Operation and maintenance method, device and computer program product based on large language model

Abstract

The present disclosure provides an operation and maintenance method, device, electronic device, storage medium and computer program product based on a large language model, which relates to the field of computer technology, specifically to artificial intelligence large models, natural language understanding, and operation and maintenance technology, and can be applied to operation and maintenance scenarios. The specific implementation scheme is: the following operations are performed iteratively through an intelligent agent: determining the link operation and maintenance results corresponding to the completed links in the operation and maintenance task; through a large language model, according to the link operation and maintenance results corresponding to the completed links, determining the target tool for continuing to perform the operation and maintenance task from the tool set; calling the target tool to generate the link operation and maintenance results corresponding to the next link in the operation and maintenance task. The present disclosure adopts the Agent Loop method to determine the target tool for the next link in the operation and maintenance task based on the link operation and maintenance results corresponding to the completed links in the operation and maintenance task, thereby improving the applicability of the operation and maintenance solution in various operation and maintenance scenarios and the accuracy of the operation and maintenance results.

Description

Technical Field

[0001] The present disclosure relates to the field of computer technology, specifically to the field of artificial intelligence big models, natural language understanding, and operation and maintenance technology, and in particular to an operation and maintenance method, device, electronic device, storage medium, and computer program product based on a big language model, which can be applied in operation and maintenance scenarios. Background Art

[0002] Network operations and maintenance primarily refers to the daily maintenance, management, monitoring, and optimization of the network to ensure stable, efficient, and secure operation and provide reliable support for various network applications and services. This encompasses the configuration, debugging, and management of network equipment (such as routers, switches, and servers), network performance monitoring and analysis, troubleshooting and resolution, and the development and implementation of network security policies. Current network operations and maintenance solutions rely on predefined rule bases for alarm correlation and handling simple operations and maintenance events. Summary of the Invention

[0003] The present disclosure provides an operation and maintenance method, apparatus, electronic device, storage medium, and computer program product based on a large language model.

[0004] According to the first aspect, an operation and maintenance method based on a large language model is provided, including: iteratively performing the following operations through an intelligent agent: determining the link operation and maintenance results corresponding to the completed links in the operation and maintenance task; determining, through the large language model, a target tool for continuing to perform the operation and maintenance task from a tool set based on the link operation and maintenance results corresponding to the completed links; calling the target tool to generate the link operation and maintenance results corresponding to the next link in the operation and maintenance task.

[0005] According to the second aspect, an operation and maintenance device based on a large language model is provided, comprising the following processing units arranged in an intelligent body to perform iterative operations: a result determination unit, configured to determine the link operation and maintenance results corresponding to the completed links in the operation and maintenance task; a tool determination unit, configured to determine, through the large language model, the link operation and maintenance results corresponding to the completed links, from a tool set, a target tool for continuing to perform the operation and maintenance task; and a result generation unit, configured to call the target tool and determine the link operation and maintenance results corresponding to the next link in the operation and maintenance task.

[0006] According to a third aspect, an electronic device is provided, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so as to enable the at least one processor to execute the method described in any implementation manner of the first aspect.

[0007] According to a fourth aspect, a non-transitory computer-readable storage medium storing computer instructions is provided, where the computer instructions are used to cause a computer to execute the method as described in any implementation of the first aspect.

[0008] According to a fifth aspect, a computer program product is provided, comprising: a computer program, which implements the method described in any implementation manner of the first aspect when executed by a processor.

[0009] According to the technology disclosed in the present invention, an operation and maintenance method and device based on a large language model are provided. The AgentLoop method is used to determine the target tool for the next link in the operation and maintenance task based on the operation and maintenance results of the links corresponding to the completed links in the operation and maintenance task, so as to call the target tool to complete the next link in the operation and maintenance task, thereby improving the applicability of the operation and maintenance solution in various operation and maintenance scenarios and the accuracy of the operation and maintenance results.

[0010] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present disclosure, nor are they intended to limit the scope of the present disclosure. Other features of the present disclosure will become readily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] The accompanying drawings are provided to facilitate a better understanding of the present invention and do not constitute a limitation of the present disclosure.

[0012] Figure 1 is an exemplary system architecture diagram in which an embodiment of the present disclosure may be applied;

[0013] Figure 2 is a flowchart of an embodiment of an operation and maintenance method based on a large language model according to the present disclosure;

[0014] Figure 3 is a schematic diagram of an application scenario of the operation and maintenance method based on a large language model according to this embodiment;

[0015] Figure 4 is a flowchart of another embodiment of an operation and maintenance method based on a large language model according to the present disclosure;

[0016] Figure 5 is a structural diagram of an embodiment of an operation and maintenance device based on a large language model according to the present disclosure;

[0017] Figure 6 It is a schematic diagram of the structure of a computer system suitable for implementing the embodiments of the present disclosure. DETAILED DESCRIPTION

[0018] The following description of exemplary embodiments of the present disclosure is made in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding. These details should be considered as merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0019] In the technical solutions disclosed herein, the collection, storage, use, processing, transmission, provision and disclosure of user personal information involved comply with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0020] Figure 1 An exemplary architecture 100 is shown to which the operation and maintenance method and apparatus based on a large language model of the present disclosure can be applied.

[0021] like Figure 1 As shown, system architecture 100 may include terminal devices 101, 102, and 103, a network 104, and a server 105. The communication connections between terminal devices 101, 102, and 103 constitute a topological network, and network 104 is used to provide a medium for communication links between terminal devices 101, 102, and 103 and server 105. Network 104 may include various connection types, such as wired or wireless communication links or fiber optic cables.

[0022] Terminal devices 101, 102, and 103 can be hardware devices or software that support network connection for data interaction and data processing. When terminal devices 101, 102, and 103 are hardware, they can be various electronic devices that support network connection, information acquisition, interaction, display, processing, and other functions, including but not limited to smartphones, tablet computers, e-book readers, laptop computers, and desktop computers. When terminal devices 101, 102, and 103 are software, they can be installed in the electronic devices listed above. They can be implemented as multiple software or software modules, for example, to provide distributed services, or they can be implemented as a single software or software module. No specific limitations are given here.

