Intelligent generation method and system of land disaster prevention and control report based on indoor and outdoor work integration

Abstract

The invention relates to the technical field of ground disaster prevention and control report generation, in particular to an intelligent ground disaster prevention and control report generation method and system based on indoor and outdoor work integration. Retrieving geological information of the associated area, and determining the type of a hidden danger point; matching a target template; generating a target template directory structure; calling a workflow agent module according to a directory structure sequence to generate a first draft text; inserting chart data into the first draft text to obtain a complete report; and performing compliance examination and optimization on the complete report, and outputting a final report. The generation system comprises a data acquisition module, an intelligent retrieval engine, a template management unit, a workflow scheduling center and an examination optimization module. The data acquisition module, the intelligent retrieval engine, the template management unit, the workflow scheduling center and the review optimization module are connected in sequence and realize data transmission; synchronous real-time updating of indoor and outdoor data is met, data dispersion and information omission are avoided, and the generation efficiency and accuracy of a ground disaster prevention and control report are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of geological disaster prevention and control report generation technology, and in particular to an intelligent generation method and system for geological disaster prevention and control reports based on integrated indoor and outdoor operations. Background Art

[0002] A geological disaster is a geological process or phenomenon caused by natural or human factors that can cause loss of life and property, damage to the environment, and so on. The temporal and spatial distribution of geological disasters is not only governed by the natural environment but also by human activities, often resulting from the interaction between these two environments.

[0003] In the existing technology, geological disaster reports are generated through manual analysis and investigation to determine where hidden dangers exist, and then a geological disaster prevention and control report is generated. This has the following technical problems: ① Field data relies on manual records, and internal data requires cross-system retrieval of geological data, which makes the generation of prevention and control reports time-consuming and prone to missing key information; ② Report templates are not dynamically matched with disaster types and project stages and require manual adjustment; ③ Budget calculations rely on empirical formulas, icon generation requires external professional software operation, and content generation lacks automated verification; ④ Field data collection and internal data utilization are separated, making it difficult to achieve real-time synchronous update of data.

[0004] It can be seen that the technical problems existing in the generation of existing geological disaster prevention and control reports can be summarized as follows: data dispersion leads to inefficiency, poor template adaptability leads to insufficient positioning accuracy of image and text fusion and easily leads to typesetting dislocation, lack of dedicated hardware for vectorized retrieval leads to low intelligence, and insufficient coordination between internal and external data. Summary of the Invention

[0005] Based on the above technical problems, the present invention provides an intelligent generation method and system for geological disaster prevention and control reports based on the integration of indoor and outdoor operations. Through modular processing, it realizes the mutual coordination and intelligent generation of geological disaster data collection, retrieval, report generation and compliance review, meets the synchronous real-time update and utilization of indoor and outdoor data, improves the intelligence and automation of geological disaster prevention and control report generation, centralizes data, processes it in real time and synchronously, avoids data dispersion and omission of key information, and improves the efficiency and accuracy of geological disaster prevention and control report generation.

[0006] The specific technical solutions are:

[0007] One of the purposes of the present invention is to provide an intelligent method for generating geological disaster prevention and control reports based on integrated indoor and outdoor operations, comprising the following steps:

[0008] S1: Collect disaster information at geological disaster risk points; obtain it in real time through GNSS equipment, INSAR monitoring system and drone photography;

[0009] S2: Based on RAG technology, disaster information is converted, features are extracted, and vectorized, and related geological information is retrieved to determine the type of hidden danger;

[0010] S3: Match the target template from the preset template library according to the type of potential risk point and requirements;

[0011] S4: parsing the structural features of the target template and generating a target template generation directory structure;

[0012] S5: Calling a workflow agent module to perform multi-source search, project cost budget, and chart generation based on the directory structure; the workflow agent module generates instruction templates based on the chapter topics and prompt words in the directory structure, and performs multi-source search, project cost budget, and chart generation based on the disaster information of the geological disaster risk points;

[0013] S6: Generate a draft text of the geological disaster prevention and control report and perform logic verification and terminology correction; the logic verification and terminology correction are performed on the draft text through a reflection mechanism;

[0014] S7: Insert the generated chart data into the draft text to form a complete report;

[0015] S8: The complete report is reviewed and optimized using the LLM model for compliance, and the final report is output. Optimization includes data consistency verification, terminology standardization, and chapter cohesion correction.

