Network element fault repair method and device, equipment, storage medium and product

Abstract

The invention relates to the technical field of computers, and particularly provides a network element fault repairing method and device, equipment, a storage medium and a product, and the method comprises the steps: receiving a work order push message, the work order push message comprises at least one piece of alarm information, and the alarm information refers to information sent when a corresponding network element has a fault; selecting a target agent group of each piece of alarm information from a plurality of agent groups constructed in advance, wherein the agent group comprises at least one agent; determining a target agent of each piece of alarm information by using the target agent group of each piece of alarm information; and according to the target intelligent agent of each piece of alarm information, carrying out fault repair processing on the network element corresponding to each piece of alarm information. According to the technical scheme, the network element repairing efficiency is improved.

Description

Technical Field

[0001] This disclosure relates to the field of computer technology, and in particular to a method, apparatus, device, storage medium, and product for repairing network element faults. Background Technology

[0002] With the continuous development of communication technology, the structure and application methods of network elements such as base stations and satellite base stations are becoming increasingly complex, and the frequency of network element failures is also gradually increasing. In order to ensure the normal operation of the network, when a network element fails, it is necessary to repair it in a timely manner to reduce problems such as communication quality degradation caused by network element failures.

[0003] To address network element fault repair needs, the usual method is to manually issue repair commands and distribute them to the faulty network element. However, this manual repair method is inefficient and cannot handle a large number of fault repair issues, resulting in untimely and inaccurate network element fault repair. Summary of the Invention

[0004] This disclosure is made in view of the above-mentioned problems. This disclosure provides a method, apparatus, device, storage medium, and product for repairing network element faults.

[0005] According to one aspect of this disclosure, a method for repairing network element faults is provided, comprising:

[0006] Receive work order push messages. The work order push message includes at least one alarm message. The alarm message refers to the information sent when the corresponding network element has a fault.

[0007] Select the target intelligent agent group for each alarm message from multiple pre-built intelligent agent groups, wherein the intelligent agent group includes at least one intelligent agent;

[0008] By utilizing the target agent group of each alarm message, the target agent of each alarm message can be determined;

[0009] Based on the target intelligent agent of each alarm message, fault repair processing is performed on the network element corresponding to each alarm message.

[0010] According to one aspect of this disclosure, a network element fault repair device is provided, comprising:

[0011] The message receiving unit is used to receive work order push messages. The work order push message includes at least one alarm message, which refers to the information sent when the corresponding network element has a fault.

[0012] A group selection unit is used to select the target intelligent agent group for each alarm information from multiple pre-built intelligent agent groups, wherein the intelligent agent group includes at least one intelligent agent;

[0013] The target determination unit is used to determine the target intelligent agent for each alarm information by utilizing the target intelligent agent group of each alarm information;

[0014] The fault repair unit is used to perform fault repair processing on the network element corresponding to each alarm information based on the target intelligent agent of each alarm information.

[0015] According to one aspect of this disclosure, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of any of the methods described above.

[0016] According to one aspect of this disclosure, a computer-readable storage medium is provided that stores a computer program / instructions thereon, which, when executed by a processor, implements the steps of any of the methods described above.

[0017] According to one aspect of this disclosure, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps of any of the methods described above.

[0018] As will be described in detail below, the network element fault repair method according to embodiments of this disclosure groups multiple intelligent agents into multiple intelligent agent groups. Each intelligent agent group is associated with a corresponding alarm category, thereby classifying multiple alarm information into the corresponding target intelligent agent group, achieving intelligent classification. Different levels of alarm information correspond to different processing measures, solving the problem of low efficiency in manual processing caused by complex alarm information, and achieving the goal of effectively improving operation and maintenance efficiency and reducing costs. It should be understood that both the foregoing general description and the following detailed description are exemplary and are intended to provide further illustration of the claimed technology. Attached Figure Description

[0019] The above and other objects, features, and advantages of this disclosure will become more apparent from the more detailed description of the embodiments thereof in conjunction with the accompanying drawings. The drawings are provided to further illustrate the embodiments of this disclosure and form part of the specification. They are used together with the embodiments of this disclosure to explain the disclosure and do not constitute a limitation thereof. In the drawings, the same reference numerals generally represent the same components or steps.

[0020] Figure 1 This is an example diagram illustrating a network element fault repair system according to an embodiment of this disclosure;

[0021] Figure 2 This is a flowchart illustrating a network element fault repair method according to an embodiment of the present disclosure;

[0022] Figure 3 This is an example diagram illustrating the interaction of an intelligent agent according to an embodiment of the present disclosure;

[0023] Figure 4 This is an example diagram illustrating a topology of multiple intelligent agents according to an embodiment of the present disclosure;

[0024] Figure 5 This is a flowchart illustrating another network element fault repair method according to an embodiment of the present disclosure;

[0025] Figure 6 This is an example diagram illustrating the interaction between two nodes according to an embodiment of the present disclosure;

[0026] Figure 7 This is a signaling diagram illustrating the interaction between two intelligent agents according to an embodiment of the present disclosure;

[0027] Figure 8 This is an illustration of an example of agent interaction between different groups according to an embodiment of the present disclosure;

[0028] Figure 9 This is an illustration of an example of user-smart agent interaction according to an embodiment of the present disclosure;

[0029] Figure 10 This is an illustration of another example of user-smart agent interaction according to an embodiment of the present disclosure;

[0030] Figure 11 This is an illustration of an application example of an intelligent agent used for repair according to an embodiment of the present disclosure;

[0031] Figure 12 This is an example diagram illustrating an application of a network element fault repair method according to an embodiment of the present disclosure;

[0032] Figure 13 This is a block diagram illustrating a network element fault repair device according to an embodiment of the present disclosure;

[0033] Figure 14 This is a hardware block diagram illustrating an electronic device according to an embodiment of the present disclosure. Detailed Implementation

[0034] To make the objectives, technical solutions, and advantages of this disclosure more apparent, exemplary embodiments according to this disclosure will now be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this disclosure, and not all embodiments of this disclosure. It should be understood that this disclosure is not limited to the exemplary embodiments described herein.

[0035] The technical solution disclosed herein can be applied to scenarios of automatic repair of network elements such as base stations. By using intelligent agents to automatically process alarm information, the network elements can be automatically repaired, reducing manual intervention in the network element repair process and improving the efficiency of network element repair.

[0036] In related technologies, network elements, such as base stations, are becoming increasingly complex and prone to operational failures. Generally, network element fault repair is performed manually. That is, a person makes the repair instructions and issues them to the faulty network element to achieve fault repair. However, manual repair is inefficient and cannot solve a large number of fault repair problems, leading to untimely and inaccurate network element fault repair.

[0037] To address the aforementioned issues, this disclosure employs intelligent agents in the automatic repair of network elements. In practical applications, due to the numerous and complex types of alarm information, a small number of intelligent agents results in low task recognition rates. Furthermore, network element fault repair is complex, and a single intelligent agent cannot complete the repair process. Therefore, this disclosure designs multiple intelligent agents and groups them based on different alarm types, constructing multiple intelligent agent groups. Based on these groups, a target intelligent agent group is determined for each alarm information. Then, based on the target intelligent agent groups, a target intelligent agent is determined to process each alarm information, thereby using the target intelligent agent to repair the corresponding network element. Therefore, the network element repair process can be completed without human intervention, reducing repair costs while improving repair efficiency.