[0023] Server 105 can be a server that provides various services, such as a backend processing server that receives operation and maintenance tasks sent by terminal devices 101, 102, and 103 and uses a large language model to determine the corresponding operation and maintenance results for each step in the operation and maintenance tasks. Optionally, the server can also provide feedback on the step operation and maintenance results to the terminal devices. For example, server 105 can be a cloud server.

[0024] It should be noted that a server can be either hardware or software. When a server is hardware, it can be implemented as a distributed server cluster consisting of multiple servers, or as a single server. When a server is software, it can be implemented as multiple software programs or software modules (for example, software or software modules used to provide distributed services), or as a single software program or software module. This is not specifically limited here.

[0025] It should also be noted that the large language model-based operation and maintenance methods provided in the embodiments of the present disclosure are generally executed by a server, but this does not rule out the possibility of execution by a terminal device, or of a server and a terminal device cooperating with each other. Accordingly, the various components (e.g., various units) included in the large language model-based operation and maintenance apparatus can be entirely located in the server, entirely located in the terminal device, or separately located in the server and the terminal device.

[0026] It should be understood that Figure 1 The number of terminal devices, networks, and servers in the system is merely illustrative. Any number of terminal devices, networks, and servers may be used as needed. When the electronic device on which the large language model-based operation and maintenance method is running does not need to transmit data to other electronic devices, the system architecture may only include the electronic device (e.g., terminal device or server) on which the large language model-based operation and maintenance method is running.

[0027] Please refer to Figure 2 , Figure 2 This is a flowchart of an operation and maintenance method based on a large language model provided by an embodiment of the present disclosure, wherein process 200 is iteratively executed by an intelligent agent. Process 200 includes the following steps:

[0028] Step 201: Determine the link operation and maintenance results corresponding to the completed links in the operation and maintenance task.

[0029] In this embodiment, the execution subject of the operation and maintenance method based on the large language model (for example, Figure 1 The server in the operation and maintenance task can determine the operation and maintenance results of the corresponding links in the operation and maintenance task.

[0030] Operation and maintenance tasks include, but are not limited to, network operation and maintenance tasks, system operation and maintenance tasks, application operation and maintenance tasks, data operation and maintenance tasks, security operation and maintenance tasks, and cloud operation and maintenance tasks. Network operation and maintenance tasks include, for example, network equipment management and maintenance, network performance monitoring and optimization, network troubleshooting and resolution, network security maintenance, network optimization and upgrades; system operation and maintenance tasks include, for example, server, operating system, and file system management and maintenance; application operation and maintenance tasks include, for example, application deployment, monitoring, and maintenance, and middleware management and maintenance; data operation and maintenance tasks include, for example, database management and maintenance, data warehouse and data lake management, and data quality management; security operation and maintenance tasks include, for example, security monitoring and auditing, vulnerability management, and security policy development and execution; and cloud operation and maintenance tasks include, for example, cloud resource management and cloud service management.

[0031] A complete operation and maintenance task typically includes multiple steps. For example, network vulnerability scanning and remediation tasks include scanning and assessment, remediation plan development, and remediation implementation and verification. Network equipment inspection tasks include device status inspection, performance indicator monitoring, and configuration verification.

[0032] In this embodiment, an intelligent agent is provided in the execution subject. In the process of determining the link operation and maintenance results corresponding to each link in the operation and maintenance task, the intelligent agent executes steps 201-203, that is, the intelligent agent first executes the above step 201 to determine the link operation and maintenance results corresponding to the completed link in the operation and maintenance task; then, the link operation and maintenance results corresponding to the completed link are input into the large language model, and based on the subsequent step 202, the target tool for the next link in the operation and maintenance task is determined by the large language model according to the link operation and maintenance results corresponding to the completed link in the operation and maintenance task; finally, the intelligent agent calls the target tool to generate the link operation and maintenance results corresponding to the next link in the operation and maintenance task. Among them, the completed link represents the link in which the target tool determined by the large language model generates its corresponding link operation and maintenance results. The next link is the next unfinished link after the completed link. In this way, the partial operation and maintenance task corresponding to a link in the operation and maintenance task is completed.

[0033] By iteratively executing steps 201-203, the links in the operation and maintenance task are processed sequentially by the agent. This iterative agent processing method is known as an Agent Loop. For example, the execution subject receives model prompt data (e.g., prompt words) from a large language model and, based on the large language model's powerful natural language understanding and data analysis capabilities, determines the operation and maintenance task represented by the model prompt data. It then splits the operation and maintenance task into multiple links, either serial, parallel, or a combination of serial and parallel. It then determines the processing order of the links based on the logical relationships between the multiple links. Each link is processed sequentially in this order. For the first link, there is no corresponding completed link, meaning the link operation and maintenance result for the completed link is empty. The large language model can directly determine the target tool based on the model prompt data representing the operation and maintenance task and the relevant information for the first link, thereby obtaining the link operation and maintenance result for the first link through subsequent steps 202 and 203.

[0034] For a non-first link in the operation and maintenance task, in response to executing the link, the link operation and maintenance result corresponding to the completed link in the operation and maintenance task is determined.

[0035] In some optional implementations of this embodiment, the execution entity may perform step 201 in the following manner: determining link operation and maintenance results corresponding to a preset number of completed links in the operation and maintenance task up to the current time.

[0036] The preset number can be flexibly set according to actual conditions, for example, the preset number is 10.

[0037] By setting a preset number, you can limit the number of link operation and maintenance results fed into the large language model. In some cases, an operation and maintenance task includes many links. As the operation and maintenance task executes, the link operation and maintenance results corresponding to completed links increase. Furthermore, the link operation and maintenance results corresponding to the completed links ranked earlier may not provide any assistance to the large language model's data processing process, but may instead increase the data processing pressure on the large language model.

[0038] In this implementation, the number of link operation and maintenance results input to the large language model is limited by a preset number, which reduces the data processing pressure of the large language model while ensuring the accuracy of the processing results of the large language model and improves the data processing efficiency of the large language model.

[0039] In step 202 , a target tool for continuing to perform the operation and maintenance task is determined from the tool set based on the operation and maintenance results of the completed links through the large language model.