[0016] Preferably, step S1 is to obtain disaster information of geological disaster risk points in real time through GNSS equipment, INSAR monitoring system and drone photography; the disaster information includes terrain deformation data, crack distribution characteristics and on-site impact data.

[0017] Preferably, step S2 is based on RAG technology to convert disaster information into searchable text through OCR technology and extract features of geological disaster risk points, vectorize the features of geological disaster risk points through BERT embedding model, and use FAISS vector database to retrieve related regional geological information based on the vectorized geological disaster risk point features to determine the type of risk; the related regional geological information includes regional hydrological data, historical disaster records, transportation network distribution and geotechnical engineering parameters.

[0018] Preferably, the target template includes an emergency investigation report, a prevention and control project proposal, and a standardized on-site investigation form.

[0019] Preferably, the multi-source search includes linked knowledge base search, network search and engineering database query; the engineering cost budget is generated by a cost estimation model; and the chart generation includes generating geological profiles and monitoring data visualization charts.

[0020] Preferably, the cost estimation model is constructed by the following steps:

[0021] S51: Extract historical project cost data, including material unit price, labor cost and machine shift data;

[0022] S52: Based on the data extracted in step S51, establish a project quantity-cost mapping relationship based on the LSTM neural network;

[0023] S53: Based on the engineering quantity-cost mapping relationship established in S52, the model parameters are dynamically adjusted in combination with the regional economic index to obtain the cost estimation model.

[0024] Preferably, the graph generation comprises the following steps:

[0025] S511: Analyze the geographic coordinates of the geological disaster risk points and generate a three-dimensional geological model base map;

[0026] S512: Overlaying InSAR deformation monitoring data on the base map and rendering to form a deformation cloud map;

[0027] S513: Mark the high-risk area boundaries and monitoring point numbers on the deformation cloud map to generate a chart.

[0028] The second purpose of the present invention is to provide an intelligent generation system for geological disaster prevention and control reports based on the integration of indoor and outdoor work, including:

[0029] Data acquisition module, used to obtain disaster information of geological disaster risk points in real time through GNSS equipment, INSAR monitoring system and drone photography;

[0030] Intelligent retrieval engine, used to vectorize disaster information through RAG technology and retrieve relevant regional geological information from the geological database; connected with the data acquisition module;

[0031] Template management unit, used to store preset template library and matching rule library; connected with intelligent search engine;

[0032] The workflow scheduling center (workflow agent module) is used to perform AI workflow orchestration and includes a directory generation unit for generating a directory structure for a target template, a prompt assembler for generating and storing prompt words, a cost estimator for calculating a project cost budget, a content generation and iteration unit for generating a draft text based on the directory structure, and a graphic-text fusion unit for inserting the generated chart data into the draft text; the directory generation unit, prompt assembler, cost estimator, content generation and iteration unit, and graphic-text fusion unit are sequentially connected to realize data transmission; and are connected to the template management unit;

[0033] Review and optimization module, used for compliance review and optimization of complete reports; connected with the workflow scheduling center;

[0034] Among them, the data acquisition module, intelligent search engine, template management unit, workflow scheduling center and review optimization module are connected in sequence to realize data transmission.

[0035] Preferably, the data acquisition module includes:

[0036] Integrated GNSS receiver, configured to obtain coordinate data of hidden danger points in real time;

[0037] a drone aerial photography unit configured to collect on-site impact data;

[0038] Ground sensor network,using LoRa modules to transmit crack monitoring data.

[0039] Preferably, the intelligent search engine includes:

[0040] OCR processor for converting unstructured geological documents into searchable text;

[0041] BERT embedding model, used to generate 768-dimensional feature vectors;

[0042] FAISS vector database, configured for multimodal data retrieval based on cosine similarity;

[0043] Preferably, the workflow scheduling center includes:

[0044] Multi-core processors for running the cue word assembler and illustration generation agent in parallel;

[0045] FPGA chip, used for dynamic matching template type encoding;

[0046] Cost estimation coprocessor with built-in LSTM neural network unit;

[0047] Preferably, the image-text fusion unit includes:

[0048] OpenCV image recognition module, used to locate the placeholder coordinates with a positioning error of ≤ 5 pixels;

[0049] 3D geological model rendering engine, used to generate deformation cloud maps superimposed with InSAR deformation data.

