A method and system for realizing autonomous operation and maintenance of an MES system through an AI large model

Abstract

The application relates to the field of industrial automation and discloses a method and system for realizing autonomous operation and maintenance of an MES system through an AI large model, which comprises the following steps: receiving an operation and maintenance interaction instruction, sending the operation and maintenance interaction instruction to a large language model for analysis, receiving a processing suggestion returned by the large language model, and receiving a confirmation operation for the processing suggestion to trigger a first business processing task; creating an asynchronous thread through a timing task scheduling mechanism, and periodically triggering a second business processing task according to a preset execution interval in the asynchronous thread; and realizing explicitization and inheritance of implicit knowledge through an AI knowledge retrieval function. New employees can obtain operation experience and processing suggestions of experienced employees through an AI assistant.

Description

Technical Field

[0001] This invention relates to the field of industrial automation, and in particular to a method and system for achieving autonomous operation and maintenance of an MES system through a large AI model. Background Technology

[0002] As the manufacturing industry moves towards intelligence and automation, the Manufacturing Execution System (MES) has become the core information platform for modern manufacturing. With the large-scale application of MES systems, enterprises have incurred a significant amount of maintenance work to ensure their stable operation. For example, some enterprises implement 24 / 7 production, requiring personnel to be on duty at all times to guarantee the MES system's continuous operation, leading to problems such as high personnel input and low maintenance efficiency.

[0003] (1) Low operation and maintenance efficiency Traditional MES system maintenance relies heavily on manual labor, with a large amount of repetitive work consuming significant human resources. For example, tasks such as physical property data entry, inspection task processing, and report generation all require manual operation, resulting in low efficiency. According to statistics from a large manufacturing company, quality control personnel spend approximately 60% of their time each day on repetitive data entry tasks.

[0004] (2) Slow response speed Anomaly handling relies on manual discovery and processing, resulting in long response times. Traditional inspection methods typically occur every 2-3 hours, making it impossible to detect anomalies in the production process in real time. This leads to untimely problem handling and may result in a batch of defective products.

[0005] (3) Decision-making relies on experience The lack of intelligent analysis and decision support means that decisions rely primarily on the experience of operators. The significant differences in experience levels among operators can lead to subjectivity and inconsistency in decision-making.

[0006] (4) Difficulty in knowledge transmission Tacit knowledge accumulated during the production process is difficult to pass on effectively. When experienced employees retire or leave the company, their accumulated experience and knowledge are often difficult to fully transfer to new employees, leading to the loss of the company's knowledge assets.

[0007] A complete AI-driven autonomous operation and maintenance solution needs to be designed. Summary of the Invention

[0008] To address the aforementioned technical problems, this invention provides a method for achieving autonomous operation and maintenance of an MES system through a large AI model, comprising the following steps: Receive operation and maintenance interaction instructions, send the operation and maintenance interaction instructions to the large language model for parsing, receive the processing suggestions returned by the large language model, and receive the confirmation operation for the processing suggestions to trigger the first business processing task. An asynchronous thread is created through a timed task scheduling mechanism, and a second business processing task is periodically triggered in the asynchronous thread according to a preset execution interval. Wherein, the first business processing task and / or the second business processing task is an automatic copying task of physical property data, and the steps for executing the automatic copying task of physical property data include: Query the inspection task results table and identify the master volume physical property inspection task whose task status is "passed". Based on the parent volume serial number of the parent volume physical property inspection task, query all sub-volume inspection tasks belonging to the same batch. Extract the physical property data of the master volume physical property inspection task, and copy the physical property data to each of the sub-volume inspection tasks according to the three-layer structure of inspection record header, inspection item and inspection record tail. Extract the date, time, personnel, specifications, roll number, machine number, order number, and customer information corresponding to each roll, and populate them into the corresponding roll inspection task; The application programming interface of the manufacturing execution system is called to save the inspection data of each of the small roll inspection tasks, and the inspection pass operation is performed on each of the small roll inspection tasks.

