A method and system for generating intelligent operation and maintenance reports based on large models

Abstract

This invention relates to the field of intelligent analysis and data processing technology, specifically to an intelligent operation and maintenance report generation method based on a large model. This method addresses the problems of low report generation efficiency, high data query thresholds, weak data analysis capabilities, and limited interaction methods in traditional operation and maintenance management. The method encapsulates database operations into standardized tool interfaces through the MCP tool layer, enabling secure and regulated access to operation and maintenance data by the large model. It dynamically generates SQL permission filtering conditions based on user organizational permissions to ensure data access security and prevent unauthorized access. The method employs SSE streaming transmission to achieve real-time response from the large model, supporting both HTTP and NL2SQL data sources to enhance the user experience. The report generation layer supports multiple report types, automatically calculates month-on-month time periods, and aggregates multi-dimensional operation and maintenance data to generate structured reports. The multi-dimensional analysis layer integrates intelligent data querying, data analysis, trend prediction, and intelligent recommendation functions to achieve comprehensive intelligent analysis of operation and maintenance data.

Description

Technical Field

[0001] This invention relates to the field of intelligent analysis and data processing technology, specifically to a method and system for generating intelligent operation and maintenance reports based on a large model. Background Technology

[0002] With the rapid development of the social economy and the widespread application of information technology, the operation and maintenance management of enterprise information technology (IT) systems and industrial equipment is becoming increasingly complex. Under the wave of digital transformation, the operation and maintenance process has generated an explosive growth of multi-dimensional data, encompassing inspection task records, maintenance work order history, real-time equipment alarms, and equipment health assessment indicators. Therefore, how to efficiently utilize this massive amount of data to achieve intelligent and automated operation and maintenance has become crucial for improving enterprise operational efficiency and ensuring business continuity. However, traditional operation and maintenance management systems are gradually revealing significant technical bottlenecks when dealing with the current complex data environment and the demands for intelligent operation and maintenance, making it difficult to meet the requirements of efficient operation and maintenance.

[0003] First and foremost, the most prevalent problem is the inefficiency of report generation. When conducting periodic reports or troubleshooting, operations and maintenance (O&M) personnel must manually export data from multiple heterogeneous monitoring platforms, ticketing systems, and databases. They then use spreadsheets and other tools for tedious data cleaning, organization, and analysis before finally writing the O&M report. This process is not only time-consuming and labor-intensive, often taking hours or even days, but the manual operation is also highly prone to data errors and omissions of crucial information, severely impacting the accuracy and timeliness of the report.

[0004] Secondly, the technical barrier to data retrieval is too high. Operational data is typically stored in structured relational databases, and retrieving specific information requires writing Structured Query Language (SQL) statements. Ordinary operations and maintenance personnel lack database expertise and cannot independently obtain the required data, thus relying on technical support.

[0005] Secondly, existing systems have weak data analysis capabilities. Most current operations and maintenance management systems remain at the level of data visualization, lacking in-depth intelligent analysis capabilities. These systems cannot automatically identify abnormal patterns in the data, discover potential correlations, or effectively predict future trends based on historical data. This passive response model results in a lack of strong data support for operations and maintenance decisions, making it difficult to shift from "passive firefighting" to "proactive prevention."

[0006] Furthermore, the human-computer interaction methods are too simplistic. Users can only operate through the system's preset graphical user interface (GUI) and fixed menu options, unable to express their needs using natural language. When faced with complex, non-standardized query or analysis requirements, the system cannot flexibly understand and respond to users' personalized intentions, limiting the flexibility and convenience of operation and maintenance work.

[0007] Finally, the access control mechanism is complex and rigid. Operational data often involves sensitive information from multiple departments, different regions, and different levels, making access management itself extremely complex. Existing systems struggle to provide flexible and convenient data access capabilities for users with different roles while ensuring data security and preventing unauthorized access, which to some extent restricts data sharing and collaboration.