[0038] Figure 1 This is an example diagram of a network element fault repair system provided in an embodiment of this disclosure. The network element fault repair system may include: multiple network elements 10 and a server 20 connected to each network element 10 via a network. Any network element 10 can be connected to a terminal, such as user equipment (UE), mobile station (MS), mobile terminal (MT), access terminal, user unit, user station, mobile station, mobile station, remote station, remote terminal, mobile device, user terminal, terminal, wireless communication equipment, user agent, or user device, etc. This embodiment does not impose excessive limitations on this aspect.

[0039] When any network element 10 malfunctions, it can report the fault to server 20. Server 20 can collect multiple alarm messages reported by multiple network elements 10. Server 20 can also communicate with electronic device 30. Server 20 can send multiple alarm messages to electronic device 30 as work order push messages. Electronic device 30 is configured with the network element fault repair method provided in this disclosure. After receiving the work order push message sent by server 20, it executes the network element fault repair method of this disclosure to automatically repair the network element corresponding to each alarm message, thereby improving fault repair efficiency.

[0040] Figure 2This is a flowchart illustrating a network element fault repair method provided in this embodiment. The method can be configured as a network element fault repair device, which can be located within a network element (e.g., a cloud server, a backend server). The network element fault repair method may include the following steps:

[0041] S201. Receive work order push message. The work order push message may include at least one alarm message. The alarm message refers to the information sent when the corresponding network element has a fault.

[0042] In one possible design, each network element can send alarm information to electronic devices in the form of work order push messages. The electronic devices can receive the work order push messages sent by each network element and read the alarm information from the work order push messages.

[0043] In another possible design, each network element can send alarm information to a server. The server can receive multiple alarm messages from multiple network elements and send these multiple alarm messages to electronic devices in the form of work order push messages. Accordingly, electronic devices can receive work order push messages and read multiple alarm messages from them.

[0044] The network element can be a device used for communication with the terminal. For example, it can be a base station, a TRP (Transmit / Receive Point) or a multi-TRP (Multiple Transmission and Reception Point), an evolved NodeB (eNB or eNodeB) in a communication system, a home base station (e.g., home evolved NodeB or home Node B, HNB), a base band unit (BBU), a wireless controller in a cloud radio access network (CRAN) scenario, or a relay station, access point, vehicle-mounted equipment, wearable devices, and network equipment in 5G networks or future evolved PLMN networks. It can be an access point (AP) in a WLAN, a gNB (the next generation Node B) in a new radio (NR) system, a satellite base station in a satellite communication system, or various forms of equipment that perform base station functions. The embodiments of this application are not limited to these.

[0045] It is understood that a work order push message can refer to a message that carries at least one alarm message (i.e., the alarm message body) or at least one alarm message access information (e.g., an access link / address for at least one alarm message). Work order push messages can take the form of a message body (MSG), a message queue, a message pop-up, or a message webpage, etc.

[0046] Of course, the work order push message is merely a carrier of at least one alarm message or related information of at least one alarm message. It is only used to distinguish the message by name and does not have any restrictive meaning. This application does not impose too many restrictions on the name of this message.

[0047] Optionally, alarm information can refer to fault information reported by a network element to relevant equipment (such as a server or electronic device) when a fault occurs. Alarm information can carry fault details, network element information, etc. The fault details can indicate the specific fault occurring in the network element. The network element information can indicate the network element that has experienced the fault. Network element information can, for example, refer to the network element's identifier, type, name, location, etc. The faulty network element can be identified through the network element information.

[0048] S202. Select the target intelligent agent group for each alarm message from a pre-built group of intelligent agents. The intelligent agent group includes at least one intelligent agent.

[0049] Optionally, an agent is an AI (Artificial Intelligence) system trained around a large language model (LLM). It can be used to handle corresponding tasks and possesses autonomy, the ability to handle complex tasks, and the ability to use tools. The agent integrates modules such as LLM, planning, memory, and tool use, enabling it to autonomously complete complex goals set by the user without continuous human intervention.

[0050] For example, an intelligent agent can be formally defined as: A = (L, R, S), where L is the large model used by the agent and its hyperparameters; R represents the role of the agent, defining its responsibilities and goals, guiding it to perform tasks and interactions; and S represents the agent's state, including its environmental awareness, long-term and short-term memory, reasoning ability, and ability to use external tools. The agent's state can be continuously updated based on the tasks performed or interactions with other agents or the environment.

[0051] For ease of understanding, Figure 3 An example diagram of agent interaction is shown.

[0052] The intelligent agent 300 may internally include a memory retrieval module 301, a reasoning and planning module 302, a tool invocation module 303, an observation and perception module 304, and an action execution module 305. Specifically, the memory retrieval module 301 can be used for retrieval of long-term and / or short-term memory. The reasoning and planning module 302 may include functions such as large-scale modeling, summarizing learning, and reflective decision-making. The tool invocation module 303 can be used to invoke preset tools, such as API tools (Application Programming Interface tools) and hardware tools. The observation and perception module 304 can be used for visual, auditory, and textual perception operations.

[0053] The intelligent agent 300 can interact with the user, the environment, or other intelligent agents. For example, the user can interact with the inference and planning module 302 by giving instructions, and the action execution module 305 and the observation and perception module 304 can interact with the environment.

[0054] This disclosure provides a system composed of multiple agents. A multi-agent system is a cooperative network of multiple agents, each with different responsibilities and expertise, which cooperate to accomplish more complex tasks. Multi-agent systems are characterized by division of labor and autonomous decision-making, improving overall efficiency and robustness, and have wide applications in reinforcement learning, robot control, and other fields.

[0055] Optionally, multiple alarm messages can belong to multiple alarm types. Related technologies do not focus on the type of alarm messages, but the number of alarm messages is quite large. If the alarm messages are not sorted, a large number of remediation decisions and corresponding quality issues would need to be processed manually in a short period of time. In this disclosure, multiple intelligent agents are grouped to obtain multiple intelligent agent groups.

[0056] Optionally, before S202, the process may further include: identifying multiple groups of intelligent agents that have been grouped. Different intelligent agent groups are defined for different alarm categories. For example, these may include: a power and environment intelligent agent group, a device and transmission intelligent agent group, a base station software intelligent agent group, etc.

[0057] Each agent group can include multiple agents. For example, a power and environment agent group could include a data center environment agent and an equipment monitoring agent. Each agent can have corresponding dialogue and tool invocation capabilities, but different agents can have different roles, capability configurations, and responsibilities. Each agent group uniformly schedules the activities of agents under the corresponding category, which may include dialogue recording, dialogue order arrangement, and dialogue end determination.

[0058] List 1 below shows example tables of several agent groups.

[0059]

[0060]

[0061] Table 1

[0062] S203. Using the target intelligent agent group of each alarm information, determine the target intelligent agent of each alarm information.