[0009] Secondly, this invention also relates to a system for autonomous operation and maintenance of an MES system through a large AI model, comprising: The AI ​​intelligent interaction module is used to receive operation and maintenance interaction instructions, send the operation and maintenance interaction instructions to the large language model for parsing, receive the processing suggestions returned by the large language model, and receive the confirmation operation for the processing suggestions to trigger the first business processing task. The scheduled task module is used to create an asynchronous thread through a scheduled task scheduling mechanism, and periodically trigger a second business processing task in the asynchronous thread according to a preset execution interval. The automatic physical property data copying module is used to query the inspection task result table and identify the parent volume physical property inspection task whose task status is passed when the first business processing task and / or the second business processing task is an automatic physical property data copying task; according to the parent volume serial number of the parent volume physical property inspection task, it queries all small volume inspection tasks belonging to the same batch; it extracts the physical property data of the parent volume physical property inspection task and copies the physical property data to each of the small volume inspection tasks according to the three-layer structure of inspection record header, inspection item and inspection record tail; it extracts the date, time, personnel, specifications, volume number, machine number, order number and customer information corresponding to each small volume and fills them into the corresponding small volume inspection task. The MES system integration module is used to call the application programming interface of the manufacturing execution system to save the inspection data of each small roll inspection task, and to perform inspection pass operation on each small roll inspection task.

[0010] Thirdly, the present invention also relates to a computer-readable storage medium having a computer program stored thereon that, when executed by a processor, implements the steps of the aforementioned method.

[0011] Fourthly, the present invention also relates to a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps of the aforementioned method.

[0012] Compared with the prior art, the beneficial effects of the present invention are as follows: By automating operation and maintenance, the workload of quality inspectors can be reduced, saving labor costs by ≥30%. Taking a medium-sized copper foil enterprise as an example, the quality inspection department originally had 10 people, which can be reduced to 7 people after implementing this invention, saving about 500,000 yuan in labor costs annually.

[0013] It is suitable for multiple processes such as copper melting, electrolysis, and slitting, and supports the operation and maintenance of copper foil with different thicknesses from 4 to 10 μm. The system adopts a plug-in design, which can be configured and expanded according to the needs of different processes.

[0014] AI-powered knowledge retrieval enables the making of tacit knowledge explicit and facilitates its transfer. New employees can access operational experience and problem-solving suggestions from experienced employees through the AI ​​assistant. Attached Figure Description

[0015] Figure 1 The overall flowchart for the autonomous operation and maintenance of the MES system based on the AI ​​large model provided in the embodiments of the present invention is shown.

[0016] Figure 2 Detailed flowchart of the task of automatically copying physical property data; Figure 3 A schematic diagram of the modules of the system that enables autonomous operation and maintenance of the MES system through AI big data models. Detailed Implementation

[0017] like Figure 1 , Figure 2 A method for achieving autonomous operation and maintenance of an MES system through large AI models includes the following steps: Receive operation and maintenance interaction instructions, send the operation and maintenance interaction instructions to the large language model for parsing, receive the processing suggestions returned by the large language model, and receive the confirmation operation for the processing suggestions to trigger the first business processing task. An asynchronous thread is created through a timed task scheduling mechanism, and a second business processing task is periodically triggered in the asynchronous thread according to a preset execution interval. Wherein, the first business processing task and / or the second business processing task is an automatic copying task of physical property data, and the steps for executing the automatic copying task of physical property data include: Query the inspection task results table and identify the master volume physical property inspection task whose task status is "passed". Based on the parent volume serial number of the parent volume physical property inspection task, query all sub-volume inspection tasks belonging to the same batch. Extract the physical property data of the master volume physical property inspection task, and copy the physical property data to each of the sub-volume inspection tasks according to the three-layer structure of inspection record header, inspection item and inspection record tail. Extract the date, time, personnel, specifications, roll number, machine number, order number, and customer information corresponding to each roll, and populate them into the corresponding roll inspection task; The application programming interface of the manufacturing execution system is called to save the inspection data of each of the small roll inspection tasks, and the inspection pass operation is performed on each of the small roll inspection tasks.

