An engineering construction management and control method based on a large model and an intelligent agent

Abstract

The application relates to the technical field of engineering construction supervision, and provides an engineering construction management and control method and an intelligent agent based on a large model, which comprises the following steps: receiving construction supervision request information, performing semantic analysis on the construction supervision request information through a large language model deployed in a large model inference server, identifying a construction division type and a supervision scene; retrieving corresponding supervision element data and specification clause numbers from a supervision element database according to the identification result, and performing associated reasoning based on the large language model to generate a structured supervision list; sending the supervision list to a mobile terminal through an MCP communication adaptation unit and receiving supervision progress data; calculating a supervision matching degree according to the supervision progress data, and generating hierarchical early warning information when the matching degree is lower than a preset early warning threshold. The application can realize intelligent matching of construction supervision elements and dynamic early warning of the supervision process, and improve the intelligent level and supervision efficiency of engineering construction supervision.

Description

Technical Field

[0001] This invention belongs to the field of engineering construction supervision technology, specifically involving an engineering construction management and control method and intelligent agent based on a large model. Background Technology

[0002] Construction projects are characterized by large scale, diverse construction sub-projects, and high technical requirements, typically encompassing dozens of core construction sub-projects. These sub-projects differ significantly in quality standards, safety regulations, and specific technical requirements, resulting in complex and specialized regulatory elements. Traditional construction supervision relies heavily on manual verification against multiple professional specifications, making it difficult for supervisors to fully grasp the professional clauses of each sub-project and prone to oversights. Furthermore, construction sites are often located in outdoor environments with unstable network signals, and existing general-purpose monitoring tools lack offline operation capabilities and dynamic communication adaptation mechanisms, failing to meet the needs of complex scenarios. For specific requirements such as environmental protection and flood control, existing solutions largely rely on manual judgment, lacking intelligent and autonomous compliance verification capabilities. Under the background of lifelong responsibility for construction quality, manual record-keeping methods also suffer from data loss, non-standardization, and difficulty in traceability. Existing construction supervision systems are mostly centered on system integration, with intelligent modules serving only as auxiliary components, failing to form an autonomous decision-making system centered on intelligent agents, making it difficult to achieve autonomous matching of regulatory elements, autonomous compliance verification, and automatic generation of tiered early warnings. Summary of the Invention

[0003] This invention addresses the problems of low intelligence and easy oversight in existing engineering construction supervision by providing an engineering construction management and control method and intelligent agent based on a large model, which enables autonomous matching of regulatory elements, compliance judgment, and automatic generation of hierarchical early warnings.

[0004] A method for engineering construction management and control based on a large model, the method comprising the following steps: S1. Receive construction supervision request information, wherein the construction supervision request information includes at least one or more of the following: construction sub-project information and construction supervision text information; S2. The construction supervision request information is semantically parsed by the large language model deployed in the large model inference server to identify the corresponding construction sub-type and supervision scenario, and to determine the construction sub-type. The construction sub-type includes at least a key construction sub-type or a normal construction sub-type. S3. Based on the identified construction sub-types, retrieve the corresponding regulatory element data and standard clause numbers from the regulatory element database; S4. Based on the large language model, perform association reasoning on the retrieved regulatory element data, match the association relationship between the regulatory element data and the normative clause number, and generate a structured regulatory list; S5. Send the structured regulatory list to the mobile terminal through the MCP communication adapter unit, and receive the regulatory progress data returned by the mobile terminal; S6. Based on the regulatory progress data, count the number of completed regulatory items and calculate the matching degree between the number of completed regulatory items and the total number of preset regulatory elements; S7. Compare the matching degree with the preset warning threshold of the corresponding construction section. When the matching degree is lower than the preset warning threshold, generate a graded warning instruction and send the warning information to the mobile terminal.

[0005] Furthermore, the large language model mentioned in step S2 is a large language model based on the Transformer architecture. The large language model is fine-tuned and trained using a regulatory element dataset. The regulatory element dataset includes at least one or more of the following: construction diversion data, earthwork excavation data, foundation treatment data, concrete engineering data, metal structure installation data, and electromechanical equipment installation data.