[0063] Optionally, S203 may include: determining the target intelligent agent related to the repair process of each alarm message from the intelligent agents included or associated with the target intelligent agent group of each alarm message.

[0064] The target intelligent agent can be an agent within a target intelligent agent group or an agent within another intelligent agent group that needs to cooperate with the target intelligent agent group. Specifically, the target intelligent agent can be used in the network element fault repair and processing of alarm information.

[0065] S204. Based on the target intelligent agent of each alarm information, perform fault repair and maintenance on the network element corresponding to each alarm information.

[0066] Optionally, S204 may include: generating a corresponding repair instruction based on the target intelligent agent of the alarm information, and performing fault repair processing on the corresponding network element through the repair instruction.

[0067] In the technical solution disclosed herein, a network element can receive work order push messages. These work order push messages can include multiple alarm messages sent by multiple network elements. The sending of work order push messages enables the push of alarm information from different network elements. For multiple alarm messages, a target intelligent agent group for each alarm message can be selected from multiple pre-built intelligent agent groups, thereby achieving the distribution of alarm information. Based on this, the corresponding network element can be repaired through the target intelligent agent within the target intelligent agent group corresponding to each alarm message, reducing manual intervention and improving fault repair efficiency.

[0068] As an example, S202, selecting the target intelligent agent group for each alarm information from multiple intelligent agent groups includes:

[0069] By identifying agents based on intent, the target agent group for each alarm message is selected from multiple agent groups.

[0070] Optionally, the intent-recognition agent can refer to an agent with intent recognition capabilities. The intent-recognition agent can identify the target agent group for each alarm message from multiple agent groups. Selecting the target agent group for each alarm message from multiple agent groups using the intent-recognition agent can include: inputting each alarm message into the intent-recognition agent to obtain the target agent group selected by the intent-recognition agent for each alarm message from multiple agent groups.

[0071] For example, the definition of an intent-recognition agent can be as follows:

[0072] Intent recognition agent = Agent(

[0073] llm = Qwen2-finetuned,

[0074] role = "Alarm Analysis Intent Recognition Agent",

[0075] backstory = "You are an alert analysis and intent recognition agent, skilled at performing intent recognition and analysis based on alert information".

[0076] tools = [],

[0077] memory = [], )

[0079] When defining each agent, you can use the `llm` parameter to define the large model used by the agent, the `role` parameter to define the agent's name and rules, the `backstory` parameter to define the agent's functions or functional descriptions, the `tools` parameter to define the tools the agent can use, and the `memory` parameter to define the memory modules the agent can use, such as long-term memory and short-term memory.

[0080] When a multi-agent collaborative network receives an alarm, the intent recognition agent determines which group of agents should handle the alarm first. The routing mechanism is implemented by constructing a base station wireless network alarm dataset and fine-tuning the large-scale agent model to ensure it can better identify the intent based on specific alarm information. Alternatively, intent recognition can also be achieved through techniques such as building a semantic classifier or few-shot learning on a large model.

[0081] In this embodiment of the disclosure, by using an intent recognition agent, the target agent group for each alarm message can be automatically selected from multiple agent groups, thereby achieving automated classification of multiple alarm messages, reducing manual intervention in alarm message classification, and improving classification efficiency.

[0082] Furthermore, based on any of the above embodiments, multiple intelligent agent groups are respectively associated with corresponding alarm categories. Through intent-based intelligent agents, the target intelligent agent group for each alarm information is selected from the multiple intelligent agent groups, including:

[0083] By using an intent-recognition intelligent agent, multiple alarm messages in the work order push message are classified to obtain the target alarm category corresponding to each alarm message.

[0084] The intelligent agent group associated with the target alarm category of each alarm message is determined as the target intelligent agent group of each alarm message.

[0085] Multiple agents can be divided into multiple agent groups so that when multiple alarm messages exist, each alarm message can be assigned to the corresponding target agent group for processing according to the function or task type of different agent groups.

[0086] For example, alarm categories can include: wireless alarm categories, transmission alarm categories, east-ring alarm categories, clock alarm categories, antenna and feeder alarm categories, base station software alarms, comprehensive / service alarms, and other major categories.

[0087] Optionally, the work order push message can be parsed to obtain multiple alarm messages. The intent recognition agent can classify each alarm message to obtain the target alarm category for each alarm message. Each agent group is pre-associated with alarm categories, so once the target alarm category of the alarm message is determined, the target agent group for the alarm message can be identified.

[0088] In this embodiment, multiple intelligent agents are grouped into multiple intelligent agent groups. Each intelligent agent group is associated with a corresponding alarm category, thereby classifying multiple alarm information into the corresponding target intelligent agent group to achieve intelligent classification. Different levels of alarm information correspond to different processing measures, which solves the problem of low efficiency of manual processing caused by complex alarm information, and achieves the purpose of effectively improving operation and maintenance efficiency and reducing costs.

[0089] As another embodiment, the target intelligent agent for each alarm message is determined by utilizing the target intelligent agent group for each alarm message, including:

[0090] Determine the topology corresponding to multiple agent groups. The topology refers to a directed acyclic graph formed with agents as nodes and the interaction relationships between agents and / or agent groups as edges, where any two nodes on any edge can interact.

[0091] By querying the topology of the intelligent agent group administrator in the target intelligent agent group corresponding to each alarm message, the specific target intelligent agent that handles each alarm message can be obtained.

[0092] Optionally, multiple topologies corresponding to groups of agents can be constructed, using each agent as a node and according to the connection relationships between the agents. The topology can be defined as G(N, E), where N = {n i |i∈I} represents the set of all agents, which can be represented by nodes in the topological structure, E={ <n i ,n j The relationships between agents and between agents and external tools are represented by edges in the graph structure. Using directed acyclic graphs (DAGs) as the topological structure for constructing multi-agent systems enables agent scalability and improves the utilization efficiency of agent groups.

[0093] For ease of understanding, Figure 4 An example diagram of the topology of multiple agents is shown. Figure 4 It includes four intelligent agent groups, such as the device and transmission intelligent agent group 401, the base station software intelligent agent group 402, the power and environment intelligent agent group 403, and other intelligent agent groups 404.

[0094] Each agent group contains an agent group administrator and multiple agents. In addition to the agent group administrator, each agent group may contain multiple agents. For example, the device and transmission agent group 401 includes transmission agents, antenna device agents, and other types of agents. The power and environment agent group 403 may include switching power supply agents, equipment room environment agents, and other types of agents. The base station software agent group 402 may include data configuration agents, license agents, and other types of agents. Other agent extraction groups 404 contain multiple types of agents.

[0095] The administrators of each group of intelligent agents can interact with each other, and the two ends of each edge represent two intelligent agents that can communicate with each other.

[0096] In this disclosed technical solution, a topology structure corresponding to multiple agent groups is established. The topology structure refers to a directed acyclic graph formed with agents as nodes and the interaction relationships between agents and / or agent groups as edges. Each node represents an agent that interacts with other nodes through edges. By establishing the topology structure, collaborative interaction between multiple agents is achieved, enabling unified scheduling of agent activities, thereby completing more complex workflows and ensuring the accuracy and efficiency of multi-agent collaborative task execution.