[0040] In this embodiment, the execution subject may determine the target tool for continuing to perform the operation and maintenance task from the tool set according to the operation and maintenance results of the links corresponding to the completed links through the large language model.

[0041] The tool set includes tools used to perform operation and maintenance tasks. You can set up corresponding tool sets for different types of operation and maintenance tasks. Taking network operation and maintenance tasks as an example, the tool set includes device connection and login tools, device status monitoring tools, and device log analysis tools for network device inspections; network traffic monitoring tools and performance indicator monitoring tools for network performance monitoring; network connectivity testing tools and network packet capture and analysis tools for network troubleshooting; and vulnerability scanning tools, intrusion detection, and prevention tools for network security maintenance.

[0042] For example, the granularity (precision) of the types of O&M tasks can be set based on actual circumstances, allowing for a targeted tool set to be created for the type of O&M task. Specifically, O&M tasks can be divided into multiple nested levels of varying granularity based on a preset division hierarchy, and the correspondence between the types of O&M tasks and tools at the lowest level can be determined. Furthermore, users can select the type of O&M task they need based on their needs, resulting in a tool set consisting of the tools corresponding to the selected O&M task type. Tools corresponding to the type of O&M task are used to handle that type of O&M task.

[0043] For another example, the tool set includes tools applicable to various operation and maintenance tasks, so that the tool set has universal applicability to various operation and maintenance tasks.

[0044] Tools in the tool collection can support the Model Communication Protocol (MCP) or be customized to meet your needs. The MCP protocol is a communication protocol for models that defines how models interact, share data, and collaborate, allowing them to seamlessly exchange information and collaborate.

[0045] Tools in a tool collection can be implemented within the agent's process or integrated with the agent as external services via HTTP (Hypertext Transfer Protocol) or RPC (Remote Procedure Call). Implementing tools within the agent's process means developing and integrating them within the same process as the agent. This approach offers the advantages of low latency and efficient call processing, as tools and agents share the same memory space and system resources, eliminating the need for serialization and network transmission for data exchange.

[0046] Tools as external services refer to tools that call external services via protocols such as HTTP and RPC. This approach deploys tools as independent services, with agents interacting with them via standard communication protocols. Advantages include decoupling, high flexibility, and ease of scalability and maintenance. Different services can be developed, deployed, and scaled independently.

[0047] In this embodiment, the link operation and maintenance results corresponding to the completed links are input into the large language model. Based on its powerful natural language understanding and data analysis capabilities, the large language model determines the target tool to continue to perform the operation and maintenance task from the tool set.

[0048] The links in an operation and maintenance task may change. Taking the network vulnerability scanning and repair task as an example, it generally includes the scanning and assessment link, the repair plan formulation link, and the repair implementation and verification link. However, when the operation results of the link corresponding to the scanning and assessment link indicate that no vulnerability is found, the subsequent repair plan formulation link, repair implementation and verification link will no longer be executed.

[0049] For operation and maintenance tasks whose included links may change, the above-mentioned execution entity can determine the next link to be executed through the large language model based on the link operation and maintenance results corresponding to the completed links; based on the determined next link and the link operation and maintenance results corresponding to the completed links, determine the target tool from the tool collection to continue executing the operation and maintenance task.

[0050] For operation and maintenance tasks with fixed links, the above-mentioned execution entity can directly use the large language model to determine the target tool to continue executing the operation and maintenance task from the tool set based on the link operation and maintenance results corresponding to the completed links and the next link to be executed determined based on the split operation.

[0051] In some optional implementations of this embodiment, the above-mentioned execution entity can also perform the above-mentioned step 202 in the following manner: through a large language model, according to the model prompt data representing the operation and maintenance task and the link operation and maintenance results corresponding to the completed links, determine the target tool from the tool set.

[0052] For each link in the operation and maintenance task, when determining the target tool corresponding to the link, it is necessary to input the model prompt data representing the operation and maintenance task and the link operation and maintenance results corresponding to the completed link into the large language model. The large language model combines the model prompt data and the link operation and maintenance results corresponding to the completed link to determine the target tool to continue executing the operation and maintenance task from the tool set.

[0053] When the links in the operation and maintenance task may change, the large language model determines the next link to be executed based on the model prompt data and the link operation and maintenance results corresponding to the completed links; based on the prompt data, the determined next link and the link operation and maintenance results corresponding to the completed links, the target tool to continue executing the operation and maintenance task is determined from the tool collection.

[0054] For operation and maintenance tasks with fixed links, the above-mentioned execution entity can directly use the large language model to determine the target tool to continue executing the operation and maintenance task from the tool set based on the model prompt data, the link operation and maintenance results corresponding to the completed link, and the next link to be executed.

[0055] In this implementation, on the basis of considering the operation and maintenance results of the links corresponding to the completed links, the model prompt data is further considered. Under the guidance of the model prompt data, the accuracy of determining the target tool can be further improved.

[0056] In some optional implementations of this embodiment, the above-mentioned execution entity can execute the target tool determination process in the following manner: through a large language model, the target tool is determined from the tool set based on the model prompt data, the link operation and maintenance results corresponding to the completed links, and the configuration data of the tools in the tool set.

[0057] The configuration data of a tool is a collection of various parameters and settings required for the normal operation and task execution of the tool, which helps the large language model understand the functions of the tool and accurately call the tool.

[0058] The above-mentioned intelligent agent inputs the model prompt data, the link operation and maintenance results corresponding to the completed links, and the configuration data of the tools in the tool set into the large language model. The large language model deeply understands and analyzes the above three types of data and determines the target tool from the tool set.

[0059] In this implementation, the large language model further considers the configuration data of the tool based on the model prompt data and the operation and maintenance results of the completed links. This allows the large language model to determine the functions of each tool based on the configuration data of the tools in the tool set, so as to determine the tool suitable for the next link, further improving the accuracy of the target tool.

[0060] In some optional implementations of this embodiment, configuration data includes tool description data and tool parameter data. Tool description data is used to explain the tool's primary purpose and the specific problems it can solve. More detailed tool description data also includes data such as the tool's basic operating mechanism and the tool's operating steps. Tool parameter data refers to the configuration options and input values ​​required for the tool to run or perform specific tasks. These parameters are used to customize the tool's behavior, enabling it to operate according to the user's specific needs and environment.