[0050] Compared with the prior art, the technical effects created by the present invention are embodied in:

[0051] Automatic generation of geological disaster prevention and control reports such as geological disaster investigation reports and project proposals is achieved through field multi-source data collection, RAG enhanced retrieval, template dynamic matching, and intelligent workflow orchestration. By integrating GNSS / INSAR and other monitoring data, LSTM budget models, and large-scale model reflection and optimization mechanisms, closed-loop management of geological disaster investigation, inspection, cost estimation, and automatic report generation is achieved. This effectively solves technical problems such as the inefficiency caused by data dispersion in traditional methods, poor template adaptability leading to insufficient positioning accuracy for image and text fusion and easy layout dislocation, lack of dedicated hardware for vectorized retrieval leading to low intelligence, and insufficient coordination between field and field data leading to high manual dependence. This improves the intelligence and automation of geological disaster prevention and control report generation, avoids data dispersion and omission of key information, and improves the efficiency and accuracy of geological disaster prevention and control report generation. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] In order to facilitate those skilled in the art to fully understand the technical solution of the present invention, the present invention is created by combining the content of the technical solution and the provided drawings, and the following description is made of the provided drawings. It is obvious that the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0053] Figure 1 A flow chart of the generation method.

[0054] Figure 2 To generate the framework structure diagram of the system.

[0055] Figure 3 This is the framework logic diagram of the data acquisition module.

[0056] Figure 4 This is the workflow diagram of the data acquisition module.

[0057] Figure 5 This is the framework logic diagram of the intelligent search engine;

[0058] Figure 6 This is the workflow diagram of the intelligent search engine;

[0059] Figure 7 It is the framework logic diagram of the template management unit;

[0060] Figure 8 It is the workflow diagram of the template management unit;

[0061] Figure 9 Generate a skeleton logic diagram of the unit for the directory;

[0062] Figure 10 Generate a workflow diagram for the catalog unit;

[0063] Figure 11 A framework logic diagram for content generation and iteration units;

[0064] Figure 12 A workflow diagram for the content generation and iteration unit;

[0065] Figure 13 This is the framework logic diagram of the Prompt assembler;

[0066] Figure 14 Workflow diagram for Prompt assembler;

[0067] Figure 15 This is the framework logic diagram of the cost estimator;

[0068] Figure 16 This is the workflow diagram of the cost estimator;

[0069] Figure 17 This is the framework logic diagram of the image-text fusion unit;

[0070] Figure 18 This is the workflow diagram of the image-text fusion unit;

[0071] Figure 19 Optimize the framework logic diagram of the unit for review;

[0072] Figure 20 Workflow diagram for the review optimization unit. DETAILED DESCRIPTION

[0073] In order to facilitate those skilled in the art to correctly understand the present invention and enable those skilled in the art to fully understand the technical content of the present invention, the technical solution of the present invention is further explained below in conjunction with specific implementation methods, but this explanation does not limit the scope of protection required for the present invention. The scope of protection of the present invention by those skilled in the art cannot be limited to the following explanations. Any equivalent replacement or change made by any those skilled in the art or persons familiar with the technology in this field, on the basis of the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

[0074] like Figure 1 As shown in the figure, the intelligent generation method of geological disaster prevention and control report based on the integration of indoor and outdoor work includes the following steps:

[0075] S1: Use GNSS equipment, INSAR monitoring systems, and drone photography technology to obtain real-time disaster information such as terrain deformation data, crack distribution characteristics, and on-site camera data at potential disaster sites. Continuously monitor deformation at potential disaster sites through the INSAR monitoring system and synchronize the data to a cloud database in real time.

[0076] S2: Based on the RAG (Retrieval Enhanced Generation) technology, the disaster information is transformed, feature extracted, and vectorized through embedding, and the associated geological information is retrieved to determine the type of hidden danger. The implementation of the RAG technology includes:

[0077] (a) Convert unstructured documents in geological databases into searchable text using OCR technology;

[0078] (b) Use the BERT embedding model to vectorize the hidden danger point features and database content;

[0079] (c) Use the FAISS vector database to realize similarity retrieval of multimodal data.

[0080] S3: Match target templates from the preset template library based on the type of hidden danger point and user needs; the template types in the preset template library include emergency investigation reports, prevention and control project proposals and standardized on-site survey forms. The matching rules for matching target templates are dynamically adjusted based on the disaster level and project stage; dynamic weight adjustment: frequently used templates are prioritized based on the user role (monitor / engineer).