[0006] Furthermore, step S2 specifically includes the following steps: S21. Extract construction sub-keywords from the construction supervision request information; S22. Match the construction sub-section keywords with the sub-section information in the regulatory element database; S23. Determine the corresponding construction sub-type based on the matching results.

[0007] Furthermore, the regulatory element database mentioned in step S3 includes at least one or more of the following: a section information table, a regulatory element table, a regulatory clause table, and a historical case table. The division information table includes at least one or more of the following: division ID, division name, parent division ID, and division level. The regulatory element table shall include at least one or more of the following: element ID, element name, element category, judgment criteria, inspection method, and regulatory clause number; The specification clause table shall include at least one or more of the following: clause ID, specification name, clause number, clause content, and scope of application; The historical case table shall include at least one or more of the following: case ID, problem description, solution, involved divisions, and involved elements.

[0008] Furthermore, the structured regulatory list mentioned in step S4 is classified according to the construction supervision dimensions, which include at least one or more of the following: technical parameters, quality standards, safety specifications, process nodes, environmental protection standards, and flood control safety.

[0009] Furthermore, the formula for calculating the matching degree in step S6 is as follows: Matching degree = (Number of completed regulatory items / Total number of preset regulatory elements) × 100%.

[0010] Furthermore, the graded early warning instruction in step S7 sets differentiated preset early warning thresholds according to the type of construction sub-section. When the matching degree of critical construction sub-section is <98%, a first-level early warning is generated, and when the matching degree of ordinary construction sub-section is <95%, a second-level early warning is generated.

[0011] An intelligent agent for engineering construction management and control based on a large model, wherein the intelligent agent is used to implement the above method. The large-scale model inference server includes a large-scale model inference core module, a regulatory element intelligent matching module, a hierarchical early warning decision module, and a data autonomous storage module. It is used to deploy a large language model, perform semantic parsing on construction supervision request information and identify construction sub-types and supervision scenarios, retrieve and match corresponding regulatory element data and standard clause numbers from the regulatory element database, perform correlation inference on regulatory element data based on the large language model and generate a structured supervision list, count the number of completed supervision items and calculate the matching degree, generate hierarchical early warning instructions based on the comparison result of the matching degree and the preset early warning threshold, and persistently store various types of data throughout the supervision process. The MCP communication adapter unit includes a protocol switching module, an offline caching module, and an encrypted transmission module. It is used to realize data interaction and command transmission between the large model inference server and the mobile terminal. It adaptively switches the communication protocol according to the network status, performs local caching of transmitted data in the offline state and resumes interrupted transmission after network recovery, and encrypts and protects all interactive data to ensure the continuity, security and integrity of data transmission. The mobile terminal includes a construction sub-project selection module, a regulatory element display module, an early warning push module, an intelligent Q&A interaction module, a special data entry module, and a data visualization module. It is used to send construction supervision request information to the large model inference server, receive and display structured supervision lists, enter and transmit supervision progress data and special supervision data, receive tiered early warning instructions and display early warning information, support engineering supervision question interaction in natural language, and visualize supervision completion, early warning distribution, and compliance statistics in chart form.

[0012] Furthermore, the data self-storage module uses a MongoDB database as its storage medium; The MCP communication adaptation unit adopts a dual-protocol communication mechanism based on RESTful API and WebSocket. The protocol switching module automatically switches between HTTP / HTTPS protocol and WebSocket protocol according to the network latency threshold. The offline caching module uses SQLite local database to store offline transmission data and performs data synchronization or breakpoint resumption after the network is restored. The encrypted transmission module uses AES-256-GCM encryption algorithm to encrypt the transmission of communication data.

[0013] A computer device, comprising: At least one processor; and, A memory communicatively connected to the at least one processor; wherein, The memory has instructions that can be executed by the at least one processor, which, when executed by the at least one processor, enables the at least one processor to perform the above-described method.

[0014] A non-volatile computer storage medium stores computer-executable instructions configured to implement the above-described method.

[0015] The technical solution of this invention can achieve the following beneficial effects: (1) Intelligent identification and matching of construction supervision elements: By using a large language model to perform semantic analysis on construction supervision request information and combining it with the supervision element database for element retrieval and association reasoning, a structured supervision list can be automatically generated, realizing automatic matching of construction supervision elements and association with standard clauses, thereby reducing the reliance on human experience judgment.