[0097] Furthermore, based on any of the above embodiments, by querying the topology structure through the intelligent agent group administrator in the target intelligent agent group corresponding to each alarm information, the specific target intelligent agent for processing each alarm information is obtained, including:

[0098] Each alarm message is input to the agent group administrator in the corresponding target agent group, triggering the agent group administrator to run a tree search algorithm to query the target agent in the topology that handles each alarm message.

[0099] Optionally, an agent group administrator can refer to an agent within an agent group that has an administrator role and is responsible for managing the agents within the group. The agent group administrator is responsible for managing the agents within the agent group.

[0100] Optionally, in the process of building intelligent agents, the intelligent agent group may include intelligent agents and intelligent agent group administrators, specifically, an administrator belonging to the intelligent agent group and the intelligent agents to which they belong may be built.

[0101] For example, the following are examples of several agents in the power and environment intelligent agent group:

[0102] Data center environment intelligent agent = Agent (

[0103] llm=Qwen2,

[0104] role = "Intelligent Agent for Data Center Environment",

[0105] backstory = "You are an intelligent agent for the data center environment, skilled in analyzing and resolving data center environment-related faults."

[0106] tools = [Air conditioner fault repair capability, monitoring unit fault repair capability, ...]

[0107] memory=[short_term_memory, long_term_memory], )

[0109] Switching power supply intelligent agent = Agent(

[0110] llm=Qwen2,

[0111] role = "Switching Power Supply Intelligent Agent",

[0112] backstory = "You are a switching power supply intelligent agent, skilled in analyzing and resolving switching power supply faults."

[0113] tools = [AC power fault repair capability, battery module fault repair capability, ...]

[0114] memory=[short_term_memory, long_term_memory], )

[0116] Group Chat Manager (Power and Environment Intelligent Agent Group Administrator)

[0117] groupchat = GroupChat(agents = [data center environment agent, switching power supply agent, ...], ...),

[0118] llm=Qwen2,

[0119] memory=[short_term_memory, long_term_memory], )

[0121] As mentioned above, each agent can be configured with a large model and has its own role and functions. In addition, each agent has the ability to invoke tools. Furthermore, regarding the long-term memory and short-term memory in the memory module, the long-term memory is the agent's local knowledge base, and the short-term memory is the agent's context dialogue cache record.

[0122] Optionally, the local knowledge base can be stored using a hierarchical aggregation structure, storing data separately for different users, agents, and scenarios. This allows for recursive retrieval of contextual records related to the dialogue, improving the accuracy of the agent's dialogue responses. Similarly, other agents and groups of agents can also be constructed using the same approach.

[0123] In this embodiment, each alarm message is input to the agent group association node in the target agent group. The agent group administrator runs a tree search algorithm to query the target agent in the topology that handles each alarm message. By establishing the topology, the query and application efficiency of multiple agents can be improved, and the query of the target agent for each alarm message can be completed quickly, thereby improving the efficiency of fault repair.

[0124] like Figure 5 The diagram shown is a flowchart of another network element fault repair method provided in this disclosure. This network element fault repair method may include the following steps:

[0125] S501. Receive work order push message. The work order push message includes at least one alarm message. The alarm message refers to the information sent when the corresponding network element has a fault.

[0126] S502. Select the target intelligent agent group for each alarm information from multiple pre-built intelligent agent groups. The intelligent agent group includes at least one intelligent agent.

[0127] S503. Utilize the target intelligent agent group of each alarm information to determine multiple target intelligent agents for each alarm information.

[0128] Among them, multiple target agents belong to one or more agent groups.

[0129] S504. For any alarm information, based on multiple target intelligent agents of the alarm information, determine the calling order among the multiple target intelligent agents of the alarm information.

[0130] S505. Based on the calling order among multiple target intelligent agents according to alarm information, multiple target intelligent agents are called in sequence to perform fault repair processing on the network element corresponding to the alarm information.

[0131] Optionally, S505 may include: determining the first target intelligent agent according to the calling order among multiple target intelligent agents of the alarm information; inputting the alarm information into the first target intelligent agent; sequentially calling multiple target intelligent agents starting from the first target intelligent agent; during the calling process, the output of the previous target intelligent agent becomes the input of the next target intelligent agent; until the last target intelligent agent is called, the network element fault repair processing of the alarm information is completed.

[0132] In this embodiment, after obtaining multiple target intelligent agents for each alarm information, the calling order among these target intelligent agents can be determined to complete the fault handling process. Then, using the calling order, the multiple target intelligent agents are sequentially invoked to perform fault repair processing on the network element corresponding to the alarm information. By executing the corresponding fault handling procedures within the fault handling orchestration process, the fault can be successfully eliminated, ensuring a high fault repair success rate.

[0133] As an example, the invocation steps of the target intelligent agent include:

[0134] The information analysis agent is invoked to analyze the alarm information and obtain the key information of the alarm information.

[0135] Determine the target tools that the target intelligent agent needs to invoke.

[0136] Input the key information of the alarm into the target tool to obtain the running results output by the target tool.

[0137] Based on the execution results, determine the processing result of the target agent. The processing result belongs to the input of the next target agent or belongs to the fault repair result.

[0138] Optionally, each agent can be associated with at least one tool, specifically by setting at least one tool in the tool parameters (e.g., named "tools") of each agent. Determining the target tool that the target agent needs to invoke may include: identifying the target tool among the at least one tool associated with the target agent that matches the key information.

[0139] Optionally, an information analysis agent can refer to an agent capable of analyzing input information. For example, an information analysis agent may include at least one of the following: a parameter extraction agent, a knowledge question answering agent, a database dialogue agent, or a text summarization agent.

[0140] For example, a parameter extraction agent can be defined as:

[0141] Includes parameter extraction agent = Agent(

[0142] llm=Qwen2,

[0143] role="parameter extraction agent",

[0144] backstory = "You are a parameter extraction agent, skilled at extracting key field information based on alarm information".

[0145] tools = [],

[0146] memory = [], )

[0148] Optionally, the information analysis agent can be a parameter extraction agent. Specifically, the parameter extraction agent can be invoked to extract parameters from the alarm information to obtain the key information of the alarm.

[0149] Each agent can invoke corresponding tools. When invoking a tool, the agent needs to first determine the tool's input parameters. Specifically, key information can be extracted from the parameters based on the content of the alarm message. For example, if the alarm message is "DDPU clock anomaly alarm," then the key information extracted is "DDPU clock."

[0150] Optionally, using key information from the alarm message as input data and inputting it into the target tool to obtain the target tool's output results can include: constructing the target parameters according to the target tool's parameters, converting the key information into target parameters, and inputting the target parameters into the target tool to obtain the target tool's output results.

[0151] Based on the parameter construction of the target tool, key information is converted into target parameters. For example, this may include adding related hardware and software system parameters as input information to the target tool, in addition to the extracted key information. The hardware and software system parameters and key information are then input into the target tool.