[0018] The step of creating an asynchronous thread through a timed task scheduling mechanism, and periodically triggering a second business processing task in the asynchronous thread according to a preset execution interval, includes: The Task.Run method is called to create the asynchronous thread, and an infinite while loop structure is built within the asynchronous thread; Read the preset IsRunCopy flag and control the start and stop of the automatic copying task of the physical property data according to the IsRunCopy flag; When executing tasks in the infinite loop structure, the same database connection is reused, and a delayed waiting operation is performed after a single round of task traversal and processing is completed. When a database operation exception or network exception is detected, the exception information is written to the log file and the current exception task is skipped to execute the next round of the loop.

[0019] The step of sending the operation and maintenance interaction command to the large language model for parsing includes: The dialogue context history is stored locally, and the dialogue context history is combined with the operation and maintenance interaction instructions into a JSON format message. Configure the model name parameter, temperature parameter, and streaming response switch parameter of the large language model in the JSON format message; The RESTful API communication method is used to send the JSON-formatted message to the large language model via an HTTP POST request.

[0020] The application programming interface that calls the manufacturing execution system saves the inspection data for each of the small-volume inspection tasks, including: Send an authentication request to the manufacturing execution system to obtain a token, store the obtained token in a local cache, and send a refresh request before the token expires in 5 minutes; Using UTF-8 encoding, the request data is encapsulated in JSON or Multipart / form-data format in the HTTP client, the token is attached, and the application interface is called via HTTPS protocol; When a network transmission anomaly is detected by calling the application interface, a retry mechanism is triggered, and the retry mechanism is stopped when the number of retries reaches 3.

[0021] Before the step of receiving operation and maintenance interaction instructions, the method also includes a step of initializing the data cache: A singleton pattern for the data cache class is created using a double-checked locking mechanism; In the data cache class, a thread-safe collection type is instantiated, and AI role configuration information and operation permission configuration information are loaded into the collection type for centralized storage through an asynchronous lazy loading mechanism; Based on a preset time period and event triggering mechanism, a synchronous refresh operation is performed on the stored data in the data cache class in the background.

[0022] The method also includes the following security protection steps: Obtain the user code that issued the operation and maintenance interaction command, query the operation permissions corresponding to the user code in the system configuration, and filter out unauthorized operation commands based on the operation permissions; Using preset desensitization rules, the database connection string, print address, and operation address in the input content of the operation and maintenance interaction instructions and the system output content are replaced with mask characters; The input parameters of the operation and maintenance interaction commands are validated, and request parameters containing SQL injection statements and exceeding the set length limit are intercepted.

[0023] like Figure 3 The embodiments of the present invention also include a system for autonomous operation and maintenance of the MES system through a large AI model, comprising: The AI ​​intelligent interaction module is used to receive operation and maintenance interaction instructions, send the operation and maintenance interaction instructions to the large language model for parsing, receive the processing suggestions returned by the large language model, and receive the confirmation operation for the processing suggestions to trigger the first business processing task. The scheduled task module is used to create an asynchronous thread through a scheduled task scheduling mechanism, and periodically trigger a second business processing task in the asynchronous thread according to a preset execution interval. The automatic physical property data copying module is used to query the inspection task result table and identify the parent volume physical property inspection task whose task status is passed when the first business processing task and / or the second business processing task is an automatic physical property data copying task; according to the parent volume serial number of the parent volume physical property inspection task, it queries all small volume inspection tasks belonging to the same batch; it extracts the physical property data of the parent volume physical property inspection task and copies the physical property data to each of the small volume inspection tasks according to the three-layer structure of inspection record header, inspection item and inspection record tail; it extracts the date, time, personnel, specifications, volume number, machine number, order number and customer information corresponding to each small volume and fills them into the corresponding small volume inspection task. The MES system integration module is used to call the application programming interface of the manufacturing execution system to save the inspection data of each small roll inspection task, and to perform inspection pass operation on each small roll inspection task.