[0016] (2) Improved stability of data interaction in complex construction scenarios: By setting up the MCP communication adapter unit, adaptive switching between different communication protocols can be achieved. Combined with the offline caching mechanism, the continuous transmission and caching of monitoring data can be maintained in unstable network or offline environments, thereby improving the applicability of the system in field construction scenarios.

[0017] (3) Improved construction supervision efficiency: By automatically generating a structured supervision list containing the basis of the standard clauses, supervisors can directly carry out inspection work based on the list, reducing the time spent manually consulting construction standards and sorting out supervision elements, thereby improving the efficiency of construction supervision work.

[0018] (4) Enhanced traceability of construction supervision data: By uniformly storing and managing construction supervision records, supervision progress data and related reasoning data, the continuous accumulation and related management of supervision process data can be realized, which is conducive to subsequent traceability of engineering quality responsibility and supervision information query, and improves the integrity and manageability of supervision data. Attached Figure Description

[0019] Figure 1 This is a flowchart of the engineering construction management and control method based on a large model in an embodiment of the present invention; Figure 2 This is a schematic diagram of the three-stage training process of the large model in an embodiment of the present invention; Figure 3 This is a schematic diagram of the structure of the engineering construction management and control intelligent agent based on a large model in an embodiment of the present invention; Figure 4 This is a schematic diagram of the architecture of the MCP communication adapter unit in an embodiment of the present invention; Figure 5 This is a schematic diagram of the architecture of the mobile terminal in an embodiment of the present invention; Figure 6 This is a schematic diagram of the workflow of the engineering construction management and control intelligent agent based on a large model in an embodiment of the present invention. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will now be described in detail and completely with reference to the accompanying drawings. It should be noted that the described embodiments are merely some examples of the present invention and do not represent all possible implementations of the present invention. Any other implementations obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the protection scope of the present invention.

[0021] Example 1 like Figure 1 As shown in the figure, this embodiment provides a method for engineering construction management and control based on a large model. The method includes the following steps: S1. Receive construction supervision request information, wherein the construction supervision request information includes at least one or more of the following: construction sub-project information and construction supervision text information.

[0022] S2. The construction supervision request information is semantically parsed by a large language model deployed in a large model inference server to identify the corresponding construction sub-type and supervision scenario, and to determine the construction sub-type, wherein the construction sub-type includes at least a key construction sub-type or a general construction sub-type.

[0023] Specifically, step S2 includes the following steps: S21. Extract construction sub-keywords from the construction supervision request information; S22. Match the construction sub-section keywords with the sub-section information in the regulatory element database; S23. Determine the corresponding construction sub-type based on the matching results.

[0024] Specifically, the large language model is a large language model based on the Transformer architecture. The large language model is fine-tuned and trained using a regulatory element dataset. The regulatory element dataset includes at least one or more of the following: construction diversion data, earthwork excavation data, foundation treatment data, concrete engineering data, metal structure installation data, and electromechanical equipment installation data.

[0025] More specifically, the large language model is a generalized, thousand-question large language model based on the Transformer architecture, and the parameter scale of the large language model is 8 billion.

[0026] Specifically, such as Figure 2 As shown, the large language model adopts a three-stage training method: During the fine-tuning phase of supervision, the labeled regulatory element dataset is used to fine-tune the parameters of the Tongyi Thousand Questions basic model and establish the correlation between regulatory elements and regulatory clauses. In the reinforcement learning phase, the model's interaction logic and decision-making ability in regulatory scenarios are optimized through reinforcement learning based on human feedback. During the incremental learning phase, newly collected regulatory case data is used periodically to incrementally update the model that has completed the first two phases of training, maintaining the timeliness of the model's knowledge.

[0027] S3. Based on the identified construction sub-type, retrieve the corresponding regulatory element data and standard clause number from the regulatory element database.

[0028] Specifically, the regulatory element database should include at least one or more of the following: a section information table, a regulatory element table, a regulatory clause table, and a historical case table. The division information table should include at least one or more of the following: division ID, division name, parent division ID, and division level; The regulatory element table shall include at least one or more of the following: element ID, element name, element category, judgment criteria, inspection method, and regulatory clause number; The specification clause table shall include at least one or more of the following: clause ID, specification name, clause number, clause content, and scope of application; The historical case table should include at least one or more of the following: case ID, problem description, solution, involved divisions, and involved elements.