[0152] For example, if the key information is "BBU high temperature alarm", the equipment temperature information can be obtained through temperature sensors or a computer room temperature acquisition system, and the BBU high temperature alarm and equipment temperature information can be input into the target tool as target parameters.

[0153] Of course, one or more intelligent agents, such as knowledge-based question-answering agents, database-based dialogue agents, or text-based summary agents, can be applied to the process of obtaining key information or target parameters of alarm information. For details, please refer to the above content.

[0154] In this embodiment, during the fault repair process for network elements receiving alarm information, an information analysis agent is invoked to analyze the alarm information and obtain key information. Since the calling order among the multiple target agents for the alarm information is determined, the first target agent to be called can be identified. Therefore, after inputting the key information into the first target agent, the system can then sequentially call each target agent according to the calling order to perform fault repair processing on the network elements corresponding to the alarm information. This sequential execution of fault repair improves the success rate and efficiency of fault repair.

[0155] As another embodiment, multiple target intelligent agents are sequentially invoked to perform fault repair processing on the network elements corresponding to the alarm information, including:

[0156] In the process of sequentially calling multiple target agents, two target agent groups with call associations are divided into one agent group, resulting in at least one agent group;

[0157] If two target agents within any group of agents belong to the same group, then the instruction agent executes the instruction interaction between the two target agents within the group. The instruction agent is used to generate interaction instructions according to the input natural language sequence.

[0158] Alternatively, if two target agents within any agent group are not in the same group, then the instruction interaction between the two target agents is executed by the agent group administrator of the agent to which each target agent belongs.

[0159] Optionally, if two target agents within the same agent group belong to the same group, they can interact. Specifically, two target agents within the same group can interact through an instruction agent.

[0160] Taking a typical meshed directed acyclic graph as an example, the agent represented by each node interacts through the edges between nodes, and the connecting edge is handled by the corresponding instruction agent of that edge. Figure 6 The diagram illustrates an example of interaction between two nodes. For instance, the first agent 601 and the second agent 602 can interact via the instruction agent 603 corresponding to this edge.

[0161] For ease of understanding, Figure 7A signaling diagram illustrating the interaction between two agents is shown. The interaction steps between the two agents can be as follows:

[0162] 701. The first intelligent agent sends an instruction generation request to the instruction intelligent agent. 702. The instruction intelligent agent responds to the instruction generation request and generates a first instruction. 703. The first instruction is sent to the first intelligent agent, which optimizes the first instruction to obtain optimization information. 704. The optimization information is sent to the instruction intelligent agent. 705. The instruction intelligent agent regenerates a second instruction according to the optimization information of the first instruction. 706. The instruction intelligent agent sends the second instruction to the second intelligent agent. 707. The second intelligent agent executes the second instruction. 708. The second intelligent agent feeds back the execution result of the second instruction to the instruction intelligent agent.

[0163] Alternatively, each node and each edge can be represented using the following definition:

[0164]

[0165] Where μ(x) represents the agent operation instruction for x, α i To represent the node n i The intelligent agent, α ij To represent the edge n, assign it to the edge. ij The intelligent agent.

[0166] Optionally, if the two target agents within an agent group are in different groups, the interaction between the two agents needs to be executed by the agent group administrators corresponding to the two agent groups respectively.

[0167] like Figure 8 The diagram shown is an example of agent interaction between different groups provided in an embodiment of this disclosure. Taking two agents, a clock agent and an optical module agent, as an example, the clock agent includes a data query agent, a software configuration agent, an agent group administrator 1, and other types of agents. The optical module agent may include a clock loss agent, a laser agent, an agent group administrator 2, and other types of agents.

[0168] The data query agent determines that it needs to query the data of the clock loss agent. The data query agent needs to interact with agent group administrator 1, sending the requested clock loss information to agent group administrator 1. Agent group administrator 1 executes: 2. Determines that the optical module agent needs assistance. Therefore, agent group administrator 1 sends a query command to agent group administrator 2. 3. Responding to the query command, agent group administrator 2 determines that the clock loss agent needs to handle the issue. Agent group administrator 2 sends a query command to the clock loss agent, and, triggered by the query command, retrieves the clock loss information. This clock loss information is then sent to the data query agent via agent group administrator 1 and agent group administrator 2.

[0169] Optionally, when the repair process involves collaboration between agents, one agent group can send an assistance request or communication request to another agent group. The administrator of the requested agent group automatically selects an agent to handle the request. After the agent completes the task, it reports the result back to the administrator of the other agent group and further broadcasts the result to all agents within the group until the fault repair dialogue ends.

[0170] In this embodiment, during the sequential invocation of multiple target intelligent agents, two target intelligent agents with invocation associations can be grouped into an intelligent agent group, resulting in at least one intelligent agent group. Within each intelligent agent group, two target intelligent agents have invocation associations, thus refining the management of invocation associations between multiple intelligent agents related to alarm information. This allows for separate interaction between two target intelligent agents belonging to the same group or two target intelligent agents belonging to different groups, ensuring that two target intelligent agents within intelligent agent groups of different types or attributes can interact normally. Furthermore, it ensures normal interaction between multiple target intelligent agents related to alarm information, ultimately achieving normal repair of the network element corresponding to the alarm information.

[0171] As another embodiment, multiple target intelligent agents are sequentially invoked to perform fault repair processing on the network elements corresponding to the alarm information, including:

[0172] In the process of sequentially calling multiple target agents, the target agent that needs to interact with the user is determined.

[0173] The instruction agent executes the instruction interaction between the user and the target agent that needs to interact with the user.

[0174] Optionally, each intelligent agent can interact with the user. During the actual network element repair process, each intelligent agent can actively interact with the user or automatically interact when triggered by the user.

[0175] In the technical solution disclosed herein, in order to ensure the normal repair of network element faults, human participation is enabled in the fault repair process. Although the repair process requires user interaction with the target intelligent agent, human participation in the repair operation can still be reduced. The repair process can be completed through simple command interaction, ensuring the normal execution of network element repair while improving repair efficiency.

[0176] Furthermore, based on any of the above embodiments, the instruction interaction between the user and the target intelligent agent that needs to interact with the user is executed by the instruction intelligent agent, including:

[0177] The target intelligent agent that needs to interact with the user sends a first natural language sequence to the instruction intelligent agent, which then generates a first instruction corresponding to the first natural language sequence and sends the first instruction to the user terminal.

[0178] Alternatively, the system can detect the intervention message sent by the user, send the intervention message to the instruction agent, control the instruction agent to generate a second instruction corresponding to the intervention message, and send the second instruction to the target agent that needs to interact with the user.

[0179] Optionally, the intelligent agent can interact with the user through instructions. The following section combines... Figure 9 and Figure 10 An example diagram illustrating the interaction between a user and an intelligent agent.

[0180] Figure 9 This is an example diagram of interaction with an intelligent agent. (Example:) Figure 9 As shown, the process may include the following steps: S901, the third intelligent agent sends an instruction generation request to the instruction intelligent agent; S902, the instruction intelligent agent responds to the instruction generation request and generates a third instruction; S903, the first instruction is sent to the user terminal so that the user terminal performs the corresponding operation according to the first instruction and obtains the execution result; S904, the execution result is fed back to the instruction intelligent agent.