[0024] The various modules of this invention are described as follows: (1) AI intelligent interaction module This function is used to receive operation and maintenance interaction commands, send the operation and maintenance interaction commands to a large language model for parsing, receive processing suggestions returned by the large language model, and receive confirmation operations for the processing suggestions to trigger the first business processing task. 1. Large Language Model Integration: It uses a RESTful API to connect with the DeepSeek large language model, supports HTTP POST requests, and uses JSON format for messages, including configuration items such as model name, temperature parameters, and streaming response switch.

[0025] 2. Multi-turn dialogue support: Maintains dialogue context history, supports multi-turn dialogue interaction, and automatically manages the message list.

[0026] 3. Diverse AI functions: Supports AI chat, AI knowledge retrieval, AI data analysis, AI confirmation and execution, and other functions.

[0027] 4. Role-based access control: Supports configuration of multiple AI roles and filters executable operations based on user permissions.

[0028] 5. Security protection mechanism: Automatically blocks sensitive information to prevent information leakage and malicious attacks.

[0029] 6. Standardized Interface: Provides a unified command interface, supports multiple command types, and facilitates system integration and expansion.

[0030] This invention enables intelligent interaction, reducing problem-solving time from an average of 30 minutes to 5 minutes and increasing the problem-solving rate from 70% to 90%.

[0031] (2) Self-animal data replication module This mechanism is used to query the inspection task result table and identify the parent volume physical property inspection task whose task status is "passed" when the first business processing task and / or the second business processing task are physical property data automatic copying tasks; based on the parent volume serial number of the parent volume physical property inspection task, it queries all small volume inspection tasks belonging to the same batch; it extracts the physical property data of the parent volume physical property inspection task and copies the physical property data to each of the small volume inspection tasks according to the three-layer structure of inspection record header, inspection item and inspection record tail; it extracts the date, time, personnel, specifications, volume number, machine number, order number and customer information corresponding to each small volume and fills them into the corresponding small volume inspection task: 1. Inspection task identification mechanism: By querying the inspection task result table, it identifies the passed physical property inspection tasks, supporting filtering by time range and multi-condition filtering.

[0032] 2. Batch Association Query: Query all sub-volume inspection tasks in the same batch based on the parent volume number.

[0033] 3. Intelligent data population: Automatically populates date, time, personnel, specifications, roll number, machine number, order number, customer information, etc.

[0034] 4. Multi-level data replication: Supports replication of the header, inspection items, and tail of inspection records.

[0035] 5. Automated inspection process: Automatically performs data saving and inspection pass operations, and records detailed logs.

[0036] 6. Exception handling mechanism: Supports exception capture and logging. When replication fails, an exception is automatically thrown and error information is recorded.

[0037] This invention reduces the single-volume processing time from 5 minutes to 10 seconds, improves data accuracy from 95% to 99.9%, and saves the workload of 2 people per shift.

[0038] (3) MES system integration module The application programming interface (API) of the manufacturing execution system is used to save the inspection data of each of the small roll inspection tasks and to perform inspection pass operations on each of the small roll inspection tasks. 1. Token authentication mechanism: Automatically acquire, cache, and refresh tokens, valid for 5 minutes.

[0039] 2. HTTP client configuration: 5-minute timeout, supports JSON and Multipart / form-data formats, UTF-8 encoding.

[0040] 3. Data transmission security: HTTPS support, exception handling, and automatic retry mechanism.

[0041] 4. Interface encapsulation: The API of the MES system is encapsulated into a unified method for easy calling and maintenance.

[0042] This invention achieves seamless integration with the MES system, with a data transmission success rate of ≥99.9% and an average response time of ≤500ms.

[0043] (4) Scheduled Task Module This is used to create an asynchronous thread through a timed task scheduling mechanism, and to periodically trigger a second business processing task in the asynchronous thread according to a preset execution interval: 1. Asynchronous thread execution: Use Task.Run() to create an independent asynchronous thread to execute the scheduled task without blocking the main thread.

[0044] 2. Configurable execution interval: The default execution interval is 4 minutes, which can be adjusted through configuration parameters.

[0045] 3. Infinite loop mechanism: It adopts a while(1==1) infinite loop structure to continuously execute the task.

[0046] 4. Task Status Management: Controls task start and stop via the IsRunCopy flag, and records execution logs and result statistics.