[0081] S4: Analyze the structural features of the selected target template, automatically generate a cover page and a multi-level directory outline that complies with the "Geological Hazard Prevention and Control Code", and logically bind the directory hierarchy to the chapters of the template to obtain the directory structure;

[0082] S5: Calling the workflow agent module (agent module or AI workflow scheduling center) according to the directory structure to perform prompt word assembly, multi-source search, project cost budget and chart generation; the workflow agent module generates instruction templates based on the chapter topics and prompt words in the directory structure, and performs multi-source search, project cost budget and chart generation based on the disaster information of the geological disaster risk points; multi-source search: linking knowledge base search, network search and engineering database query; data calculation: generating a project cost budget table through a cost estimation model;

[0083] The cost estimation model is constructed by the following steps:

[0084] S51: Extract historical project cost data, including material unit price, labor cost and machine shift data;

[0085] S52: Based on the data extracted in step S51, establish a project quantity-cost mapping relationship based on the LSTM neural network;

[0086] S53: Based on the engineering quantity-cost mapping relationship established in S52, the model parameters are dynamically adjusted in combination with the regional economic index to obtain the cost estimation model.

[0087] Chart generation: Call the illustration generation agent to output geological profiles and monitoring data visualization charts;

[0088] The running process of the illustration generation agent includes:

[0089] (a) Analyze the geographical coordinates of the hidden danger points to generate a three-dimensional geological model base map;

[0090] (b) Deformation cloud map rendered by overlaying InSAR deformation monitoring data on the base map;

[0091] (c) Automatically mark the boundaries of high-risk areas and monitoring point numbers on the deformation cloud map.

[0092] S6: Generate a first draft of the geological disaster prevention and control report and perform logic verification and terminology correction. The logic verification and terminology correction are performed on the first draft through a reflection mechanism. Specifically, the workflow components are called in the order of the directory, the search results and calculation data are input into the large model to generate a first draft, and the text is logically verified and terminology corrected through the reflection mechanism.

[0093] S7: Insert the generated chart data into the draft text to form a complete report; specifically:

[0094] Insert illustrations and tables generated by the chart generation module into the specified chapter, with their coordinates positioned based on the placeholder markers preset in the template;

[0095] Placeholder positioning: The template presets the illustration insertion position (such as "geological cross-section"), and automatically matches the coordinates to generate a 3D geological model;

[0096] S8: The complete report is reviewed and optimized using the LLM model for compliance, and the final report is output. Optimization includes data consistency verification, terminology standardization, and chapter cohesion correction.

[0097] Large model verification report terminology (such as the terminology library of the "Geological Hazard Prevention and Control Code") and data consistency (correlation between inspection records and budget tables).

[0098] like Figures 2 to 20 As shown in the figure, the intelligent generation system of geological disaster prevention and control reports based on the integration of indoor and outdoor work includes:

[0099] Data acquisition module (such as Figure 2 The perception layer shown, Figure 3The framework logic diagram shown is used to obtain disaster information of geological disaster risk points in real time through GNSS equipment, INSAR monitoring system and drone photography; the data acquisition module is configured to obtain real-time coordinate data of risk points through the integrated GNSS receiver, drone aerial photography unit, configured to collect on-site impact data and ground sensor network, used to obtain real-time dynamic data of risk points; the GNSS receiver and drone are connected via Bluetooth / Wi-Fi to transmit the coordinates of risk points in real time at a transmission frequency of 2.4GHz; the ground sensor network uses LoRa module to upload crack monitoring data to the cloud, with a transmission distance of ≥1km. The workflow of this module is as follows Figure 4 shown.

[0100] Intelligent search engines (such as Figure 2 The data layer shown, such as Figure 5 The framework logic diagram shown is used to vectorize disaster information through RAG technology and retrieve related regional geological information from the geological database; it is connected to the data acquisition module; a hybrid retrieval system of the RAG architecture is deployed to support structured database queries and unstructured document vector retrieval; the server cluster of the RAG architecture has a built-in OCR processor and a FAISS vector database, the OCR processor is connected to the geological database, and the FAISS library receives the hidden danger point feature vector generated by the BERT model; the FAISS vector database supports cosine similarity retrieval, and the index dimension is 768 dimensions; the BERT model is the BERT-base version, and the embedding vector length is 768. The OCR processor is used to convert unstructured geological documents into searchable text; the BERT embedding model is used to generate a 768-dimensional feature vector; the FAISS vector database can be configured to perform multimodal data retrieval based on cosine similarity; the workflow is as follows Figure 6 shown.