[0061] In this implementation, the above-mentioned execution entity can execute the target tool determination process in the following manner: through a large language model, based on the model prompt data, the link operation and maintenance results corresponding to the completed links, and the tool description data and tool parameter data of the tools in the tool collection, determine the target tool from the tool collection and determine the target parameter value corresponding to the tool parameter data of the target tool.

[0062] The above-mentioned intelligent agent inputs the model prompt data, the link operation and maintenance results corresponding to the completed links, and the tool description data and tool parameter data of the tools in the tool collection into the large language model. The large language model determines the target tool from the tool collection and, based on the tool parameter data of the target tool, determines the target parameter value corresponding to the tool parameter data, so that the target tool can execute the next link of the operation and maintenance task based on the target parameter value.

[0063] For example, the large language model first identifies each parameter's name, type, and value range based on the target tool's parameter data. It then matches the completed phase's operational results with the target tool's parameter requirements to determine appropriate initial values ​​for each parameter. Finally, based on actual conditions and historical experience, the parameter values ​​are adjusted and optimized to ensure the tool performs its task optimally.

[0064] In this implementation, the large language model further determines the target parameter value corresponding to the tool parameter data of the target tool on the basis of determining the target tool of the next link, thereby improving the richness of the generated information and providing accurate data basis for the subsequent tool calling process.

[0065] In some optional implementations of this embodiment, the execution subject may further perform the following operations: setting configuration data of the tools in the tool set through at least one of a model context protocol, a configuration file, and a database.

[0066] MCP is a protocol for passing information within a model context. Tool configuration data can be passed from related services or modules through MCP to the large language model that requires the tool. When executing a task, the large language model can use the configuration data received through MCP to select the appropriate tool to assist in completing the task.

[0067] The configuration method based on the model context protocol can pass tool information updates to the large language model in real time, and can dynamically adjust the passed configuration data according to the needs of the large language model or the scenario of the operation and maintenance task.

[0068] In the configuration file-based configuration method, tool configuration data is stored in the configuration file in JSON (JavaScript Object Notation) format. JSON is a lightweight data exchange format that is easy to read and write, and also easy for machines to parse and generate. Information such as the tool name, description, and parameters are organized in the JSON file as key-value pairs.

[0069] In a database-based configuration approach, a database is used to store tool configuration data. A database table structure can be designed to store data such as tool names, tool descriptions, and tool parameter data. For example, a configuration data table might have the following fields: tool ID (Identity Document), tool name, tool description, and parameter definitions (which can be a JSON string field to store detailed parameter information such as parameter name, type, and description). A database query is used to retrieve a list of tool information, which is then formatted into JSON and provided to models or other components that require tool information.

[0070] This implementation provides multiple configuration methods for configuration files, which improves the flexibility of the configuration process.

[0071] Step 203: call the target tool to generate the link operation and maintenance result corresponding to the next link in the operation and maintenance task.

[0072] In this embodiment, the execution entity may call the target tool to generate a link operation and maintenance result corresponding to the next link in the operation and maintenance task.

[0073] Continuing with the example of a network vulnerability scanning and remediation task consisting of scanning and assessment, remediation plan development, and remediation implementation and verification, the target tool for the scanning and assessment configuration phase is Nessus (a vulnerability scanning tool). Nessus sends various detection requests to the target devices corresponding to the O&M task, simulating potential attack behaviors to identify vulnerabilities on the devices. During the scan, Nessus displays real-time information such as the scan progress and the number of discovered vulnerabilities. After the scan is complete, Nessus generates a detailed scan report, including the vulnerability name, description, severity, and affected devices. Based on the vulnerability severity classification in the report, high-risk vulnerabilities are prioritized, and the potential impact of each vulnerability is analyzed, such as whether it could cause a system crash or data leakage.

[0074] When developing a remediation plan, the corresponding tool is Qualys Vulnerability Management. First, Qualys Vulnerability Management further analyzes and categorizes the imported vulnerability data. Vulnerabilities are prioritized based on factors such as severity, difficulty of exploitation, and the importance of the affected assets. For example, high-risk vulnerabilities that are easily exploitable and impact critical business systems are given a higher priority. Based on the analysis results, Qualys then generates remediation recommendations for each vulnerability. These recommendations include specific remediation steps, official patch links, and configuration modification methods. Taking into account different enterprise technical environments and business needs, remediation recommendations may offer a variety of solutions, such as immediate patch installation and temporary configuration changes. Finally, Qualys uses the remediation recommendations to develop a detailed remediation plan, combining the enterprise's specific situation with the remediation recommendations. Remediation timelines and responsible individuals are determined for each vulnerability, and remediation is prioritized to ensure timely remediation of high-risk vulnerabilities. Furthermore, reminders and tracking mechanisms are implemented for each remediation task to ensure smooth progress.

[0075] During the patch implementation and verification phase, the corresponding tool is Microsoft Patch Management. First, obtain the corresponding vulnerability patch based on the patch link or information provided in the patch plan. Before officially deploying the patch, establish a test environment similar to the production environment, install the patch on the devices in the test environment, and conduct comprehensive functional and stability testing. This testing includes verifying system operation, application compatibility, and whether business processes are impacted. After passing the testing, use MS Patch Management to deploy the patch to the target devices in the production environment. You can choose between batch or gradual deployment according to the pre-defined deployment plan and strategy to ensure a smooth patch installation process. Furthermore, closely monitor the device status during the patch installation process and promptly address any installation failures or system anomalies. After the patch is installed, scan the patched devices again using a vulnerability scanning tool (such as Nessus) to verify that the vulnerability has been successfully fixed. Check the scan report to confirm that the vulnerability status has changed from "Unfixed" to "Fixed." Manually check the device configuration and functionality to ensure system operation and security. Finally, document the entire patch process, including the time of repair, device, vulnerability information, and patch version. If problems arise during the repair process or the patch causes new failures, initiate a rollback plan promptly to restore the system to its previous stable state. You can use previously backed-up system configurations or data for the rollback operation to ensure business continuity and stability.