[0008] Therefore, how to build a new type of intelligent operation and maintenance system that can realize natural language interactive query, automated report generation and intelligent analysis and prediction to overcome the shortcomings of existing technologies has become an urgent technical problem to be solved in this field. Summary of the Invention

[0009] The purpose of this invention is to provide an intelligent operation and maintenance report generation method based on a large model, which aims to solve the problems of low report generation efficiency, high data query threshold, weak data analysis capability, and single interaction method in traditional operation and maintenance management.

[0010] To solve the above-mentioned technical problems, the technical solution provided by the present invention is as follows: This invention provides a method for generating intelligent operation and maintenance reports based on a large model, comprising the following steps: Step S1: Collect operation and maintenance data, process the data, store it, and output a structured operation and maintenance database; the operation and maintenance data includes inspection task data, maintenance work order data, equipment alarm data, and equipment health data. Step S2: Based on the structured operation and maintenance database obtained in step S1, the functions of querying the logical structure schema of the structured operation and maintenance database, executing structured query statements (SQL), and generating report data are encapsulated into multiple standardized tool interfaces, which are then accessed by large models through the Model Context Protocol (MCP) tool call protocol. Step S3: Receive the user's natural language input, construct request parameters containing user permissions, organizational structure, and regional scope context, and conduct real-time intelligent interaction with the large model through Server-Sent Events (SSE) streaming. Step S4: Based on the time range, regional range, and content type parameters configured in the report plan, automatically aggregate multi-dimensional operation and maintenance data, and call the large model to generate a structured operation and maintenance report.

[0011] Based on the above technical solution, step S5 is also included: multi-dimensional intelligent analysis of operation and maintenance data, including: intelligent data query, data analysis, trend prediction and intelligent recommendation.

[0012] Based on the above technical solution, the data processing and storage process in step S1 includes the following steps: S21, Data Cleaning: Remove duplicate records with the same task ID, fill null fields with default values ​​or the previous record, and convert time fields to a standard time format. S22, Data Conversion: Convert Unix timestamps to standard time format, map status enumeration values ​​to standard codes, and perform numerical normalization on health scores; S23, Data storage: Write the processed data into the corresponding table structure of the database and create indexes to optimize queries; the output includes at least a structured operation and maintenance database containing inspection task table, maintenance work order table, equipment alarm table, and equipment health table.

[0013] Based on the above technical solution, the standardization tools in step S2 include: Data Table Name Retrieval Tool: Retrieves a list of names for all tables in a structured operations and maintenance database, returning a result in string array format; Tool to obtain the complete structure of a specified table: Obtain the complete structure information of a specified table, including column definitions, primary keys, and unique key metadata; Tool to retrieve information about all columns in a specified table: Retrieves information about all columns in a specified table, including column name, data type, and whether it is nullable; Query execution tool: Executes query statements and returns results, supports permission filtering and parameterized queries; Tools for obtaining operation and maintenance information: Generate structured report data based on time range, region, and content type parameters; Permission verification tool: Parses user tokens to obtain organization ID, queries organization region mapping table to obtain list of accessible regions, and automatically adds region filtering conditions to SQL queries; Parameterized query tool: Uses prepared statements to replace string concatenation, performs type validation and length limit on input parameters, and escapes special characters.

[0014] Based on the above technical solution, step S3 includes the following process: S31, Create a user session and generate a unique session identifier; parse the user ID and organization ID from the user token and associate the user and organization; establish a session context and create a mapping relationship in the session storage; S32, based on the organization region association table, obtain the list of accessible regions according to the user's organization permissions, and construct the data access permission context; S33, identify the interaction type and select the corresponding processing flow, where the interaction types include: SQL query type, analysis type, prediction type and recommendation question type; use a large model to perform semantic analysis on user input, match predefined intent templates, and output intent classification results and confidence scores; S34 establishes a Server-Sent Events (SSE) streaming connection, pushes large model responses in character chunks, and renders the response content in real time on the front end.