[0181] Figure 10 This is another example diagram of user-agent interaction. (Example:) Figure 10 As shown, the process may include the following steps: S1001, the user terminal sends an active intervention request to the command agent upon user triggering. S1002, the command agent responds to the active intervention request and generates a fourth instruction. S1003, the fourth instruction is sent to the fourth agent, which executes the fourth instruction and obtains the execution result. S1004, the fourth agent feeds back the execution result of the fourth instruction to the command agent.

[0182] Furthermore, user interaction with intelligent agents also occurs through user terminals. The user terminal can be defined as an intelligent agent, and the communication and interaction process between intelligent agents can be as follows:

[0183] IP(α i α ij α j )=(IP(α i α ij ), IP(α) ij α j ))

[0184] r←IP(α i α ij ), s←IP(α ij α i )↑,

[0185] p←IP(α ij α j ),q←IP(α i α ij )↑

[0186] Where IP(·) refers to the interactive operation between agents, and ↑ indicates that any agent performs iterative execution in the same request-response pattern to obtain the final execution result after the interaction is completed.

[0187] In this disclosed technical solution, human participation in the fault repair process of network elements is designed from two main aspects: one is that the target intelligent agent sends corresponding instructions to the human, who then responds; the other is that the human actively intervenes in the working process of the target intelligent agent. These two different participation methods enable human involvement in the fault repair process of network elements. This provides a more flexible network element repair solution and improves the success rate of network element repair.

[0188] As another embodiment, after performing fault repair processing on the network elements corresponding to each alarm message, the method further includes:

[0189] Query the repair status of the network element corresponding to each alarm message;

[0190] If the repair status of any network element is not repaired, then return to the target intelligent agent group selected from the pre-built intelligent agent groups to continue execution.

[0191] Optionally, after the network element repair is completed, the network element status can be queried. If the alarm no longer appears on the network element, it means that the fault self-repair is complete, and the process ends. If the alarm still appears, it is reported that the alarm in the agent group has not been cleared, and the decision needs to be reconsidered through the agent reflection process.

[0192] Furthermore, when the agent responsible for handling alarm information receives a message indicating a failure in the repair process, it re-executes the fault repair steps through the reflection mechanism in the tree search algorithm. If the processing step requires assistance from other types of agents, the processing information will be sent to the corresponding responsible agent group administrator.

[0193] In the technical solution disclosed herein, after fault repair processing of network elements, the repair status of the network elements can be queried. Furthermore, if the repair status of a network element is "not repaired," a new target intelligent agent group is selected for the alarm information, achieving dynamic feedback on the repair results. This further optimizes and improves the fault repair capability of network elements, continuously improving the repair behavior of network elements based on feedback from intelligent agent fault repair, thereby enhancing the repair efficiency and accuracy of the base station.

[0194] To facilitate understanding of the technical solutions disclosed herein Figure 11 This diagram illustrates an application example of an intelligent agent used for repair, as provided in this embodiment. Repairing a wireless base station fault involves multiple decision-making processes and various stakeholders, including network management, maintenance, manufacturers, and the configuration and replacement of various software and hardware. Taking a wireless base station clock alarm fault as an example, the fault repair employs clock alarm intelligent agent groups and optical module alarm intelligent agent groups. This intelligent agent repair method may include, for example, the following modules:

[0195] S1101, Clock alarm agent receives clock phase deviation abnormality alarm.

[0196] S1102, Clock alarm agent queries clock phase deviation.

[0197] S1103, Clock alarm intelligent agent interface transmission network management.

[0198] S1104, Software Configuration of the Clock Alarm Agent: The agent executes the following: Determine whether to transmit the carrier device clock and 1588 state lock. If yes, execute S1109; otherwise, execute S1105.

[0199] S1105. The software configuration of the clock alarm agent determines whether the configuration is missing or incorrect. If yes, proceed to S1106; otherwise, proceed to S1107.

[0200] S1106, Clock alarm intelligent agent transmission network management background processing.

[0201] S1107. The clock alarm agent queries the data to determine whether a large number of base station alarms occur at the same time. If yes, then execute S1108; otherwise, execute S1118.

[0202] S1108, the clock alarm intelligent agent performs troubleshooting on the SPN aggregation, backbone, and local OTN networks.

[0203] S1109. Is the clock alarm agent configured with the 1588PTP port? If not, execute step 1110; otherwise, execute step 1111.

[0204] S1110, clock alarm intelligent agent background processing.

[0205] S1111: Does the clock alarm intelligent agent transmit data on devices carrying multiple base stations? If not, proceed to step 1112; if yes, proceed to step 1113.

[0206] S1112. Check if the optical module alarm intelligent agent access port receives optical power normally. If yes, proceed to 1116; otherwise, proceed to 1117.

[0207] S1113. Does the clock alarm agent have a clock alarm at one of its base stations? If yes, execute 1114; otherwise, execute 1115.

[0208] S1114, Clock alarm intelligent agent resets main control, restarts BBU.

[0209] S1115. Asymmetric compensation is performed when the deviation values ​​are the same.

[0210] S1116, On-site processing of optical module alarm intelligent agent.

[0211] S1117, For optical module alarm intelligent agents, please contact the manufacturer for handling.

[0212] S1118. The clock alarm agent queries the data. The agent determines whether there is a hardware problem based on the phase deviation. If not, it executes 1119; if so, it executes 1120.

[0213] S1119, Clock alarm, SPN side fault, hardware repair or replacement required.

[0214] S1120, clock alarm, intelligent agent PTP hardware failure, perform asymmetric compensation.

[0215] certainly Figure 11 The agent applications shown are merely illustrative and do not constitute a specific limitation on agent group applications.

[0216] Figure 12 This diagram illustrates an application example of a network element fault repair method provided in this embodiment of the disclosure. The specific details of the network element fault repair process are as follows:

[0217] The electronic device receives a work order push message, then parses it to obtain at least one alarm message. The alarm message indicates a fault in the corresponding network element. The target agent group for each alarm message is determined through intent recognition. Each alarm message is distributed to the agent group administrator of the corresponding target agent group. The agent group administrator can coordinate with agents (Agent 1, Agent 2, up to Agent n) within the agent group to process the corresponding alarm messages. For example, Agent 2 can perform fault repair processing on the network element corresponding to the input alarm message. This includes generating repair instructions, which can be done collaboratively with multiple agents within other agent groups. After generating the repair instructions, Agent 2 can call a tool to execute the corresponding instructions sequentially. After the instructions are executed, the network element's status can be queried to obtain its operating status. If the network element has recovered, the repair process ends; otherwise, it returns to the collaborative processing of multiple agents within the group.

[0218] In fact, the process of generating instructions for the above-mentioned intelligent agent and / or the application process of the tools can be referred to the relevant content of the above embodiments, and will not be repeated here.

[0219] Figure 13 This is a schematic diagram of a network element fault repair device provided in an embodiment of the present disclosure. The network element fault repair device may include:

[0220] The message receiving unit 1301 is used to receive work order push messages. The work order push message includes at least one alarm message, which refers to the information sent when the corresponding network element has a fault.

