Operation and maintenance method and system for park private network, and electronic equipment

Abstract

The invention discloses an operation and maintenance method and system for a park private network and electronic equipment. The method comprises the following steps: receiving an operation and maintenance problem input by a user, and collecting network operation data of a park private network; performing semantic analysis on the operation and maintenance problem to generate a corresponding semantic vector; based on the semantic vector, retrieving a matched historical fault case from an operation and maintenance knowledge base to obtain a general processing scheme; meanwhile, based on the semantic vector and the network operation data, a Bayesian network is adopted to position a root cause node, the fault influence is evaluated in combination with a topology centrality algorithm, and a root cause positioning result is determined; according to the general processing scheme and the root cause positioning result, generating a target processing scheme matched with the operation and maintenance problem; according to the target disposal scheme, repairing a fault in the park private network, and recording a corresponding operation log and an execution result; and based on the operation log and the execution result, generating a structured operation and maintenance report to realize the operation and maintenance of the park private network. According to the method, the precision and intelligentization of the operation and maintenance of the park private network can be realized.

Description

Technical Field

[0001] This invention relates to the field of network technology, and in particular to a method, system, and electronic equipment for the operation and maintenance of a campus private network. Background Technology

[0002] As enterprises deepen their digital transformation, campus private networks, as critical information infrastructure carrying core businesses such as production, office, and security, are experiencing continuous expansion in network scale and significantly increased device heterogeneity, posing increasingly severe challenges to overall operation and maintenance. Currently, campus private network operation and maintenance still largely relies on traditional methods, heavily depending on the personal experience of maintenance personnel. Specifically, this involves manually logging into various devices and parsing command-line logs one by one to complete fault diagnosis and location. This method is not only inefficient and requires extremely high professional skills from maintenance personnel, but it also struggles to effectively handle sudden and complex fault scenarios.

[0003] In recent years, intelligent operation and maintenance (O&M) technologies have been gradually applied to this field, attempting to assist in fault handling through rule engines or simple keyword matching. However, existing O&M methods mostly rely on static experience and fail to be linked with real-time changes in network topology and performance indicators, resulting in a disconnect between the developed O&M plans and the dynamic state of the network. Summary of the Invention

[0004] The technical problem this invention aims to solve is to address the aforementioned shortcomings of existing technologies by proposing an operation and maintenance method, system, and electronic equipment for a campus private network. This method can dynamically generate operation and maintenance plans that precisely match the current network status, thereby effectively improving operation and maintenance efficiency and accuracy.

[0005] In a first aspect, the present invention provides a method for the operation and maintenance of a private network in a campus, the method comprising the following steps:

[0006] It receives operation and maintenance questions from users and collects network operation data of the park's private network;

[0007] Perform semantic parsing on operation and maintenance issues to generate corresponding semantic vectors;

[0008] Based on semantic vectors, matching historical fault cases are retrieved from the operation and maintenance knowledge base to obtain general handling solutions; at the same time, based on semantic vectors and network operation data, Bayesian network is used to locate the root cause node, and the impact of the fault is evaluated by combining the topology centrality algorithm to determine the root cause location result.

[0009] Based on the general handling plan and root cause identification results, generate a target handling plan that matches the operation and maintenance problem;

[0010] According to the target handling plan, repair the faults in the park's private network and record the corresponding operation logs and execution results;

[0011] Based on operation logs and execution results, a structured operation and maintenance report is generated to realize the operation and maintenance of the park's private network.

[0012] Furthermore, it receives user-inputted maintenance and operation questions, specifically including:

[0013] An embedded floating assistant is used to establish a user interaction channel to receive text input or voice input from the user.

[0014] The user interaction channel is implemented based on Kubernetes' multi-tenant containerized deployment method. When a user initiates an interaction through the user interaction channel, an independent session space with resource isolation is allocated to the user or the user's group, and the corresponding visual operation components are dynamically rendered according to the user's permissions.

[0015] Furthermore, semantic parsing is performed on the operation and maintenance issues to generate corresponding semantic vectors, specifically including:

[0016] When maintenance issues are addressed via voice input, a voice recognition model converts the voice input into text information.

[0017] The semantic vector model is used to vectorize textual operation and maintenance questions, generating corresponding semantic vectors.

[0018] Furthermore, based on semantic vectors, matching historical fault cases are retrieved from the operations and maintenance knowledge base to obtain general handling solutions, specifically including:

[0019] Calculate the similarity between the semantic vector and the vectors of each knowledge item in the operation and maintenance knowledge base;

[0020] Based on vector similarity, candidate knowledge items with similarity higher than a preset threshold are selected.

[0021] The candidate knowledge items are sorted according to their decay weights, and the processing scheme corresponding to the knowledge item with the highest weight is selected as the general processing scheme.