[0015] Based on the above technical solution, the generation of the structured operation and maintenance report in step S4 includes the following process: S41, based on the report plan, obtain the time range, area range, device range, and content type configuration parameters; S42, calculate the month-on-month time period based on the report type, where the report types include: annual report, quarterly report, monthly report, weekly report and custom report; S43 aggregates multi-dimensional data, calls the MCP tool's interface for obtaining operation and maintenance information, and generates structured data that includes basic report information, core indicators, inspection data, work order data, equipment alarm data, and equipment health data. S44 uses structured data as input parameters to call the large model and generates a structured operation and maintenance report in natural language form. The structured operation and maintenance report includes natural language text, key indicator tables, and report metadata.

[0016] Based on the above technical solution, the multi-dimensional intelligent analysis in step S5 includes the following process: Intelligent Questioning: It receives users' natural language questions, converts the natural language into SQL query statements through a large model and schema context, executes the query after SQL permission verification and returns the results, and supports multi-turn dialogue and contextual understanding; Data Analysis: Intelligent analysis of query results, using statistical methods to identify data points that deviate from the normal range, performing year-on-year and month-on-month calculations and rate of change analysis, identifying data anomalies and trends, and generating analytical conclusions and recommendations; Trend prediction: Based on historical data and time series analysis algorithms, predict future trends and output predicted values ​​and confidence intervals for the next N days or N weeks to support operation and maintenance decisions; Intelligent recommendation: Based on users' historical interaction records and behavior patterns, it counts users' high-frequency query types, areas of interest, and commonly used indicators, and performs collaborative filtering based on the analysis dimensions of similar users' interests to generate Top-N related questions and analysis dimensions.

[0017] This invention provides an intelligent operation and maintenance report generation system based on a large model, employing the aforementioned intelligent operation and maintenance report generation method, including: The operations and maintenance data acquisition layer collects operations and maintenance data, processes the data, stores it, and outputs a structured operations and maintenance database. The MCP tool layer, based on a structured operations and maintenance database, encapsulates the functions of querying the logical structure schema of the structured operations and maintenance database, executing structured query statements (SQL), and generating report data into multiple standardized tool interfaces, which are then accessed by large models through the Model Context Protocol (MCP) tool call protocol. The intelligent interaction layer receives natural language input from users, constructs request parameters that include user permissions, organizational structure, and regional context, and performs real-time intelligent interaction with the large model through Server-Sent Events (SSE) streaming. The operation and maintenance report generation layer automatically aggregates multi-dimensional operation and maintenance data and calls large models to generate structured operation and maintenance reports based on the time range, regional range and content type parameters configured in the report plan. The multi-dimensional analysis layer provides intelligent analysis of operational data across multiple dimensions, including intelligent data retrieval, data analysis, trend prediction, and intelligent recommendation.

[0018] The beneficial effects of the technical solution provided by this invention are as follows: (1) Improve report generation efficiency: Through automated data aggregation and large model generation technology, the time for generating operation and maintenance reports is reduced from several hours to minutes, significantly improving work efficiency.

[0019] (2) Lowering the threshold for data query: Through natural language interaction and NL2SQL technology, ordinary operation and maintenance personnel can query data independently without mastering SQL skills, thus lowering the technical threshold.

[0020] (3) Enhance data analysis capabilities: integrate intelligent analysis, trend prediction and anomaly identification functions to automatically discover data patterns and potential problems, and support scientific decision-making.

[0021] (4) Improve the interactive experience: By adopting natural language interaction and streaming response technology, users can communicate with the system in everyday language and get real-time responses, thus improving the user experience.

[0022] (5) Ensure data security: Through dynamic permission filtering and context permission control, ensure that users can only access data within the authorized scope to prevent data leakage.

[0023] (6) Supports flexible expansion: The MCP tool layer adopts a standardized interface design, which can easily expand new data sources and analysis tools to adapt to business changes.