[0076] In this embodiment, the execution entity may iteratively execute steps 201-203 until a preset termination condition is met. The preset termination condition includes, but is not limited to, completion of the operation and maintenance task, the number of tool calls exceeding a preset threshold, and, based on the operation and maintenance results of the completed stage, determining that the target tool for the next stage does not exist in the tool set.

[0077] In some optional implementations of this embodiment, the execution subject may perform step 203 as follows: calling a target tool that uses a target parameter value to generate a link operation and maintenance result corresponding to the next link in the operation and maintenance task.

[0078] In this implementation, the target tool using the target parameter value is more compatible with the next link in the operation and maintenance task, which helps to further improve the accuracy of the link operation and maintenance results corresponding to the next link.

[0079] In some optional implementations of this embodiment, the execution entity may execute the process of generating the link operation and maintenance results in the following manner:

[0080] The first step is to determine the target data source based on the target tool and target parameter values.

[0081] For example, first, analyze the target tool's configuration options to determine the supported data source types (such as local files, databases, network services, etc.). Then, locate the specific data source based on the tool parameter values ​​and the target tool's data analysis requirements. Finally, check the data source's availability and data integrity to ensure that the data can be read correctly.

[0082] As another example, first, based on the business logic and process of the operation and maintenance task, identify the data sources required for each step. Then, based on the target tool's parameter values ​​and business requirements, infer reasonable data sources. Finally, verify the inferred data sources are reasonable and make adjustments if any deviations are found.

[0083] The second step is to call the target tool using the target parameter value and generate the link operation and maintenance results corresponding to the next link in the operation and maintenance task based on the data in the target data source.

[0084] After determining the target data source, the above-mentioned intelligent agent calls the target tool, and the target tool adopts the target parameter value to generate the link operation and maintenance results corresponding to the next link in the operation and maintenance task based on the data in the target data source.

[0085] This implementation dynamically and intelligently selects and orchestrates the most appropriate analytical tools and data sources. This "thinking" automated process makes the O&M solution applicable to a wider range of problem scenarios.

[0086] In some optional implementations of this embodiment, the above-mentioned execution entity can perform the above-mentioned second step in the following manner: in the process of processing the data in the target data source by the target tool using the target parameter value, combined with the large language model to perform auxiliary analysis and reasoning, and generate the link operation and maintenance results corresponding to the next link in the operation and maintenance task.

[0087] During the target tool's information processing, some or all of the data processing operations can be analyzed and reasoned using the large language model. That is, the tool itself can call the large language model for reasoning and analysis. For example, the target tool can call the large language model through an API (Application Programming Interface). Another example is that the large language model's SDK (Software Development Kit) is embedded in the tool's codebase. Within the target tool, the data to be analyzed is organized according to the format specified by the SDK. Simultaneously, the SDK is used to initialize and configure the large language model, including setting model parameters (such as maximum generation length and temperature). Data is then input into the embedded model, and the SDK's reasoning interface is called to initiate model analysis.

[0088] In this implementation, the data processing process of the target tool can be further combined with a large language model for auxiliary analysis and reasoning, which helps to further improve the accuracy of the link operation and maintenance results.

[0089] Continue to see Figure 3 , Figure 3 It is a schematic diagram of an application scenario 300 of the operation and maintenance method based on a large language model according to the present embodiment. User 301 sends model prompt data representing the operation and maintenance task to the intelligent agent in the server 303 through the terminal device 302. The intelligent agent inputs the model prompt data into the large language model, and the large language model determines the target tool for executing the first link in the operation and maintenance task, that is, the first link target tool, and calls the first link target tool to generate the link operation and maintenance result corresponding to the first link in the operation and maintenance task, that is, the first link operation and maintenance result. In the subsequent processing process, the intelligent agent iteratively performs the following operations: determining the link operation and maintenance result corresponding to the completed link in the operation and maintenance task; through the large language model, according to the link operation and maintenance result corresponding to the completed link, determining the target tool for continuing to execute the operation and maintenance task from the tool set; calling the target tool to generate the link operation and maintenance result corresponding to the next link in the operation and maintenance task.

[0090] For example, after completing the first stage of the operation and maintenance task, the intelligent agent first determines the first stage operation and maintenance result corresponding to the first stage in the operation and maintenance task; then, through the large language model, based on the first stage operation and maintenance result corresponding to the first stage, it determines the second stage target tool from the tool set to continue to execute the operation and maintenance task; finally, it calls the second stage target tool to generate the second stage operation and maintenance result corresponding to the second stage in the operation and maintenance task.

[0091] After completing the second stage of the operation and maintenance task, the intelligent agent first determines the first stage operation and maintenance results corresponding to the first stage in the operation and maintenance task and the second stage operation and maintenance results corresponding to the second stage; then, through the large language model, based on the first stage operation and maintenance results corresponding to the first stage and the second stage operation and maintenance results corresponding to the second stage, the intelligent agent determines the third stage target tool from the tool set to continue to execute the operation and maintenance task; finally, the third stage target tool is called to generate the third stage operation and maintenance results corresponding to the third stage in the operation and maintenance task.

[0092] In this embodiment, an operation and maintenance method based on a large language model is provided. The Agent Loop method is used to determine the target tool for the next link in the operation and maintenance task based on the operation and maintenance results of the links corresponding to the completed links in the operation and maintenance task, so as to call the target tool to complete the next link in the operation and maintenance task, thereby improving the applicability of the operation and maintenance solution in various operation and maintenance scenarios and the accuracy of the operation and maintenance results.

[0093] In some optional implementations of this embodiment, the above-mentioned execution entity may also perform the following operations: during the execution of the operation and maintenance task, in response to the number of tool calls up to the current time exceeding a preset threshold, a total operation and maintenance result is generated based on the operation and maintenance results corresponding to the completed links.

[0094] The preset number threshold can be flexibly set according to actual conditions, for example, the number of tool calls is 20.

[0095] By setting a preset threshold, you can limit the number of times a large language model can call a tool, avoiding the illusion of a large language model and always returning a callable tool. During the execution of an operation and maintenance task, when the number of tool calls reaches the preset threshold, the agent will no longer initiate new calls and will exit the agent loop, returning a prompt to the user indicating that the agent has exited the loop.

[0096] The large language model can summarize and analyze the data of the operation and maintenance results corresponding to the completed links to obtain the overall operation and maintenance results.