[0029] More specifically, the method for constructing a regulatory element database includes the following steps: S31. Design and create data tables: Project Division Information Table: - division_id: Unique identifier for the division - division_name: Division name - parent_id: Parent branch ID - level: Branch level (levels 1-3) - is_critical: Whether the part is critical (0 / 1) Supervision Elements Table: - element_id: Unique identifier for the element - element_name: element name - element_type: Element category (technical parameters / quality standards / safety specifications, etc.) - judgment_criteria: Judgment criteria - inspection_method: Inspection method - specification_id: ID of the associated specification clause Specification Clauses: - clause_id: Unique identifier for the clause - specification_name: Specification name - clause_number: Clause number - clause_content: Clause content - applicable_scope: Scope of application Historical Cases Table: - case_id: Unique identifier for the case - problem_description: Problem description - solution: solution - related_divisions: related divisions - related_elements: related elements S32. Establish relationships between data tables: A division can contain multiple regulatory elements; One regulatory element can be associated with multiple regulatory provisions; A historical case can involve multiple parts and elements.

[0030] S4. Based on the large language model, perform association reasoning on the retrieved regulatory element data, match the association relationship between the regulatory element data and the normative clause number, and generate a structured regulatory list.

[0031] Specifically, the structured regulatory list is classified according to construction supervision dimensions, which include at least one or more of the following: technical parameters, quality standards, safety specifications, process nodes, environmental protection standards, and flood control safety.

[0032] S5. Send the structured regulatory list to the mobile terminal through the MCP communication adapter unit, and receive the regulatory progress data returned by the mobile terminal.

[0033] S6. Based on the regulatory progress data, count the number of completed regulatory items and calculate the matching degree between the number of completed regulatory items and the total number of preset regulatory elements.

[0034] Specifically, the formula for calculating the matching degree is as follows: Matching degree = (Number of completed regulatory items / Total number of preset regulatory elements) × 100%.

[0035] S7. Compare the matching degree with the preset warning threshold of the corresponding construction section. When the matching degree is lower than the preset warning threshold, generate a graded warning instruction and send the warning information to the mobile terminal.

[0036] Specifically, the graded early warning instruction sets differentiated preset early warning thresholds according to the type of construction sub-section. When the matching degree of critical construction sub-section is <98%, a first-level early warning is generated, and when the matching degree of ordinary construction sub-section is <95%, a second-level early warning is generated.

[0037] Specifically, the implementation method of early warning decision-making includes the following steps: S71. Matching degree calculation: Python def calculate_completion_rate(user_progress, required_elements): completed_count = sum(1 for element in user_progress if element['status'] == 'completed') total_count = len(required_elements) return (completed_count / total_count) * 100 ``` S72, Threshold Judgment: Python def determine_warning_level(completion_rate, is_critical_division): if is_critical_division: return 'level1' if completion_rate<98 else 'normal' else: return 'level2' if completion_rate<95 else 'normal' ``` S73, Warning Generation: Python def generate_warning(warning_level, division_name, completion_rate): warning_message = f"{division_name}completion rate{completion_rate}%" if warning_level == 'level1': warning_message += "Below the 98% threshold for critical segments, please conduct further checks immediately" elif warning_level == 'level2': warning_message += "Below the 95% threshold for normal divisions, please complete as soon as possible" return warning_message Example 2 like Figure 3 As shown, this embodiment provides an intelligent agent for engineering construction management and control based on a large model. The intelligent agent includes: The large-scale model inference server includes a large-scale model inference core module, a regulatory element intelligent matching module, a hierarchical early warning decision module, and a data autonomous storage module. It is used to deploy a large language model, perform semantic parsing on construction supervision request information, identify construction sub-types and supervision scenarios, retrieve matching regulatory element data and standard clause numbers from the regulatory element database, perform correlation inference on the regulatory element data based on the large language model and generate a structured supervision list, count the number of completed supervision items and calculate the matching degree, generate hierarchical early warning instructions based on the comparison result of the matching degree and the preset early warning threshold, and persistently store various types of data throughout the supervision process.