[0024] (7) Reduce operation and maintenance costs: Reduce manual data processing and report writing work, free up the energy of operation and maintenance personnel, and reduce labor costs.

[0025] (8) Improve decision-making quality: Provide data-driven decision support through intelligent analysis and trend prediction, and reduce the subjectivity and error of experience-based judgment. Attached Figure Description

[0026] Figure 1 This is a flowchart illustrating the intelligent operation and maintenance report generation method of this invention; Figure 2 This is a schematic diagram of the architecture of the intelligent operation and maintenance report generation system in this invention; Figure 3 This is a flowchart illustrating the intelligent interaction process in step S3 of the present invention. Figure 4 This is a flowchart illustrating the intelligent question processing and analysis process in step S5 of the present invention. Detailed Implementation

[0027] The present invention will be further described below with reference to the accompanying drawings and embodiments: In this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," "linking," and "fixing," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0028] In the description of this invention, it should be understood that the terms "left", "right", "front", "rear", "top", "bottom", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0029] like Figures 1 to 4 As shown, a method for generating intelligent operation and maintenance reports based on a large model includes the following steps: Step S1: Collect operation and maintenance data, process the data, store it, and output a structured operation and maintenance database; the operation and maintenance data includes inspection task data, maintenance work order data, equipment alarm data, and equipment health data. Specifically, in step S1, the inspection task data includes the task, area, task status, planned start time, actual completion time, and inspection personnel information; the maintenance work order data includes the work order name, area, work order status, work order type, handling personnel, and handling time information; the equipment alarm data includes the alarm name, alarm level, alarm time, handling status, and alarm type information; and the equipment health data includes the equipment name, health score, health trend, and assessment time information.

[0030] Based on the above technical solution, the data processing and storage process in step S1 includes the following steps: S11, Data Cleaning: Remove duplicate records with the same task ID, fill null fields with default values ​​or the previous record, and convert time fields to a standard time format. This refers to the data cleaning process that removes duplicate records, fills in missing values, and standardizes the data format.

[0031] S12, Data Conversion: Convert Unix timestamps to standard time format, map status enumeration values ​​to standard codes, and perform numerical normalization on health scores; This involves unifying timestamps, mapping enumerated values, and normalizing numerical values, such as normalizing health scores from 0 to 100. S13, Data storage: Write the processed data into the corresponding table structure of the database and create indexes to optimize queries; the output includes at least a structured operation and maintenance database containing inspection task table, maintenance work order table, equipment alarm table, and equipment health table.

[0032] The structured operations and maintenance database output in this step serves as the data source for the MCP tool layer in step S2, allowing the tool interface to query and read it.

[0033] Step S2: Based on the structured operation and maintenance database obtained in step S1, the functions of querying the logical structure schema of the structured operation and maintenance database, executing structured query statements (SQL), and generating report data are encapsulated into multiple standardized tool interfaces, which are then accessed by large models through the Model Context Protocol (MCP) tool call protocol. The standardization tools in step S2 include: Data Table Name Retrieval Tool: Retrieves a list of names for all tables in a structured operations and maintenance database, returning a result in string array format; Tool to obtain the complete structure of a specified table: Obtain the complete structure information of a specified table, including column definitions, primary keys, and unique key metadata; Tool to retrieve information about all columns in a specified table: Retrieves information about all columns in a specified table, including column name, data type, and whether it is nullable; Query execution tool: Executes query statements and returns results, supports permission filtering and parameterized queries; Tools for obtaining operation and maintenance information: Generate structured report data based on time range, region, and content type parameters; The permission verification tool filters the accessible data range based on the user's organization ID: it parses the user token to obtain the organization ID, queries the organization region mapping table to obtain the list of accessible regions, and automatically adds region filtering conditions to the SQL query. A parameterized query tool that prevents SQL injection and supports dynamic parameter binding: It uses prepared statements to replace string concatenation, performs type validation and length limits on input parameters, and escapes special characters.