[0101] Template management unit (such as Figure 2 As shown in the processing layer and application layer, Figure 7 The framework logic diagram shown in the figure is used to store the preset template library and matching rule library; connect with the intelligent search engine; have a built-in template version traceability mechanism; SSD solid-state hard disk stores the standardized template library and has a built-in version traceability chip; the FPGA chip dynamically calls the matching rule according to the disaster type code (such as landslide = 01); the workflow is as follows Figure 8 shown.

[0102] Workflow scheduling center (workflow agent module, such as Figure 2 The processing layer is used for AI workflow orchestration and includes a directory generation unit for generating a directory structure for a target template (the framework logic diagram is shown in FIG Figure 9 As shown, the workflow is as follows Figure 10 As shown), Prompt assembler for generating and storing prompt words (framework logic diagram as shown Figure 13 As shown, the workflow is as follows Figure 14 As shown), cost estimator for calculating engineering cost budget (framework logic diagram as shown Figure 15 As shown, the workflow is as follows Figure 16 As shown), a content generation and iteration unit for generating a first draft text according to the directory structure (the framework logic diagram is as shown Figure 11 As shown, the workflow is as follows Figure 12 As shown, a graphic-text fusion unit is used to insert the generated chart data into the draft text; the directory generation unit, prompt assembler, cost estimator, content generation and iteration unit, and graphic-text fusion unit are sequentially connected to implement data transmission; and the unit is connected to the template management unit. A multi-core processor runs the prompt word assembler and the illustration generation agent. The cost estimator coprocessor is connected via a PCIe bus. The cost estimator coprocessor has a built-in LSTM neural network unit and supports FP16 half-precision calculations. The placeholder locator uses the OpenCV image recognition algorithm with a positioning error of ≤5px. The multi-core processor is used to run the prompt word assembler and the illustration generation agent in parallel; an FPGA chip is used to dynamically match the template type code; OpenCV image recognition is used to locate the placeholder coordinates with a positioning error of ≤5 pixels; and a 3D geological model rendering engine is used to generate deformation cloud maps superimposed with InSAR deformation data.

[0103] Review optimization module (framework logic diagram as shown Figure 19 As shown, the workflow is as follows Figure 20 The report's graphical locator automatically inserts charts based on placeholder coordinates (e.g., X=120px, Y=300px).

[0104] Among them, the data acquisition module, intelligent search engine, template management unit, workflow scheduling center and review optimization module are connected in sequence to realize data transmission.

[0105] The invention acquires dynamic data of disaster sites in real time through GNSS, INSAR, and drone oblique photography technologies; realizes multimodal association retrieval of geological data based on RAG enhanced retrieval technology, BERT embedding model, and FAISS vector library; can automatically select templates such as emergency investigation reports and project proposals according to disaster level and engineering stage; adopts an intelligent agent workflow engine of an AI orchestration system with integrated prompt word assembly, multi-source retrieval, budget calculation, and chart generation modules, and realizes text logic verification and compliance review through reflection mechanism; realizes real-time synchronization of field data through Bluetooth / Wi-Fi / LoRa multi-protocol integration; FPGA chips dynamically match templates, and multi-core processors parallelly schedule intelligent agent tasks to optimize processing efficiency; placeholder locators automatically align images and text, reducing manual adjustment time by 70% and simplifying human-computer interaction.

[0106] Other matters not covered by the present invention may be achieved by conventional technical means with reference to the prior art or common knowledge known to those skilled in the art.

Claims

1. The intelligent generation method of geological disaster prevention and control reports based on the integration of indoor and outdoor work is characterized by: The following steps are involved: S1: Collect disaster information of geological disaster risk points; S2: Vectorize the disaster information based on RAG technology, retrieve related geological information, and determine the type of hidden danger; S3: Match the target template from the preset template library according to the type of potential risk point and requirements; S4: parsing the structural features of the target template and generating a target template generation directory structure; S5: Call the workflow agent module according to the directory structure to perform multi-source retrieval, project cost budgeting and chart generation; S6: Generate the first draft of the geological disaster prevention report and perform logic verification and terminology correction; S7: Insert the generated chart data into the draft text to form a complete report; S8: The complete report is reviewed and optimized through the LLM model for compliance, and the final report is output.

2. The intelligent generation method of geological disaster prevention and control report based on the integration of indoor and outdoor work as claimed in claim 1 is characterized in that: The step S1 is to obtain disaster information of geological disaster risk points in real time through GNSS equipment, INSAR monitoring system and drone photography; the disaster information includes terrain deformation data, crack distribution characteristics and on-site impact data.