[0097] In this implementation, by setting a preset threshold, the number of times the large language model calls the tool is limited, thereby avoiding the hallucination problem of the large language model and ensuring the effectiveness of the data processing process of the large language model.

[0098] In some optional implementations of this embodiment, the above-mentioned execution entity may also perform the following operations: in response to the large language model, based on the link operation and maintenance results corresponding to the completed links, it is determined that the target tool does not exist in the tool set, and based on the link operation and maintenance results corresponding to the completed links, a total operation and maintenance result is generated.

[0099] The large language model determines that the target tool does not exist in the tool set based on the operation and maintenance results of the links corresponding to the completed links, indicating that the operation and maintenance task cannot be continued. The large language model can summarize and analyze the data of the operation and maintenance results of the links corresponding to the completed links to obtain the overall operation and maintenance results.

[0100] In this implementation, a method for generating a total operation and maintenance result is provided, which improves the comprehensiveness of the output processing.

[0101] In some optional implementations of this embodiment, the above-mentioned execution entity may also perform the following operations: in response to completing the operation and maintenance task, determine the overall operation and maintenance result of the operation and maintenance task based on the link operation and maintenance result corresponding to the last link of the operation and maintenance task.

[0102] The completed operation and maintenance task representation uses the tools corresponding to each stage of the operation and maintenance task to complete the partial operation and maintenance tasks of the stage and obtain the stage processing results. In this case, the execution entity can directly determine the stage operation and maintenance results corresponding to the last stage of the operation and maintenance task as the overall operation and maintenance result of the operation and maintenance task. Alternatively, the large language model can focus on the stage operation and maintenance results corresponding to the last stage, and combine the stage operation and maintenance results corresponding to other completed stages to summarize and analyze the data to obtain the overall operation and maintenance result.

[0103] In this implementation, a method for generating a total operation and maintenance result is provided, which improves the comprehensiveness of the output processing.

[0104] In some optional implementations of this embodiment, the above-mentioned execution entity can also perform the following operations: call the first output tool to generate and display the total result data in a first specified format based on the total operation and maintenance results of the operation and maintenance task; and / or call the second output tool to generate and display the link result data in a second specified format based on the link operation and maintenance results.

[0105] Both the first and second output tools have the function of outputting data results. The first and second output tools can be the same or different. For different steps in the operation and maintenance task, the corresponding second output tools can be the same or different. In this implementation, the destination node for output data can also be determined based on user needs, including but not limited to IM (Instant Messaging) work groups, text messages, phone calls, web pages, etc.

[0106] In this implementation, you can flexibly select at least one of the following output methods based on your needs:

[0107] 1. After the operation and maintenance task is completed, the first output tool is called to generate and display the total result data in the first specified format according to the total operation and maintenance result of the operation and maintenance task.

[0108] 2. During the execution of the operation and maintenance task, the second output tool is called to generate and display the link result data in the second specified format according to the link operation and maintenance results.

[0109] 3. During the execution of the operation and maintenance task, the second output tool is called to generate and display the link result data in the second specified format based on the link operation and maintenance results; after the operation and maintenance task is completed, the first output tool is called to generate and display the total result data in the first specified format based on the total operation and maintenance results of the operation and maintenance task.

[0110] In this implementation, the output method of the link operation and maintenance results and the total result data can be flexibly set, which helps to improve the user's information acquisition efficiency and experience.

[0111] Continue to refer Figure 4 , shows a schematic process 400 of another embodiment of the operation and maintenance method based on a large language model according to the present disclosure. In the process 400, the following steps are included:

[0112] Step 401, perform the following operations through agent iteration:

[0113] Step 4011: Determine the link operation and maintenance results corresponding to a preset number of completed links in the operation and maintenance task up to the current time.

[0114] Step 4012, through the large language model, based on the model prompt data, the link operation and maintenance results corresponding to the completed link, and the tool description data and tool parameter data of the tools in the tool set, determine the target tool from the tool set and determine the target parameter value corresponding to the tool parameter data of the target tool.

[0115] Step 4013: Determine the target data source based on the target tool and target parameter values.

[0116] Step 4014: call the target tool using the target parameter value, and generate the link operation and maintenance results corresponding to the next link in the operation and maintenance task based on the data in the target data source.

[0117] Step 402 : During the execution of the operation and maintenance task, in response to the number of tool calls to date exceeding a preset threshold, a total operation and maintenance result is generated based on the operation and maintenance results of the corresponding completed links.

[0118] Step 403 , in response to determining that the target tool does not exist in the tool set based on the link operation and maintenance results corresponding to the completed links through the large language model, a total operation and maintenance result is generated based on the link operation and maintenance results corresponding to the completed links.

[0119] Step 404 , in response to the completion of the operation and maintenance task, the overall operation and maintenance result of the operation and maintenance task is determined according to the link operation and maintenance result corresponding to the last link of the operation and maintenance task.

[0120] The process 400 of the operation and maintenance method based on the large language model in this embodiment specifically illustrates the process of determining the link operation and maintenance results and the process of determining the overall operation and maintenance results. It adopts the Agent Loop method to determine the target tool for the next link in the operation and maintenance task based on the link operation and maintenance results corresponding to the completed links in the operation and maintenance task, so as to call the target tool to complete the next link in the operation and maintenance task, and generate the overall operation and maintenance results in a targeted manner based on different situations, thereby further improving the applicability of the operation and maintenance plan in various operation and maintenance scenarios and the accuracy of the operation and maintenance results.

[0121] To further illustrate the present disclosure, the specific data processing process is described by taking the operation and maintenance task of latency analysis in a high-performance network as an example:

[0122] 1. The user queries the current latency data of the target training task related to the data center on the latency page;

[0123] 2. Receive the user's click operation on the "Smart Analysis" button on the web page;

[0124] 3. Pass the ID and delay data of the target training task to the agent, and start the agent loop;

[0125] 4. The agent submits the prompt words and tool set representing the delay analysis to the large language model;

[0126] 5. The large language model returns the link operation and maintenance results, which include the call to the latency analysis tool and the parameter values ​​of the latency analysis tool;

[0127] 6. The latency analysis tool re-pulls the latency data of the target training task based on its ID and parameters, and also pulls the topology data of the target training task.