[0034] The MCP tool interface output in this step serves as the calling interface for the intelligent interaction layer in step S3, the report generation layer in step S4, and the multidimensional analysis layer in step S5, providing unified data access capabilities.

[0035] Step S3: Receive the user's natural language input, construct request parameters containing user permissions, organizational structure, and regional scope context, and conduct real-time intelligent interaction with the large model through Server-Sent Events (SSE) streaming. Step S3 includes the following process: S31, Create a user session and generate a unique session identifier; parse the user ID and organization ID from the user token and associate the user and organization; establish a session context and create a mapping relationship in the session storage; S32, based on the organization region association table, retrieves the list of accessible regions according to the user's organization permissions, and constructs a data access permission context to ensure data security; S33, identify the interaction type and select the corresponding processing flow, where the interaction types include: SQL query type, analysis type, prediction type and recommendation question type; use a large model to perform semantic analysis on user input, match predefined intent templates, and output intent classification results and confidence scores; S34 establishes a Server-Sent Events (SSE) streaming connection, pushing large model responses in character chunks, with the front end rendering the response content in real time. In other words, it returns fragments of the large model's response content in real time via SSE streaming, improving the user experience.

[0036] This step outputs a real-time response stream, with text snippets pushed via SSE; a query result set, returned when the interaction type is an SQL query; and an updated session context supporting multi-turn dialogue. The query results output in this step also serve as data input for the next step's report generation process, or are directly returned to the user; the updated session context supports the multi-turn dialogue functionality of step S5's multidimensional analysis.

[0037] Step S4: Based on the time range, regional range, and content type parameters configured in the report plan, automatically aggregate multi-dimensional operation and maintenance data, and call the large model to generate a structured operation and maintenance report.

[0038] The generation of the structured operation and maintenance report in step S4 includes the following process: S41, based on the report plan, obtain the time range, area range, device range, and content type configuration parameters; S42, calculate the month-on-month time period based on the report type, where the report types include: annual report, quarterly report, monthly report, weekly report and custom report; The monthly report covers the period from the 1st to the end of the current month, with a month-on-month comparison period from the 1st to the end of the previous month; the quarterly report covers the period from the first month to the last month of the current quarter, with a month-on-month comparison period of the previous quarter; the annual report covers the period from January 1st to December 31st of the current year, with a month-on-month comparison period of the previous year. S43 aggregates multi-dimensional data, calls the MCP tool's interface for obtaining operation and maintenance information, and generates structured data that includes basic report information, core indicators, inspection data, work order data, equipment alarm data, and equipment health data. The API calls the MCP tool to obtain operation and maintenance information and calculates core indicators including inspection completion rate, work order closure rate, alarm handling timeliness rate, and average health score. Aggregated inspection data includes total number of tasks, number of completed tasks, number of overdue tasks, and personnel distribution. Aggregated work order data includes total number of work orders, distribution by type, distribution by status, and average processing time. Aggregated alarm data includes total number of alarms, distribution by level, unprocessed alarms, and alarm trend. Aggregated health score data includes equipment health score distribution, list of equipment with declining health score, and trend analysis. S44 uses structured data as input parameters to call the large model and generates a structured operation and maintenance report in natural language form. The structured operation and maintenance report includes natural language text, key indicator tables, and report metadata.

[0039] Preferably, the operation and maintenance report supports multiple report templates and formats: build report template prompts, attach structured data JSON, call large models to generate report content in natural language, and render the report content into multiple output formats such as Markdown, PDF, and HTML.

[0040] This step generates a structured operations and maintenance report, including natural language text and key indicator tables; a report file available for download in various formats such as PDF or HTML; and report metadata, including generation time, coverage, and data source version. The structured operations and maintenance report can be directly delivered to users, and the report metadata can be used as a historical interaction record for intelligent recommendations in the S5 multidimensional analysis layer.