[0047] 5. Database connection management: Reuse database connections to reduce the overhead of establishing and closing connections.

[0048] 6. Task execution flow: Query pending tasks → Iterate through and process them → Delay and wait → Next round.

[0049] 7. Exception Handling: Capture and handle database operation exceptions, network exceptions, etc., log errors, and continue to execute other tasks.

[0050] This invention enables automated operation and maintenance, automatically executing a physical property data copying task (second business processing task) every 4 minutes, reducing manual intervention by ≥90%.

[0051] (5) Data caching module A singleton pattern for the data cache class is created using a double-checked locking mechanism; In the data cache class, a thread-safe collection type is instantiated, and AI role configuration information and operation permission configuration information are loaded into the collection type for centralized storage through an asynchronous lazy loading mechanism. Based on a preset time period and event triggering mechanism, a synchronous refresh operation is performed on the stored data in the data cache class in the background.

[0052] 1. Singleton pattern design: The singleton pattern is implemented with double-checked locking to ensure that the global instance is unique.

[0053] 2. Centralized data management: Caches data such as AI roles, AI operations, printers, and configuration parameters.

[0054] 3. Data access interface: Provides operations such as data acquisition, updating, and clearing.

[0055] 4. Asynchronous loading mechanism: Delayed loading, background updates, without affecting system startup speed.

[0056] 5. Data synchronization strategies: timed refresh and event-triggered refresh.

[0057] 6. Thread safety guarantee: Data is stored using a thread-safe collection type, supporting concurrent access.

[0058] This invention reduces the number of database queries by ≥80% and the data access response time by ≤1ms.

[0059] (6) Safety protection module Obtain the user code that issued the operation and maintenance interaction command, query the operation permissions corresponding to the user code in the system configuration, and filter out unauthorized operation commands based on the operation permissions; Using preset desensitization rules, the database connection string, print address, and operation address in the input content of the operation and maintenance interaction instructions and the system output content are replaced with mask characters; The input parameters of the operation and maintenance interaction commands are validated, and request parameters containing SQL injection statements and exceeding the set length limit are intercepted. 1. Sensitive Information Masking: Automatically masks sensitive information such as database connection strings, print addresses, and operation addresses.

[0060] 2. User permission verification: Filters executable operations based on user code, and supports configuring operation permissions for different users.

[0061] 3. Input validation: Perform validity checks on input parameters to prevent SQL injection attacks and buffer overflows.

[0062] 4. Log recording: Record operation logs, exception logs, and access logs to facilitate auditing and traceability.

[0063] 5. Interface security: Verify the legality and integrity of command IDs and input parameters.

[0064] 6. Data anonymization: Anonymize sensitive data, such as phone numbers and ID card numbers.

[0065] This invention effectively prevents information leakage and illegal operations, improving system security by ≥90%.

[0066] The following are specific examples: Example 1: Automatic Copying of Physical Property Data Scenario: In the production process of lithium battery copper foil, the small rolls cut from the same master roll have the same physical properties, and it is necessary to automatically copy the physical property data of the master roll to each small roll.

[0067] Implementation steps: 1. Triggering condition: The system automatically triggers the copying task after the physical property verification task of the parent volume passes. 2. Query Sub-rolls: Query all sub-roll inspection tasks in the same batch based on the parent roll number; 3. Data replication: The physical property data of the master volume is copied to each sub-volume, and information such as volume number, date, and personnel is intelligently filled in; 4. Data Storage: Save the inspection data by calling the MES API; 5. Automatic Pass: Automatically performs the inspection and pass operation for small rolls; 6. Log recording: Record operation logs, including task ID, volume number, operation results, etc.

[0068] Implementation results: - Single-volume processing time: reduced from 5 minutes to 10 seconds; - Data accuracy: improved from 95% to 99.9%; - Labor saving: Saves the workload of 2 people per shift.

[0069] Example 2: AI-powered intelligent question answering Scenario: Operators need to query production anomaly handling solutions and obtain professional advice through an AI assistant.