[0128] 7. The latency analysis tool submits latency data and topology data to the large language model, which generates analysis results.

[0129] 8. The large language model returns the latency analysis results of this target training task. The latency analysis results are in Markdown (Lightweight Markup Language) format.

[0130] 9. The agent calls a web reporting tool to generate a professional and beautiful latency report based on the Markdown format results, uploads the report to the specified URL (Uniform Resource Locator), and stores it in the database;

[0131] 10. The agent returns the report URL to the user.

[0132] Continue to refer Figure 5 As an implementation of the methods shown in the above figures, the present disclosure provides an embodiment of an operation and maintenance device based on a large language model. Figure 2 Corresponding to the method embodiment shown, the system can be specifically applied to various electronic devices.

[0133] like Figure 5As shown, the large language model-based operation and maintenance device 500 includes the following processing units, located within an intelligent agent and performing iterative operations: a result determination unit 501, configured to determine the operation and maintenance results for a completed step in the operation and maintenance task; a tool determination unit 502, configured to use the large language model to determine, from a tool set, a target tool to continue executing the operation and maintenance task based on the step operation and maintenance results for the completed step; and a result generation unit 503, configured to invoke the target tool and determine the step operation and maintenance result for the next step in the operation and maintenance task.

[0134] In some optional implementations of this embodiment, the tool determination unit 502 is further configured to: determine the target tool from the tool set through a large language model, according to the model prompt data representing the operation and maintenance task and the link operation and maintenance results corresponding to the completed links.

[0135] In some optional implementations of this embodiment, the tool determination unit 502 is further configured to: determine the target tool from the tool set through a large language model according to the model prompt data, the link operation and maintenance results corresponding to the completed link, and the configuration data of the tools in the tool set.

[0136] In some optional implementations of this embodiment, the configuration data includes tool description data and tool parameter data, and the tool determination unit 502 is further configured to: through a large language model, determine the target tool from the tool collection based on the model prompt data, the link operation and maintenance results corresponding to the completed link, and the tool description data and tool parameter data of the tools in the tool collection, and determine the target parameter value corresponding to the tool parameter data of the target tool.

[0137] In some optional implementations of this embodiment, the result generation unit 503 is further configured to: call a target tool that adopts a target parameter value to generate a link operation and maintenance result corresponding to the next link in the operation and maintenance task.

[0138] In some optional implementations of this embodiment, the result generation unit 503 is further configured to: determine the target data source based on the target tool and the target parameter value; call the target tool that adopts the target parameter value, and generate the link operation and maintenance results corresponding to the next link in the operation and maintenance task based on the data in the target data source.

[0139] In some optional implementations of this embodiment, the result generation unit 503 is further configured to: in the process of processing the data in the target data source by the target tool using the target parameter value, combine the large language model to perform auxiliary analysis and reasoning, and generate the link operation and maintenance results corresponding to the next link in the operation and maintenance task.

[0140] In some optional implementations of this embodiment, the above-mentioned device also includes: a tool configuration unit (not shown in the figure), which is configured to: set the configuration data of the tools in the tool collection through at least one of a model context protocol, a configuration file, and a database.

[0141] In some optional implementations of this embodiment, the result determination unit 501 is further configured to: determine the link operation and maintenance results corresponding to a preset number of completed links in the operation and maintenance task up to the current time.

[0142] In some optional implementations of this embodiment, the result generation unit 503 is also configured to: during the execution of the operation and maintenance task, in response to the number of tool calls up to the current time exceeding a preset threshold, generate a total operation and maintenance result based on the link operation and maintenance results corresponding to the completed links.

[0143] In some optional implementations of this embodiment, the result generation unit 503 is further configured to: in response to the large language model, determine that the target tool does not exist in the tool set based on the link operation and maintenance results corresponding to the completed links, and generate a total operation and maintenance result based on the link operation and maintenance results corresponding to the completed links.

[0144] In some optional implementations of this embodiment, the result generation unit 503 is further configured to: in response to completing the operation and maintenance task, determine the total operation and maintenance result of the operation and maintenance task according to the link operation and maintenance result corresponding to the last link of the operation and maintenance task.

[0145] In some optional implementations of this embodiment, the above-mentioned device also includes a result output unit (not shown in the figure), which is configured to: call a first output tool to generate and display total result data in a first specified format based on the total operation and maintenance results of the operation and maintenance task; and / or call a second output tool to generate and display link result data in a second specified format based on the link operation and maintenance results.

[0146] In this embodiment, an operation and maintenance device based on a large language model is provided. The Agent Loop method is adopted to determine the target tool for the next link in the operation and maintenance task according to the operation and maintenance results corresponding to the completed links in the operation and maintenance task, so as to call the target tool to complete the next link in the operation and maintenance task, thereby improving the applicability of the operation and maintenance plan in various operation and maintenance scenarios and the accuracy of the operation and maintenance results.

[0147] According to an embodiment of the present disclosure, the present disclosure also provides an electronic device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can implement the operation and maintenance method based on the large language model described in any of the above embodiments when executing.

[0148] According to an embodiment of the present disclosure, the present disclosure further provides a readable storage medium, which stores computer instructions, and the computer instructions are used to enable a computer to implement the operation and maintenance method based on a large language model described in any of the above embodiments when executed.

[0149] An embodiment of the present disclosure provides a computer program product, which, when executed by a processor, can implement the operation and maintenance method based on a large language model described in any of the above embodiments.

[0150] Figure 6 A schematic block diagram of an example electronic device 600 that can be used to implement embodiments of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are provided as examples only and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0151] like Figure 6 As shown, device 600 includes a computing unit 601, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 602 or a computer program loaded from a storage unit 608 into a random access memory (RAM) 603. RAM 603 may also store various programs and data required for the operation of device 600. Computing unit 601, ROM 602, and RAM 603 are interconnected via a bus 604. An input / output (I / O) interface 605 is also connected to bus 604.