[0038] Specifically, the data self-storage module uses MongoDB database as the storage medium.

[0039] The MCP communication adapter unit includes a protocol switching module, an offline caching module, and an encrypted transmission module. It is used to realize data interaction and command transmission between the large model inference server and the mobile terminal. It adaptively switches the communication protocol according to the network status, caches the transmitted data locally in the offline state and resumes the transmission after the network is restored. It encrypts and protects all interactive data to ensure the continuity, security and integrity of data transmission.

[0040] Specifically, the MCP communication adaptation unit adopts a dual-protocol communication mechanism based on RESTful API and WebSocket. The protocol switching module automatically switches between HTTP / HTTPS protocol and WebSocket protocol according to the network latency threshold. The offline caching module uses SQLite local database to store offline transmission data and performs data synchronization or breakpoint resumption after the network is restored. The encrypted transmission module uses AES-256-GCM encryption algorithm to encrypt the transmission of communication data.

[0041] Specifically, the implementation process of the dual-protocol communication mechanism includes the following steps: S1. Protocol switching logic implementation: Dynamically select the optimal communication protocol based on network status. Set the network latency threshold to 200ms. When the real-time monitored network latency is <200ms and the network status is stable, select the WebSocket protocol to achieve real-time communication; when the network latency is ≥200ms, automatically switch to the HTTP / HTTPS protocol to ensure the stability of data transmission.

[0042] Protocol switching logic: Java public class ProtocolSwitcher { private static final int DELAY_THRESHOLD = 200; / / ms public CommunicationProtocol selectProtocol(NetworkStatus status) { if (status.getLatency() <DELAY_THRESHOLD&&status.isStable()) { return new WebSocketProtocol(); } else { return new HttpProtocol(); } } } ``` S2. The offline caching mechanism is implemented by using a local SQLite database to store the operation data during offline periods. The cache capacity is limited to 100MB. When the network connection is restored, the cached data is automatically synchronized to the large model inference server to ensure that the offline operation data is not lost.

[0043] Offline caching mechanism: Java public class OfflineCache { private static final int MAX_CACHE_SIZE = 100 * 1024 * 1024; / / 100MB public void cacheData(UserOperation operation) { if (getCurrentCacheSize() + operation.getSize()<= MAX_CACHE_SIZE) { sqliteDB.insert(operation); } } public void syncWhenOnline() { List <useroperation>cachedOps = sqliteDB.getAllUnsynced(); for (UserOperation op : cachedOps) { serverAPI.upload(op); sqliteDB.markAsSynced(op.getId()); } } } ``` S3. Data encryption transmission is implemented by integrating the AES-256-GCM encryption algorithm into the MCP communication adapter unit. The symmetric encryption method is used to encrypt all data exchanged between the intelligent agent and the mobile terminal, ensuring the security of engineering supervision data from the data transmission level.

[0044] Encrypted data transmission: Java public class DataEncryption { private static final String ALGORITHM = "AES-256-GCM"; public EncryptedData encrypt(String plainText) { SecretKey key = getSecretKey(); Cipher cipher = Cipher.getInstance(ALGORITHM); cipher.init(Cipher.ENCRYPT_MODE, key); byte[] encryptedData = cipher.doFinal(plainText.getBytes()); return new EncryptedData(encryptedData, cipher.getIV()); } } ``` The S4 and MCP communication adapter units are integrated, combining core components such as protocol switching, offline caching, and encrypted data transmission into the MCP communication adapter unit to form a complete dual-protocol communication solution, specifically including: (1) Network status monitoring module: Real-time monitoring of key network indicators such as network latency, bandwidth, and packet loss rate; (2) Protocol selector: Automatically selects the optimal communication protocol based on preset delay threshold rules; (3) Data Cache Manager: Responsible for offline data storage, capacity management, and synchronization after network recovery; (4) Security encryption module: Performs AES-256-GCM encryption on all transmitted data to ensure data security.