3. The intelligent generation method of geological disaster prevention and control report based on the integration of indoor and outdoor work as claimed in claim 1 is characterized in that: The step S2 is based on RAG technology to convert the disaster information into searchable text through OCR technology and extract the features of geological disaster risk points, vectorize the features of geological disaster risk points through BERT embedding model, and use FAISS vector database to retrieve related regional geological information based on the vectorized geological disaster risk point features to determine the type of risk; the related regional geological information includes regional hydrological data, historical disaster records, transportation network distribution and geotechnical engineering parameters.

4. The method for intelligently generating geological disaster prevention and control reports based on the integration of indoor and outdoor work as claimed in claim 1 is characterized in that: The target templates include emergency investigation reports, prevention and control project proposals and standardized on-site survey forms.

5. The intelligent generation method of geological disaster prevention and control report based on the integration of indoor and outdoor work as claimed in claim 1 is characterized in that: The multi-source search includes linked knowledge base search, network search and engineering database query; the engineering cost budget is generated by a cost estimation model; and the chart generation includes generating geological profiles and monitoring data visualization charts.

6. The method for intelligently generating geological disaster prevention and control reports based on the integration of indoor and outdoor work as claimed in claim 5 is characterized in that: The cost estimation model is constructed by the following steps: S51: Extract historical project cost data, including material unit price, labor cost and machine shift data; S52: Based on the data extracted in step S51, establish a project quantity-cost mapping relationship based on the LSTM neural network; S53: Based on the engineering quantity-cost mapping relationship established in S52, the model parameters are dynamically adjusted in combination with the regional economic index to obtain the cost estimation model.

7. The method for intelligently generating geological disaster prevention and control reports based on the integration of indoor and outdoor work as claimed in claim 5 is characterized in that: The chart generation comprises the following steps: S511: Analyze the geographic coordinates of the geological disaster risk points and generate a three-dimensional geological model base map; S512: Overlaying InSAR deformation monitoring data on the base map and rendering to form a deformation cloud map; S513: Mark the high-risk area boundaries and monitoring point numbers on the deformation cloud map to generate a chart.

8. Based on the intelligent generation system of geological disaster prevention and control reports integrated with indoor and outdoor work, the characteristics are: include: Data acquisition module, used to obtain disaster information of geological disaster risk points in real time through GNSS equipment, INSAR monitoring system and drone photography; Intelligent retrieval engine, used to vectorize disaster information through RAG technology and retrieve relevant regional geological information from the geological database; connected with the data acquisition module; Template management unit, used to store preset template library and matching rule library; Connect with intelligent search engines; A workflow scheduling center is used to perform AI workflow orchestration, and includes a directory generation unit for generating a directory structure for a target template, a prompt assembler for generating and storing prompt words, a cost estimator for calculating a project cost budget, a content generation and iteration unit for generating a draft text based on the directory structure, and a graphic-text fusion unit for inserting the generated chart data into the draft text; the directory generation unit, prompt assembler, cost estimator, content generation and iteration unit, and graphic-text fusion unit are sequentially connected to realize data transmission; and the center is connected to a template management unit; Review and optimization module, used for compliance review and optimization of complete reports; connected with the workflow scheduling center; Among them, the data acquisition module, intelligent search engine, template management unit, workflow scheduling center and review optimization module are connected in sequence to realize data transmission.

9. The intelligent generation system for geological disaster prevention and control reports based on the integration of indoor and outdoor work as claimed in claim 8 is characterized in that: The data acquisition module includes: Integrated GNSS receiver, configured to obtain coordinate data of hidden danger points in real time; a drone aerial photography unit configured to collect on-site impact data; Ground sensor network,using LoRa modules to transmit crack monitoring data.

10. The intelligent generation system for geological disaster prevention and control reports based on the integration of indoor and outdoor work as claimed in claim 8 is characterized in that: The intelligent search engine include: OCR processor for converting unstructured geological documents into searchable text; BERT embedding model, used to generate 768-dimensional feature vectors; FAISS vector database, configured for multimodal data retrieval based on cosine similarity; and / or, The workflow scheduling center includes: Multi-core processors for running the cue word assembler and illustration generation agent in parallel; FPGA chip, used for dynamic matching template type encoding; Cost estimation coprocessor with built-in LSTM neural network unit; and / or, The image-text fusion unit includes: OpenCV image recognition module, used to locate the placeholder coordinates with a positioning error of ≤ 5 pixels; 3D geological model rendering engine, used to generate deformation cloud maps superimposed with InSAR deformation data.

Images