[0152] Various components in device 600 are connected to I / O interface 605, including an input unit 606, such as a keyboard, mouse, etc.; an output unit 607, such as various types of displays, speakers, etc.; a storage unit 608, such as a magnetic disk, optical disk, etc.; and a communication unit 609, such as a network card, modem, wireless communication transceiver, etc. The communication unit 609 allows device 600 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0153] The computing unit 601 can be any general-purpose and / or specialized processing component with processing and computing capabilities. Some examples of the computing unit 601 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various specialized artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. The computing unit 601 performs the various methods and processes described above, such as the large language model-based operation and maintenance method. For example, in some embodiments, the large language model-based operation and maintenance method can be implemented as a computer software program tangibly embodied in a machine-readable medium, such as the storage unit 608. In some embodiments, part or all of the computer program can be loaded and / or installed onto the device 600 via the ROM 602 and / or the communication unit 609. When the computer program is loaded into the RAM 603 and executed by the computing unit 601, one or more steps of the large language model-based operation and maintenance method described above can be performed. Alternatively, in other embodiments, the computing unit 601 may be configured to execute the operation and maintenance method based on the large language model in any other appropriate manner (for example, by means of firmware).

[0154] Various embodiments of the systems and techniques described above can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), system-on-chip systems (SOCs), programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include being implemented in one or more computer programs that are executable and / or interpreted on a programmable system that includes at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0155] The program code for implementing the method of the present disclosure can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable large language model-based operation and maintenance device, so that when the program code is executed by the processor or controller, the functions / operations specified in the flowchart and / or block diagram are implemented. The program code can be executed entirely on the machine, partially on the machine, as a stand-alone software package, partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0156] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of machine-readable storage media may include an electrical connection based on one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), optical fibers, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0157] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0158] The systems and techniques described herein can be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network (LAN), a wide area network (WAN), and the Internet.

[0159] A computer system may include a client and a server. The client and server are typically remote from each other and typically interact via a communication network. This client-server relationship arises through computer programs running on the respective computers, establishing a client-server relationship. The server can be a cloud server, also known as a cloud computing server or cloud host. This server is a host product within the cloud computing service ecosystem that addresses the management difficulties and limited scalability of traditional physical hosts and virtual private server (VPS) services. It can also be a server in a distributed system or a server integrated with blockchain.

[0160] According to the technical solution of the embodiment of the present disclosure, an operation and maintenance method and device based on a large language model are provided. The Agent Loop method is adopted to determine the target tool for the next link in the operation and maintenance task according to the operation and maintenance results of the links corresponding to the completed links in the operation and maintenance task, so as to call the target tool to complete the next link in the operation and maintenance task, thereby improving the applicability of the operation and maintenance solution in various operation and maintenance scenarios and the accuracy of the operation and maintenance results.

[0161] It should be understood that the various forms of the processes shown above can be used to reorder, add, or delete steps. For example, the steps described in this disclosure can be performed in parallel, sequentially, or in a different order, as long as the desired results of the technical solutions provided by this disclosure can be achieved. This is not a limitation herein.

[0162] The above specific embodiments do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions may be made based on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure shall be included within the scope of protection of this disclosure.

Claims

1. An operation and maintenance method based on a large language model, comprising: The following operations are performed through agent iteration: Determine the operation and maintenance results of the corresponding links in the operation and maintenance tasks; Determining, through a large language model, a target tool from the tool set that continues to execute the operation and maintenance task based on the link operation and maintenance results corresponding to the completed link, model prompt data, and tool description data and tool parameter data of the tools in the tool set, and determining a target parameter value corresponding to the tool parameter data of the target tool; determining a target data source according to the target tool and the target parameter value; During the process of processing the data in the target data source by the target tool using the target parameter value, the large language model is combined to perform auxiliary analysis and reasoning to generate the link operation and maintenance results corresponding to the next link in the operation and maintenance task.

2. The method according to claim 1, wherein Also includes: The tool description data and tool parameter data of the tools in the tool set are set by at least one of a model context protocol, a configuration file, and a database.

3. The method according to claim 1, wherein The determination of the link operation and maintenance results corresponding to the completed links in the operation and maintenance task includes: Determine the link operation and maintenance results corresponding to a preset number of completed links in the operation and maintenance task up to the present.

4. The method according to any one of claims 1 to 3, wherein Also includes: During the execution of the operation and maintenance task, in response to the number of tool calls up to the present exceeding a preset threshold, a total operation and maintenance result is generated according to the operation and maintenance results of the links corresponding to the completed links.

5. The method according to any one of claims 1 to 3, wherein Also includes: In response to the large language model, based on the link operation and maintenance results corresponding to the completed links, it is determined that the target tool does not exist in the tool set, and a total operation and maintenance result is generated based on the link operation and maintenance results corresponding to the completed links.

6. The method according to any one of claims 1 to 3, wherein Also includes: In response to completing the operation and maintenance task, the overall operation and maintenance result of the operation and maintenance task is determined according to the link operation and maintenance result corresponding to the last link of the operation and maintenance task.

7. The method according to any one of claims 1 to 3, wherein Also includes: Invoking a first output tool to generate and display total result data in a first specified format based on the total operation and maintenance result of the operation and maintenance task; and / or The second output tool is called to generate and display the link result data in the second specified format according to the link operation and maintenance results.

8. An operation and maintenance device based on a large language model, comprising the following processing unit disposed in an agent and performing iterative operations: A result determination unit is configured to determine the link operation and maintenance results corresponding to the completed links in the operation and maintenance task; a tool determination unit configured to determine, through a large language model, a target tool from the tool set that continues to perform the operation and maintenance task based on the operation and maintenance results of the link corresponding to the completed link, model prompt data, and tool description data and tool parameter data of the tools in the tool set, and determine a target parameter value corresponding to the tool parameter data of the target tool; The result generating unit is configured to determine the target data source according to the target tool and the target parameter value, and in the process of processing the data in the target data source by the target tool adopting the target parameter value, combine the large language model to perform auxiliary analysis and reasoning to determine the link operation and maintenance result corresponding to the next link in the operation and maintenance task.

9. An electronic 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 stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium storing computer instructions, characterized in that: The computer instructions are used to cause the computer to execute the method according to any one of claims 1 to 7.

11. A computer program product comprising: A computer program which, when executed by a processor, implements the method according to any one of claims 1 to 7.

Images