[0045] S5. Performance optimization and testing: Multi-dimensional performance optimization of the MCP communication protocol to adapt to the complex network environment of engineering construction sites. (1) Connection pool management: Reuse established network connections to reduce the resource overhead of frequent connection establishment; (2) Data compression: Compress large-volume regulatory data messages to shorten data transmission time; (3) Heartbeat mechanism: Send heartbeat packets periodically to maintain an active communication connection and avoid abnormal connection interruption; (4) Error retry: Configure intelligent retry strategy to perform graded retry for data that fails to be transmitted, thereby improving communication reliability.

[0046] The dual-protocol communication mechanism effectively adapts to the complex network environment and unstable signal at engineering construction sites, ensuring the real-time, integrity, and security of data transmission between the intelligent agent and the mobile terminal application, and providing core communication technology support for the implementation of intelligent agents for engineering construction management.

[0047] The mobile terminal includes a construction sub-project selection module, a regulatory element display module, an early warning push module, an intelligent Q&A interaction module, a special data entry module, and a data visualization module. It is used to send construction supervision request information to the large model inference server, receive and display structured supervision lists, enter and transmit supervision progress data and special supervision data, receive tiered early warning instructions and display early warning information, support engineering supervision question interaction in natural language, and visualize supervision completion, early warning distribution, and compliance statistics in chart form.

[0048] Example 3 This embodiment provides a method for constructing an intelligent agent for engineering construction management based on a large model, including the following steps: S1. Large Model Selection and Fine-Tuning Training: The Tongyi Qianwen Large Language Model with 8 billion parameters based on the Transformer architecture is used as the basic model to construct a dedicated dataset for the field of engineering construction supervision and complete the model training. S2. Standardized Data Processing: Collect relevant industry standards for engineering construction, extract the content and scope of application of the standard clauses, and construct a structured knowledge graph; S3. Supervision Element Compilation: Compile supervision elements for various construction sub-projects such as construction diversion, earthwork excavation, foundation treatment, concrete engineering, metal structure installation, and electromechanical equipment installation to form a supervision element database; S4. Three-stage training of the model: Fine-tuning the parameters of the basic model using the labeled regulatory element dataset to enable the model to learn the relationship between regulatory elements and regulatory clauses; optimizing the model's semantic understanding, interaction logic and decision-making ability in regulatory scenarios based on human feedback; and regularly updating the model with new regulatory case data to maintain the timeliness and accuracy of the model's knowledge. S5 and MCP communication adapter unit setup: Build an MCP communication adapter unit to achieve dual-protocol communication, offline caching, and encrypted transmission functions. (1) Protocol switching: Switch WebSocket and HTTP / HTTPS protocols in real time according to network latency to ensure communication stability under complex network conditions; (2) Offline caching: Offline operation data is stored in a local SQLite database and automatically synchronized to the large model inference server after the network is restored; (3) Encrypted transmission: All interactive data is encrypted using the AES-256-GCM encryption algorithm to ensure data transmission security; S6. Mobile Terminal Development: Develop a mobile terminal as the interaction medium between users and intelligent agents, and develop the following core modules: (1) Construction sub-section selection module, which displays the construction sub-sections of the project in a three-level tree structure, allowing users to select specific supervision objects level by level; (2) Regulatory element display module, which displays the regulatory list in six dimensions: technical parameters, quality standards, safety specifications, process nodes, environmental protection standards, and flood control safety, and marks the standard clause number; (3) Early warning push module, which triggers graded early warnings based on the regulatory completion threshold and reminds users in the form of pop-up windows; (4) Intelligent question-and-answer interaction module, which allows users to raise engineering supervision questions through natural language, and the large model inference server returns the analysis results with the standard basis; (5) Special data entry module, which receives special regulatory data such as flood control measures and construction wastewater treatment entered by users, and synchronizes them to the large model inference server for compliance verification; (6) Data visualization module, which displays data such as regulatory completion rate, early warning distribution and compliance statistics in the form of charts.