[0041] Based on the above technical solution, step S5 is also included: multi-dimensional intelligent analysis of operation and maintenance data, including: intelligent data query, data analysis, trend prediction and intelligent recommendation.

[0042] The multi-dimensional intelligent analysis in step S5 includes the following processes: Intelligent Questioning: It receives users' natural language questions, converts the natural language into SQL query statements through a large model and schema context, executes the query after SQL permission verification and returns the results, and supports multi-turn dialogue and contextual understanding; Data Analysis: Intelligent analysis of query results, using statistical methods to identify data points that deviate from the normal range, performing year-on-year and month-on-month calculations and rate of change analysis, identifying data anomalies and trends, and generating analytical conclusions and recommendations; Trend prediction: Based on historical data and time series analysis algorithms, predict future trends and output predicted values ​​and confidence intervals for the next N days or N weeks to support operation and maintenance decisions; Preferably, ARIMA, Prophet, or LSTM algorithms can be selected to fit a model based on historical data for trend prediction.

[0043] Intelligent Recommendation: Based on users' historical interaction records and behavioral patterns, this feature analyzes users' frequently used query types, areas of interest, and commonly used metrics. It then performs collaborative filtering based on analytical dimensions followed by similar users to generate Top-N relevant questions and analytical dimensions, thereby improving user experience.

[0044] In this step, the intelligent query function outputs query results with natural language explanations; the data analysis function outputs analysis conclusions with a suggestion list; the trend prediction function outputs prediction curves with key inflection point descriptions; and the intelligent recommendation function outputs a list of recommended questions. This step also relies on the MCP tool interface output in step S2 to obtain data, on the session context output in step S3 to support multi-turn dialogues, and on the report metadata output in step S4 to support intelligent recommendations.

[0045] This application also provides an intelligent operation and maintenance report generation system based on a large model, characterized in that it employs the aforementioned intelligent operation and maintenance report generation method, including: The operations and maintenance data acquisition layer collects operations and maintenance data, processes the data, stores it, and outputs a structured operations and maintenance database. The MCP tool layer, based on a structured operations and maintenance database, encapsulates the functions of querying the logical structure schema of the structured operations and maintenance database, executing structured query statements (SQL), and generating report data into multiple standardized tool interfaces, which are then accessed by large models through the Model Context Protocol (MCP) tool call protocol. The intelligent interaction layer receives natural language input from users, constructs request parameters that include user permissions, organizational structure, and regional context, and performs real-time intelligent interaction with the large model through Server-Sent Events (SSE) streaming. The operation and maintenance report generation layer automatically aggregates multi-dimensional operation and maintenance data and calls large models to generate structured operation and maintenance reports based on the time range, regional range and content type parameters configured in the report plan. The multi-dimensional analysis layer provides intelligent analysis of operational data across multiple dimensions, including intelligent data retrieval, data analysis, trend prediction, and intelligent recommendation.

[0046] This application also provides a computer system, including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the intelligent operation and maintenance report generation method based on a large model.

[0047] This application also provides a computer-readable medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the described intelligent operation and maintenance report generation method based on a large model.

[0048] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the aforementioned intelligent operation and maintenance report generation method based on a large model.