[0070] Implementation steps: 1. Submitting a problem: The operator enters the problem into the system (e.g., "How to handle pinhole defects in copper foil?"). 2. AI Analysis: The system sends the question to the DeepSeek large language model for analysis; 3. Returning the answer: The AI ​​model returns detailed processing suggestions and steps; 4. Execution: The operator shall take appropriate action according to the recommendations; 5. Feedback Log: Record the problem handling process and results for subsequent AI model optimization.

[0071] Implementation results: - Problem resolution time: reduced from an average of 30 minutes to 5 minutes; - Problem resolution rate: increased from 70% to 90%; - Knowledge sharing: Making tacit knowledge explicit.

[0072] Example 3: Automatic Execution of Scheduled Tasks Scenario: The system needs to periodically and automatically perform physical property data copying tasks to ensure timely updates of production data.

[0073] Implementation steps: 1. Task Configuration: Configure the scheduled task execution interval to 4 minutes; 2. Task Startup: The scheduled task thread starts automatically when the system starts; 3. Task Execution: Automatically query the parent volume data to be processed every 4 minutes and perform physical property data copying; 4. Results Recording: Record the results of each task execution, including the number of parent volumes processed, the number of successful transactions, and the number of failed transactions; 5. Exception handling: Failed tasks are logged and manual retry is supported.

[0074] Implementation results: - Automation level: 100% automatic execution, no manual intervention required; - Execution frequency: Once every 4 minutes; - Success rate: ≥99.9%.

[0075] Example 4: Multi-system data integration Scenario: It is necessary to synchronize the inspection data of the LIMS system to the MES system to achieve unified data management.

[0076] Implementation steps: 1. Data Acquisition: Acquire test data from the LIMS system; 2. Data Conversion: Converting LIMS data format to the format required by the MES system; 3. Data Upload: Upload inspection data by calling the MES system API; 4. Data Validation: Verify the correctness of the uploaded data; 5. Status synchronization: Synchronize and verify the status information of the task.

[0077] Implementation results: - Data synchronization time: ≤1 minute; - Data accuracy: ≥99.9%; - Human intervention: 0.

[0078] The results of the implementation are shown in Table 1.

[0079] Table 1 The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for achieving autonomous operation and maintenance of an MES system through a large AI model, characterized in that, Includes the following steps: Receive operation and maintenance interaction instructions, send the operation and maintenance interaction instructions to the large language model for parsing, receive the processing suggestions returned by the large language model, and receive the confirmation operation for the processing suggestions to trigger the first business processing task. An asynchronous thread is created through a timed task scheduling mechanism, and a second business processing task is periodically triggered in the asynchronous thread according to a preset execution interval. Wherein, the first business processing task and / or the second business processing task is an automatic copying task of physical property data, and the steps for executing the automatic copying task of physical property data include: Query the inspection task results table and identify the master volume physical property inspection task whose task status is "passed". Based on the parent volume serial number of the parent volume physical property inspection task, query all sub-volume inspection tasks belonging to the same batch. Extract the physical property data of the master volume physical property inspection task, and copy the physical property data to each of the sub-volume inspection tasks according to the three-layer structure of inspection record header, inspection item and inspection record tail. Extract the date, time, personnel, specifications, roll number, machine number, order number, and customer information corresponding to each roll, and populate them into the corresponding roll inspection task; The application programming interface of the manufacturing execution system is called to save the inspection data of each of the small roll inspection tasks, and the inspection pass operation is performed on each of the small roll inspection tasks.

2. The method according to claim 1, characterized in that, The step of creating an asynchronous thread through a timed task scheduling mechanism, and periodically triggering a second business processing task in the asynchronous thread according to a preset execution interval, includes: The Task.Run method is called to create the asynchronous thread, and an infinite while loop structure is built within the asynchronous thread; Read the preset IsRunCopy flag and control the start and stop of the automatic copying task of the physical property data according to the IsRunCopy flag; When executing tasks in the infinite loop structure, the same database connection is reused, and a delayed waiting operation is performed after a single round of task traversal and processing is completed. When a database operation exception or network exception is detected, the exception information is written to the log file and the current exception task is skipped to execute the next round of the loop.

