reasoning, detecting data inconsistencies, and verifying data validity through a credibility algorithm. The spatiotemporal dimension classification is based on three dimensions: "time-method-element". It classifies and integrates valid data that has undergone reliability assessment and establishes a spatiotemporal correlation index. The reasoning of the hydrocarbon accumulation elements involves calling the hydrocarbon accumulation rules in the knowledge graph library, performing weighted calculations on the spatiotemporally classified dataset, and obtaining the regional hydrocarbon accumulation matching degree. The dynamic potential assessment combines data reliability assessment results with reservoir matching data to construct a potential assessment model and output oil and gas resource potential ranges.

6. The modular intelligent geological oil and gas scheduling method according to claim 2, characterized in that, The formulation and execution of the intelligent scheduling strategy for geological oil and gas in step (2.2) is achieved through intelligent scheduling based on the following factors by the modular scheduler: (2.2.1) Dependencies of Geological Oil and Gas Survey Subtasks The module scheduler determines the execution order of each subtask based on the logical dependencies between subtasks obtained by the task resolver; (2.2.2) Optimal allocation of resources CPU-intensive tasks are assigned to CPU servers; GPU-intensive tasks are assigned to GPU servers; Memory-intensive tasks are assigned to servers with large memory. (2.2.3) Load balancing Monitor the load of each MCP Server and assign new tasks to servers with lower load. (2.2.4) Priority scheduling Supports setting task priority, with higher priority tasks being executed first; (2.2.5) Fault Tolerance Mechanism When a task fails, it will be automatically retried or reassigned to another available MCP Server.

7. The modular intelligent geological oil and gas scheduling method according to claim 6, characterized in that, The logical dependencies between subtasks include the output of the preceding subtask serving as the input of the following subtask, the outputs of multiple subtasks jointly supporting the following subtask, and the ability for non-dependent subtasks to be executed in parallel.

8. The modular intelligent geological oil and gas scheduling method according to claim 2, characterized in that, The geological oil and gas scheduling algorithm in step (2.3) is implemented using an improved genetic algorithm for optimized scheduling. The specific implementation method is as follows: (2.3.1) Chromosome coding The task allocation scheme is encoded as chromosomes, with each gene representing a task assigned to one MCP Server; (2.3.2) Construction of fitness function Fitness=α×(1-LoadBalance)+β×(1-ResourceWaste)+γ×(1-ExecutionTime); In the above formula, LoadBalance is the load balance degree; ResourceWaste is the resource waste rate; ExecutionTime is the normalized execution time; α, β, and γ are all weight coefficients that satisfy α+β+γ=1, and are used to control the weights of the influence of load balance degree, resource waste rate, and execution time on fitness, respectively. (2.3.3) Selection Operation Selecting superior individuals based on Fitness scores; (2.3.4) Crossover operation By exchanging some genes between two superior individuals, a new allocation scheme can be generated; (2.3.5) Mutation operation Randomly altering some genes in superior individuals increases population diversity; (2.3.6) Iterative optimization The process follows a multi-generational selection-crossover-mutation cycle, which includes repeating steps (2.3.4) to (2.3.5). Specifically, each generation includes one selection, one crossover, and one mutation operation. The iteration stops when either of these conditions is met: ① the average fitness value change rate of the population over three consecutive generations is less than 0.01; or ② the number of iterations reaches a preset maximum value. The iteration stops when either condition is met, and the individual with the highest fitness value is output as the optimal task allocation scheme.

9. The modular intelligent geological oil and gas scheduling method according to claim 2, characterized in that, The dynamic expansion mechanism in step (2.4) includes: (2.4.1) Automatic expansion and contraction The number of MCP Server instances can be automatically increased or decreased based on the workload. (2.4.2) Module hot-swappable Supports dynamically adding or removing MCP Servers during the operation of the modular intelligent geological oil and gas survey system without restarting the system; (2.4.3) Version Management It supports the coexistence of multiple versions of the same MCP Server, enabling smooth upgrades.

10. The modular intelligent geological oil and gas scheduling method according to claim 9, characterized in that, The specific process of step (2.4.2) is as follows: ① Add Server: Automatic access is completed through the service registration interface of the MCP protocol, and the core intelligent agent updates the service registry in real time; ② Remove Server: First, migrate any unfinished tasks on the server to other servers of the same type, then mark its status as "offline" and stop assigning new tasks.

Images