[0049] Example 4 like Figure 6 As shown, this embodiment provides a workflow for an engineering construction management and control intelligent agent based on a large model. Taking "concrete engineering → large-volume concrete pouring" as an example, it includes the following steps: S1. Users log in through a mobile terminal application and select the "Concrete Engineering → Large Volume Concrete Pouring" construction section. The mobile terminal application then transmits the monitoring requirements to the intelligent agent. S2, the core module of the intelligent agent's large model reasoning analyzes the requirements, identifies the sub-section as a key construction sub-section, and configures the corresponding early warning threshold; S3. The intelligent agent uses the intelligent matching module of regulatory elements to select regulatory elements such as concrete mix design, pouring temperature control and curing measures from the regulatory element database. S4. The intelligent agent feeds back the generated structured regulatory list to the mobile terminal application through the MCP communication adapter unit, so that users can view the regulatory requirements. S5. After completing the on-site verification, the user marks the completed and incomplete items through a mobile terminal application, and the relevant data is synchronized to the intelligent agent. S6. The hierarchical early warning decision module of the intelligent agent calculates the regulatory matching degree. When the matching degree is lower than the preset threshold, a first-level early warning instruction is generated. S7. The intelligent agent pushes the early warning information to the mobile terminal application through the MCP communication adapter unit, and simultaneously notifies the user and the project owner via SMS. S8. After the user completes the supplementary uninspected items, the regulatory record is updated and synchronized to the intelligent agent. Once the matching degree meets the standard, the warning is lifted, and all regulatory data is stored in the data autonomous storage module.

[0050] Example 5 This embodiment provides a computer device, which can be a mobile phone, laptop, tablet, desktop computer or server, etc., for executing the engineering construction supervision element management method described in the above embodiment.

[0051] The computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor; When the processor executes the computer program, it implements the steps of the engineering construction management and control method based on a large model as described in Embodiment 1.

[0052] The memory may be a high-speed random access memory, or it may include non-volatile memory, such as at least one disk storage device. The processor can be a general-purpose processor such as a central processing unit (CPU) or a digital signal processor (DSP), but this embodiment does not impose any specific limitations.

[0053] This embodiment also provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it implements the steps of the engineering construction management and control method based on a large model as described in Embodiment 1.

[0054] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to interchangeably. Each embodiment focuses on describing the differences from other embodiments. In particular, the method and system embodiments are basically similar to the method embodiments, so the descriptions are relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0055] The foregoing has described specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than that shown in the embodiments and may still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0056] The above description is merely one or more embodiments of this specification and is not intended to limit this specification. Various modifications and variations can be made to the one or more embodiments of this specification by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of one or more embodiments of this specification should be included within the scope of the claims of this specification.< / useroperation>

Claims

1. A method for engineering construction management and control based on a large model, characterized in that, The method includes the following steps: S1. Receive construction supervision request information, wherein the construction supervision request information includes at least one or more of the following: construction sub-project information and construction supervision text information; S2. The construction supervision request information is semantically parsed by the large language model deployed in the large model inference server to identify the corresponding construction sub-type and supervision scenario, and to determine the construction sub-type. The construction sub-type includes at least a key construction sub-type or a normal construction sub-type. S3. Based on the identified construction sub-types, retrieve the corresponding regulatory element data and standard clause numbers from the regulatory element database; S4. Based on the large language model, perform association reasoning on the retrieved regulatory element data, match the association relationship between the regulatory element data and the normative clause number, and generate a structured regulatory list; S5. Send the structured regulatory list to the mobile terminal through the MCP communication adapter unit, and receive the regulatory progress data returned by the mobile terminal; S6. Based on the regulatory progress data, count the number of completed regulatory items and calculate the matching degree between the number of completed regulatory items and the total number of preset regulatory elements; S7. Compare the matching degree with the preset warning threshold of the corresponding construction section. When the matching degree is lower than the preset warning threshold, generate a graded warning instruction and send the warning information to the mobile terminal.

2. The engineering construction management and control method based on a large model according to claim 1, characterized in that, The large language model mentioned in step S2 is a large language model based on the Transformer architecture. The large language model is fine-tuned and trained using a regulatory element dataset. The regulatory element dataset includes at least one or more of the following: construction diversion data, earthwork excavation data, foundation treatment data, concrete engineering data, metal structure installation data, and electromechanical equipment installation data.

3. The engineering construction management and control method based on a large model according to claim 1, characterized in that, Step S2 specifically includes the following steps: S21. Extract construction sub-keywords from the construction supervision request information; S22. Match the construction sub-section keywords with the sub-section information in the regulatory element database; S23. Determine the corresponding construction sub-type based on the matching results.