3. The method according to claim 1, characterized in that, The step of sending the operation and maintenance interaction command to the large language model for parsing includes: The dialogue context history is stored locally, and the dialogue context history is combined with the operation and maintenance interaction instructions into a JSON format message. Configure the model name parameter, temperature parameter, and streaming response switch parameter of the large language model in the JSON format message; The RESTful API communication method is used to send the JSON-formatted message to the large language model via an HTTP POST request.

4. The method according to claim 1, characterized in that, The application programming interface that calls the manufacturing execution system saves the inspection data for each of the small-volume inspection tasks, including: Send an authentication request to the manufacturing execution system to obtain a token, store the obtained token in a local cache, and send a refresh request before the token expires in 5 minutes; Using UTF-8 encoding, the request data is encapsulated in JSON or Multipart / form-data format in the HTTP client, the token is attached, and the application interface is called via HTTPS protocol; When a network transmission anomaly is detected by calling the application interface, a retry mechanism is triggered, and the retry mechanism is stopped when the number of retries reaches 3.

5. The method according to claim 1, characterized in that, Before the step of receiving operation and maintenance interaction instructions, the method also includes a step of initializing the data cache: A singleton pattern for the data cache class is created using a double-checked locking mechanism; In the data cache class, a thread-safe collection type is instantiated, and AI role configuration information and operation permission configuration information are loaded into the collection type for centralized storage through an asynchronous lazy loading mechanism. Based on a preset time period and event triggering mechanism, a synchronous refresh operation is performed on the stored data in the data cache class in the background.

6. The method according to claim 1, characterized in that, The method also includes the following security protection steps: Obtain the user code that issued the operation and maintenance interaction command, query the operation permissions corresponding to the user code in the system configuration, and filter out unauthorized operation commands based on the operation permissions; Using preset desensitization rules, the database connection string, print address, and operation address in the input content of the operation and maintenance interaction instructions and the system output content are replaced with mask characters; The input parameters of the operation and maintenance interaction commands are validated, and request parameters containing SQL injection statements and exceeding the set length limit are intercepted.

7. A system for autonomous operation and maintenance of an MES system through a large AI model, characterized in that, include: The AI ​​intelligent interaction module is used to receive operation and maintenance interaction instructions, send the operation and maintenance interaction instructions to the large language model for parsing, receive the processing suggestions returned by the large language model, and receive the confirmation operation for the processing suggestions to trigger the first business processing task. The scheduled task module is used to create an asynchronous thread through a scheduled task scheduling mechanism, and periodically trigger a second business processing task in the asynchronous thread according to a preset execution interval. The automatic physical property data copying module is used to query the inspection task result table and identify the parent volume physical property inspection task whose task status is passed when the first business processing task and / or the second business processing task is an automatic physical property data copying task; according to the parent volume serial number of the parent volume physical property inspection task, it queries all sub-volume inspection tasks belonging to the same batch; extracts the physical property data of the parent volume physical property inspection task, and copies the physical property data to each sub-volume inspection task according to the three-layer structure of inspection record header, inspection item and inspection record tail. Extract the date, time, personnel, specifications, roll number, machine number, order number, and customer information corresponding to each roll, and populate them into the corresponding roll inspection task; The MES system integration module is used to call the application programming interface of the manufacturing execution system to save the inspection data of each small roll inspection task, and to perform inspection pass operation on each small roll inspection task.

8. The system according to claim 7, characterized in that, The system also includes a security protection module, configured as follows: Obtain the user code that issued the operation and maintenance interaction command, query the operation permissions corresponding to the user code in the system configuration, and filter out unauthorized operation commands based on the operation permissions; Using preset desensitization rules, the database connection string, print address, and operation address in the input content of the operation and maintenance interaction instructions and the system output content are replaced with mask characters; The input parameters of the operation and maintenance interaction commands are validated, and request parameters containing SQL injection statements and exceeding the set length limit are intercepted.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the program implements the steps of the method described in any one of claims 1-6.

10. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the steps of the method according to any one of claims 1-6.

Images