[0221] Group selection unit 1302 is used to select the target intelligent agent group for each alarm information from a plurality of pre-built intelligent agent groups, wherein the intelligent agent group includes at least one intelligent agent.

[0222] The target determination unit 1303 is used to determine the target intelligent agent of each alarm information by utilizing the target intelligent agent group of each alarm information.

[0223] The fault repair unit 1304 is used to perform fault repair processing on the network element corresponding to each alarm information based on the target intelligent agent of each alarm information.

[0224] As one embodiment, the group selection unit 1302 may include:

[0225] The group selection module is used to select the target intelligent agent group for each alarm message from multiple pre-built intelligent agent groups by identifying the intelligent agent through intent.

[0226] As another embodiment, multiple intelligent agent groups are respectively associated with corresponding alarm categories. The group selection module includes:

[0227] The classification submodule is used to classify multiple alarm messages in the work order push message through an intent recognition intelligent agent, and obtain the target alarm category corresponding to each alarm message.

[0228] The determination submodule is used to determine the intelligent agent group associated with the target alarm category of each alarm message as the target intelligent agent group of each alarm message.

[0229] As another embodiment, the target determination unit may include:

[0230] The topology determination module is used to determine the topology structure corresponding to multiple agent groups. The topology structure refers to a directed acyclic graph formed with agents as nodes and the interaction relationships between agents and / or agent groups as edges, where any two nodes on any edge can interact.

[0231] The topology query module is used to query the topology structure by identifying the target intelligent agent group administrator in the target intelligent agent group corresponding to each alarm message, and to obtain the specific target intelligent agent that will handle each alarm message.

[0232] As another embodiment, the topology query module may include:

[0233] The search submodule is used to input each alarm information into the corresponding target intelligent agent group administrator, triggering the intelligent agent group administrator to run the tree search algorithm to query the target intelligent agent in the topology that handles each alarm information.

[0234] As one embodiment, the target intelligent agent includes multiple targets, and these multiple target intelligent agents belong to one or more intelligent agent groups. The fault repair unit 1304 may include:

[0235] The sequence determination module is used to determine the calling order among multiple target intelligent agents of any alarm information based on the alarm information.

[0236] The fault repair module is used to sequentially call multiple target intelligent agents based on the calling order of alarm information to perform fault repair processing on the network element corresponding to the alarm information.

[0237] As another embodiment, the fault repair module may include:

[0238] The key analysis submodule is used to call the information analysis intelligent agent to perform information analysis on alarm information and obtain key information of the alarm information;

[0239] The tool determination submodule is used to determine the target tool that the target intelligent agent needs to invoke;

[0240] The tool output submodule is used to input key information of alarm messages into the target tool and obtain the running results output by the target tool;

[0241] The result determination submodule is used to determine the processing result of the target agent based on the running results. The processing result belongs to the input of the next target agent or belongs to the fault repair result.

[0242] As another embodiment, the fault repair module may include:

[0243] The call analysis submodule is used to divide two target agents with call association into an agent group during the process of sequentially calling multiple target agents, so as to obtain at least one agent group;

[0244] The first interaction submodule is used to execute the instruction interaction between the two target agents in any agent group if the two target agents in the agent group are in the same group. The instruction agent is used to generate interaction instructions according to the input natural language sequence.

[0245] Alternatively, the second interaction submodule is used to execute instruction interaction between two target agents if two target agents within any agent group are not in the same group, through the agent group administrator of the agent to which the two target agents belong respectively.

[0246] As another embodiment, the fault repair module may include:

[0247] The interaction determination submodule is used to determine the target intelligent agent that needs to interact with the user during the process of sequentially calling multiple target intelligent agents.

[0248] The instruction interaction submodule is used to execute instruction interactions between the user and the target intelligent agent that needs to interact with the user through the instruction intelligent agent.

[0249] As yet another example, the instruction interaction submodule can specifically be used for:

[0250] The system controls the target agent that needs to interact with the user to send a first natural language sequence to the instruction agent, controls the instruction agent to generate a first instruction corresponding to the first natural language sequence, and sends the first instruction to the user terminal; or, it detects an intervention message sent by the user, sends the intervention message to the instruction agent, controls the instruction agent to generate a second instruction corresponding to the intervention message, and sends the second instruction to the target agent that needs to interact with the user.

[0251] As yet another embodiment, it also includes:

[0252] The status query unit is used to query the repair status of the network element corresponding to each alarm information;

[0253] The feedback execution unit is used to return to the target intelligent agent group from multiple pre-built intelligent agent groups to continue execution if the repair status of any network element is not repaired.

[0254] Figure 14 The diagram shows a schematic of an electronic device provided in an embodiment of the present disclosure. The electronic device may include a memory 1401, a processor 1402, and a computer program stored in the memory. The processor executes the computer program to implement any of the network element fault repair methods in the above embodiments.

[0255] For example, the processor can be used to perform early warning analysis on a user based on at least one user attribute data, to obtain the user's first credit score and first feature value, where each user attribute data includes the attribute data corresponding to the user in at least one user attribute; to obtain multiple early warning regions in a coordinate system, which are obtained by dividing the coordinate system based on early warning curves in the coordinate system, and each of the multiple early warning regions is associated with a corresponding early warning level, and the coordinate system is constructed with credit score and feature value as the coordinate axes; to map the first credit score and first feature value to the coordinate system to obtain a first coordinate point; to determine the early warning level associated with the early warning region of the first coordinate point in the coordinate system as the user's target early warning level; and to provide early warning prompts to the user based on the user's target early warning level.

[0256] Furthermore, embodiments of this application also provide a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements any of the network element fault repair methods described above.

[0257] Furthermore, this application also provides a computer program product, including a computer program that, when executed by a processor, implements any of the network element fault repair methods described above.

[0258] The basic principles of this disclosure have been described above with reference to specific embodiments. However, it should be noted that the advantages, benefits, and effects mentioned in this disclosure are merely examples and not limitations, and should not be considered as essential features of each embodiment of this disclosure. Furthermore, the specific details disclosed above are for illustrative and facilitative purposes only, and are not limitations. These details do not limit the scope of this disclosure to the necessity of employing the aforementioned specific details for implementation.

[0259] The block diagrams of devices, apparatuses, devices, and systems disclosed herein are merely illustrative examples and are not intended to require or imply that they must be connected, arranged, or configured in the manner shown in the block diagrams. As those skilled in the art will recognize, these devices, apparatuses, devices, and systems can be connected, arranged, and configured in any manner. Words such as “comprising,” “including,” “having,” etc., are open-ended terms meaning “including but not limited to,” and are used interchangeably with them. The terms “or” and “and” as used herein refer to the terms “and / or,” and are used interchangeably with them unless the context clearly indicates otherwise. The term “such as” as used herein refers to the phrase “such as but not limited to,” and is used interchangeably with it.

[0260] Additionally, as used herein, the “or” used in a list of items beginning with “at least one” indicates a separate list, such that a list of, for example, “at least one of A, B, or C” means A or B or C, or AB or AC or BC, or ABC (i.e., A and B and C). Furthermore, the word “exemplary” does not imply that the described example is preferred or better than other examples.