4. The engineering construction management and control method based on a large model according to claim 1, characterized in that, The regulatory element database mentioned in step S3 includes at least one or more of the following: a section information table, a regulatory element table, a regulatory clause table, and a historical case table. The division information table includes at least one or more of the following: division ID, division name, parent division ID, and division level. The regulatory element table shall include at least one or more of the following: element ID, element name, element category, judgment criteria, inspection method, and regulatory clause number; The specification clause table shall include at least one or more of the following: clause ID, specification name, clause number, clause content, and scope of application; The historical case table shall include at least one or more of the following: case ID, problem description, solution, involved divisions, and involved elements.

5. The engineering construction management and control method based on a large model according to claim 1, characterized in that, The structured regulatory list mentioned in step S4 is classified according to the construction supervision dimensions. The construction supervision dimensions include at least one or more of the following: technical parameters, quality standards, safety specifications, process nodes, environmental protection standards, and flood control safety.

6. The engineering construction management and control method based on a large model according to claim 1, characterized in that, The formula for calculating the matching degree in step S6 is: Matching degree = (Number of completed regulatory items / Total number of preset regulatory elements) × 100%; The graded early warning instruction in step S7 sets differentiated preset early warning thresholds according to the type of construction sub-section. When the matching degree of critical construction sub-section is <98%, a level one early warning is generated, and when the matching degree of ordinary construction sub-section is <95%, a level two early warning is generated.

7. An intelligent agent for engineering construction management and control based on a large model, characterized in that, The intelligent agent is used to implement the engineering construction management and control method based on a large model as described in any one of claims 1-6. The large-scale model inference server includes a large-scale model inference core module, a regulatory element intelligent matching module, a hierarchical early warning decision module, and a data autonomous storage module. It is used to deploy a large language model, perform semantic parsing on construction supervision request information and identify construction sub-types and supervision scenarios, retrieve and match corresponding regulatory element data and standard clause numbers from the regulatory element database, perform correlation inference on regulatory element data based on the large language model and generate a structured supervision list, count the number of completed supervision items and calculate the matching degree, generate hierarchical early warning instructions based on the comparison result of the matching degree and the preset early warning threshold, and persistently store various types of data throughout the supervision process. The MCP communication adapter unit includes a protocol switching module, an offline caching module, and an encrypted transmission module. It is used to realize data interaction and command transmission between the large model inference server and the mobile terminal. It adaptively switches the communication protocol according to the network status, performs local caching of transmitted data in the offline state and resumes interrupted transmission after network recovery, and encrypts and protects all interactive data to ensure the continuity, security and integrity of data transmission. The mobile terminal includes a construction sub-project selection module, a regulatory element display module, an early warning push module, an intelligent Q&A interaction module, a special data entry module, and a data visualization module. It is used to send construction supervision request information to the large model inference server, receive and display structured supervision lists, enter and transmit supervision progress data and special supervision data, receive tiered early warning instructions and display early warning information, support engineering supervision question interaction in natural language, and visualize supervision completion, early warning distribution, and compliance statistics in chart form.

8. The engineering construction management and control intelligent agent based on a large model according to claim 7, characterized in that, The data self-storage module uses MongoDB database as the storage medium; The MCP communication adaptation unit adopts a dual-protocol communication mechanism based on RESTful API and WebSocket. The protocol switching module automatically switches between HTTP / HTTPS protocol and WebSocket protocol according to the network latency threshold. The offline caching module uses SQLite local database to store offline transmission data and performs data synchronization or breakpoint resumption after the network is restored. The encrypted transmission module uses AES-256-GCM encryption algorithm to encrypt the transmission of communication data.

9. A computer device, characterized in that, include: At least one processor; as well as, A memory communicatively connected to the at least one processor; wherein, The memory has instructions executable by the at least one processor, which, when executed by the at least one processor, enables the at least one processor to perform the large-model-based engineering construction management method as described in any one of claims 1-6.

10. A non-volatile computer storage medium, characterized in that, The system stores computer-executable instructions configured to implement the large-model-based engineering construction management method as described in any one of claims 1-6.

Images