[0049] This invention provides an intelligent operation and maintenance report generation method based on a large model, aiming to solve the problems of low report generation efficiency, high data query threshold, weak data analysis capabilities, and limited interaction methods in traditional operation and maintenance management. Specifically, the method encapsulates structured operation and maintenance database operations into standardized tool interfaces through the MCP tool layer, enabling secure and standardized access to operation and maintenance data by the large model. The tool interface includes three core functions: schema query, SQL execution, and report data generation. SQL permission filtering conditions are dynamically generated based on user organizational permissions to ensure data access security and prevent unauthorized access. The method uses SSE streaming transmission to achieve real-time response from the large model, supporting both HTTP data sources and NL2SQL data sources to improve the user experience. The report generation layer supports various report types such as annual, quarterly, monthly, weekly, and custom reports, and automatically calculates... The system calculates year-over-year comparisons and aggregates multi-dimensional operational data to generate structured reports. The intelligent data analysis module supports automatic conversion from natural language to SQL, automatically recommending chart types such as line charts, bar charts, and pie charts based on data characteristics, and outputting results in a JSON array format containing SQL statements and chart configurations. Chart configuration information uses a standardized structure and supports direct rendering on the front end. Time range processing supports time zone conversion and time aggregation by year, quarter, month, week, and day. The conversation management mechanism supports multi-turn dialogue contexts. The multi-dimensional analysis layer integrates four major functional modules: intelligent data analysis, data analysis, trend prediction, and intelligent recommendation, enabling comprehensive intelligent analysis of operational data.

[0050] The foregoing has shown and described the basic principles and main features of the present invention. It will be apparent to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments. Therefore, the embodiments should be considered as exemplary and not restrictive. The scope of the present invention is defined by the appended claims rather than the foregoing description. Therefore, it is intended that all variations falling within the meaning and scope of the equivalents of the claims be included within the present invention.

[0051] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. A method for generating intelligent operation and maintenance reports based on a large model, characterized in that, Includes the following steps: Step S1: Collect operation and maintenance data, process the data, store it, and output a structured operation and maintenance database; the operation and maintenance data includes inspection task data, maintenance work order data, equipment alarm data, and equipment health data. Step S2: Based on the structured operation and maintenance database obtained in step S1, the functions of querying the logical structure schema of the structured operation and maintenance database, executing structured query statements (SQL), and generating report data are encapsulated into multiple standardized tool interfaces, which are then accessed by large models through the Model Context Protocol (MCP) tool call protocol. Step S3: Receive the user's natural language input, construct request parameters containing user permissions, organizational structure, and regional scope context, and conduct real-time intelligent interaction with the large model through Server-Sent Events (SSE) streaming. Step S4: Based on the time range, regional range, and content type parameters configured in the report plan, automatically aggregate multi-dimensional operation and maintenance data, and call the large model to generate a structured operation and maintenance report.

2. The intelligent operation and maintenance report generation method based on a large model according to claim 1, characterized in that, It also includes step S5: multi-dimensional intelligent analysis of operation and maintenance data, including: intelligent data collection, data analysis, trend prediction and intelligent recommendation.

3. The intelligent operation and maintenance report generation method based on a large model according to claim 1, characterized in that, The data processing and storage process in step S1 includes the following steps: S21, Data Cleaning: Remove duplicate records with the same task ID, fill null fields with default values ​​or the previous record, and convert time fields to a standard time format. S22, Data Conversion: Convert Unix timestamps to standard time format, map status enumeration values ​​to standard codes, and perform numerical normalization on health scores; S23, Data storage: Write the processed data into the corresponding table structure of the database and create indexes to optimize queries; the output includes at least a structured operation and maintenance database containing inspection task table, maintenance work order table, equipment alarm table, and equipment health table.

4. The intelligent operation and maintenance report generation method based on a large model according to claim 1, characterized in that, The standardization tools in step S2 include: Data Table Name Retrieval Tool: Retrieves a list of names for all tables in a structured operations and maintenance database, returning a result in string array format; Tool to obtain the complete structure of a specified table: Obtain the complete structure information of a specified table, including column definitions, primary keys, and unique key metadata; Tool to retrieve information about all columns in a specified table: Retrieves information about all columns in a specified table, including column name, data type, and whether it is nullable; Query execution tool: Executes query statements and returns results, supports permission filtering and parameterized queries; Tools for obtaining operation and maintenance information: Generate structured report data based on time range, region, and content type parameters; Permission verification tool: Parses user tokens to obtain organization ID, queries organization region mapping table to obtain list of accessible regions, and automatically adds region filtering conditions to SQL queries; Parameterized query tool: Uses prepared statements to replace string concatenation, performs type validation and length limit on input parameters, and escapes special characters.