[0261] It should also be noted that in the systems and methods of this disclosure, the components or steps can be decomposed and / or recombined. These decompositions and / or recombinations should be considered as equivalent solutions to this disclosure.

[0262] Various changes, substitutions, and modifications can be made to the technology herein without departing from the teachings defined by the appended claims. Furthermore, the scope of the claims of this disclosure is not limited to the specific aspects of the processes, machines, manufactures, events, means, methods, and actions described above. Currently existing or later-developed processes, machines, manufactures, events, means, methods, or actions that perform substantially the same function or achieve substantially the same result as the corresponding aspects herein can be utilized. Therefore, the appended claims include such processes, machines, manufactures, events, means, methods, or actions within their scope.

[0263] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use this disclosure. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein may be applied to other aspects without departing from the scope of this disclosure. Therefore, this disclosure is not intended to be limited to the aspects shown herein, but rather to be carried out within the widest scope consistent with the principles and novel features disclosed herein.

[0264] The above description has been given for purposes of illustration and description. Furthermore, this description is not intended to limit the embodiments of this disclosure to the forms disclosed herein. Although numerous exemplary aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, alterations, additions, and sub-combinations therein.

Claims

1. A method for repairing network element faults, characterized in that, include: Receive a work order push message, the work order push message including at least one alarm message, the alarm message being the information sent when the corresponding network element has a fault; Select the target intelligent agent group for each alarm message from a pre-built group of intelligent agents, wherein the intelligent agent group includes at least one intelligent agent; By utilizing the target agent group of each alarm message, the target agent of each alarm message can be determined; Based on the target intelligent agent of each alarm message, fault repair processing is performed on the network element corresponding to each alarm message.

2. The method according to claim 1, characterized in that, The step of selecting the target intelligent agent group for each alarm information from multiple pre-built intelligent agent groups includes: By identifying agents based on intent, the target agent group for each alarm message is selected from multiple pre-built agent groups.

3. The method according to claim 2, characterized in that, The multiple agent groups are respectively associated with corresponding alarm categories. The step of selecting the target agent group for each alarm information from the pre-built multiple agent groups through intent recognition includes: The intent recognition agent classifies multiple alarm messages in the work order push message to obtain the target alarm categories corresponding to the multiple alarm messages. The intelligent agent group associated with the target alarm category of each alarm message is determined as the target intelligent agent group of each alarm message.

4. The method according to claim 1, characterized in that, The group of target intelligent agents utilizing each alarm information determines the target intelligent agent for each alarm information, including... Determine the topology corresponding to the multiple intelligent agent groups. The topology refers to a directed acyclic graph formed with intelligent agents as nodes and the interaction relationships between each intelligent agent and / or each intelligent agent group as edges, where two nodes on any side can interact. By querying the topology structure through the intelligent agent group administrator in the target intelligent agent group corresponding to each alarm message, the specific target intelligent agent for handling each alarm message can be obtained.

5. The method according to claim 4, characterized in that, The step of querying the topology structure by the administrator of the agent group in the target agent group corresponding to each alarm message to obtain the specific target agent for handling each alarm message includes: Each alarm message is input to the agent group administrator in the corresponding target agent group, triggering the agent group administrator to run a tree search algorithm to query the target agent in the topology that specifically handles each alarm message.

6. The method according to any one of claims 1-5, characterized in that, The target intelligent agents include multiple ones, and these multiple target intelligent agents belong to one or more intelligent agent groups. The step of performing fault repair processing on the network elements corresponding to each alarm information based on the target intelligent agents of each alarm information includes: For any alarm information, based on the multiple target intelligent agents of the alarm information, determine the calling order among the multiple target intelligent agents of the alarm information; Based on the calling order among the multiple target intelligent agents according to the alarm information, the multiple target intelligent agents are called sequentially to perform fault repair processing on the network element corresponding to the alarm information.

7. The method according to claim 6, characterized in that, The steps for invoking the target intelligent agent include: The information analysis agent is invoked to perform information analysis on the alarm information to obtain the key information of the alarm information; Determine the target tool that the target intelligent agent needs to invoke; The key information of the alarm is input into the target tool to obtain the running results output by the target tool; Based on the running results, the processing result of the target agent is determined, and the processing result belongs to the input of the next target agent or belongs to the fault repair result.

8. The method according to claim 6, characterized in that, The step of sequentially invoking the plurality of target intelligent agents to perform fault repair processing on the network elements corresponding to the alarm information includes: During the process of sequentially calling the multiple target agents, two target agents with a calling association are divided into an agent group to obtain at least one agent group; If two target agents within any group of agents are in the same group, then the instruction agent executes the instruction interaction between the two target agents within the group. The instruction agent is used to generate interaction instructions according to the input natural language sequence. Alternatively, if two target agents within any agent group are not in the same group, then the instruction interaction between the two target agents is executed by the agent group administrator of the agent to which each target agent belongs.

9. The method according to claim 6, characterized in that, The step of sequentially invoking the plurality of target intelligent agents to perform fault repair processing on the network elements corresponding to the alarm information includes: During the process of sequentially invoking the multiple target intelligent agents, the target intelligent agent that needs to interact with the user is determined; The instruction agent executes the instruction interaction between the user and the target agent that needs to interact with the user.

10. The method according to claim 9, characterized in that, The step of executing the instruction interaction between the user and the target agent that needs to interact with the user through the instruction agent includes: The target intelligent agent that needs to interact with the user is controlled to send a first natural language sequence to the instruction intelligent agent, and the instruction intelligent agent is controlled to generate a first instruction corresponding to the first natural language sequence and send the first instruction to the user terminal; Alternatively, the system detects the intervention message sent by the user, sends the intervention message to the instruction agent, controls the instruction agent to generate a second instruction corresponding to the intervention message, and sends the second instruction to the target agent that needs to interact with the user.

11. The method according to any one of claims 1-5, characterized in that, After performing fault repair processing on the network elements corresponding to each alarm message, the process also includes: Query the repair status of the network element corresponding to each alarm message; If the repair status of any network element is not repaired, then return to the target intelligent agent group selected from the pre-built intelligent agent groups to continue execution.

12. A network element fault repair device, characterized in that, include: The message receiving unit is used to receive work order push messages, which include at least one alarm message, and the alarm message refers to the information sent when the corresponding network element has a fault. A group selection unit is used to select the target intelligent agent group for each alarm information from a plurality of pre-built intelligent agent groups, wherein the intelligent agent group includes at least one intelligent agent; The target determination unit is used to determine the target intelligent agent for each alarm information by utilizing the target intelligent agent group of each alarm information; The fault repair unit is used to perform fault repair processing on the network element corresponding to each alarm information based on the target intelligent agent of each alarm information.

13. An electronic device comprising a memory, a processor, and a computer program stored in the memory, characterized in that, The processor executes the computer program to implement the steps of the method according to any one of claims 1-11.

14. A computer-readable storage medium having a computer program / instructions stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method described in claims 1-11.

15. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method described in claims 1-11.

Images