5. The intelligent operation and maintenance report generation method based on a large model according to claim 1, characterized in that, Step S3 Includes the following processes: S31, Create a user session and generate a unique session identifier; Parse the user ID and organization ID from the user token to associate the user and organization; Establish a session context and create a mapping relationship in the session storage; S32, based on the organization region association table, obtain the list of accessible regions according to the user's organization permissions, and construct the data access permission context; S33, identify the interaction type and select the corresponding processing flow, where the interaction types include: SQL query type, analysis type, prediction type and recommendation question type; Use a large model to perform semantic analysis on user input, match predefined intent templates, and output intent classification results and confidence scores; S34 establishes a Server-Sent Events (SSE) streaming connection, pushes large model responses in character chunks, and renders the response content in real time on the front end.

6. The intelligent operation and maintenance report generation method based on a large model according to claim 1, characterized in that, The generation of the structured operation and maintenance report in step S4 includes the following process: S41, based on the report plan, obtain the time range, area range, device range, and content type configuration parameters; S42, calculate the month-on-month time period based on the report type, where the report types include: annual report, quarterly report, monthly report, weekly report and custom report; S43 aggregates multi-dimensional data, calls the MCP tool's interface for obtaining operation and maintenance information, and generates structured data that includes basic report information, core indicators, inspection data, work order data, equipment alarm data, and equipment health data. S44 uses structured data as input parameters to call the large model and generates a structured operation and maintenance report in natural language form. The structured operation and maintenance report includes natural language text, key indicator tables, and report metadata.

7. The intelligent operation and maintenance report generation method based on a large model according to claim 1, characterized in that, The multi-dimensional intelligent analysis in step S5 includes the following processes: Intelligent Questioning: It receives users' natural language questions, converts the natural language into SQL query statements through a large model and schema context, executes the query after SQL permission verification and returns the results, and supports multi-turn dialogue and contextual understanding; Data Analysis: Intelligent analysis of query results, using statistical methods to identify data points that deviate from the normal range, performing year-on-year and month-on-month calculations and rate of change analysis, identifying data anomalies and trends, and generating analytical conclusions and recommendations; Trend prediction: Based on historical data and time series analysis algorithms, predict future trends and output predicted values ​​and confidence intervals for the next N days or N weeks to support operation and maintenance decisions; Intelligent recommendation: Based on users' historical interaction records and behavior patterns, it counts users' high-frequency query types, areas of interest, and commonly used indicators, and performs collaborative filtering based on the analysis dimensions of similar users' interests to generate Top-N related questions and analysis dimensions.

8. A smart operation and maintenance report generation system based on a large model, characterized in that, The intelligent operation and maintenance report generation method according to any one of claims 1 to 7 includes: The operations and maintenance data acquisition layer collects operations and maintenance data, processes the data, stores it, and outputs a structured operations and maintenance database. The MCP tool layer, based on a structured operations and maintenance database, encapsulates the functions of querying the logical structure schema of the structured operations and maintenance database, executing structured query statements (SQL), and generating report data into multiple standardized tool interfaces, which are then accessed by large models through the ModelContext Protocol (MCP) tool call protocol. The intelligent interaction layer receives natural language input from users, constructs request parameters that include user permissions, organizational structure, and regional context, and performs real-time intelligent interaction with the large model through Server-Sent Events (SSE) streaming. The operation and maintenance report generation layer automatically aggregates multi-dimensional operation and maintenance data and calls large models to generate structured operation and maintenance reports based on the time range, regional range and content type parameters configured in the report plan. The multi-dimensional analysis layer provides intelligent analysis of operational data across multiple dimensions, including intelligent data retrieval, data analysis, trend prediction, and intelligent recommendation.

Images