[0022] The operation and maintenance knowledge base is updated using an incremental decay weight algorithm. Newly added entries are assigned an initial weight, the weight increases when an entry is accessed, and the weight of all entries decays over time and according to a preset decay factor based on the number of times they are not accessed.

[0023] Furthermore, based on semantic vectors and network operation data, a Bayesian network is used to locate the root cause node, and the impact of the fault is evaluated by combining the topological centrality algorithm to determine the root cause localization result, which specifically includes:

[0024] The semantic vectors are matched with predefined fault modes in the fault analysis matrix to determine the initial fault type;

[0025] Based on the initial fault type, relevant performance indicators and log data are extracted from network operation data;

[0026] A Bayesian network model is constructed using performance metrics and log data as input. The posterior probability of each network node becoming a root cause node is calculated, and the node with the highest posterior probability is identified as the root cause node.

[0027] Based on the network topology, the topological centrality algorithm is applied to calculate the root cause nodes and identify key fault points.

[0028] Based on the network topology link information, and according to the root cause node, key fault point, and network operation data, the root cause localization result, including a heat map of the spread of fault impact, is generated.

[0029] Furthermore, based on the general handling plan and root cause analysis results, a target handling plan matching the operational issue is generated, specifically including:

[0030] The consistency between the general treatment plan and the root cause localization results is verified.

[0031] When the verification result is a match, the general handling solution will be used as the target handling solution;

[0032] When the verification result is a mismatch or there are multiple root causes, the general handling solution is tailored, supplemented or reordered based on the causal path in the knowledge graph and the device context information to generate a target handling solution that is adapted to the current fault scenario.

[0033] Furthermore, based on the operation logs and execution results, a structured operation and maintenance report is generated, specifically including:

[0034] Collect operation logs and execution results;

[0035] Based on a preset report template, operation logs and execution results are integrated through a drag-and-drop canvas editing interface to generate a structured preliminary operation and maintenance report;

[0036] The preliminary operation and maintenance report is dynamically converted into Word or PPT format documents through a multi-format output engine, and visualized charts of key data nodes are rendered to generate a structured operation and maintenance report.

[0037] Furthermore, after generating the structured operation and maintenance report, the method also includes:

[0038] Structured operation and maintenance reports are fed back to the operation and maintenance knowledge base for iterative updates.

[0039] Secondly, the present invention provides an operation and maintenance system for a campus private network, the system comprising:

[0040] The data interaction module is used to collect network operation data of the park's private network and also to receive operation and maintenance questions input by users.

[0041] The semantic parsing module, connected to the data interaction module, is used to perform semantic parsing on operation and maintenance issues and generate corresponding semantic vectors.

[0042] The knowledge retrieval module, connected to the semantic parsing module, is used to retrieve matching historical fault cases from the operation and maintenance knowledge base based on semantic vectors to obtain general handling solutions.

[0043] The root cause analysis module is connected to the data interaction module and the semantic parsing module respectively. It is used to locate the root cause node based on semantic vectors and network operation data, using Bayesian network, and to evaluate the impact of the fault in combination with the topological centrality algorithm to determine the root cause location result.

[0044] The solution generation module, connected to the root cause analysis module, is used to generate a target solution that matches the operation and maintenance problem based on the general solution and the root cause location results.

[0045] The solution execution module, connected to the solution generation module, is used to repair faults in the park's private network according to the target disposal solution, and record the corresponding operation logs and execution results.

[0046] The report generation module, connected to the solution execution module, is used to generate structured operation and maintenance reports based on operation logs and execution results.

[0047] Thirdly, the present invention provides an electronic device including a memory and a processor. The memory stores a computer program, and when the processor runs the computer program stored in the memory, the processor executes the operation and maintenance method of the campus private network according to the first aspect.

[0048] This invention, by constructing an intelligent operation and maintenance closed loop that integrates historical knowledge retrieval and real-time network status analysis, effectively overcomes the core shortcomings of traditional operation and maintenance methods, such as low efficiency, and the disconnect between fault case retrieval and real-time network analysis in existing intelligent operation and maintenance solutions. Specific beneficial effects are as follows:

[0049] (1) Accurate root cause localization of faults: By parallel execution of historical case retrieval based on semantic vectors and real-time data analysis based on Bayesian networks and topological centrality algorithms, this invention deeply integrates historical experience knowledge with the current real-time state of the network. This solves the problem of "insufficient accuracy in root cause localization" mentioned in the background technology, enabling the system to perform probabilistic reasoning by combining historical patterns and real-time evidence, thereby accurately locating the root cause node of the fault and assessing its impact range, significantly improving the accuracy and reliability of localization.

[0050] (2) A dynamic handling scheme highly matched with the real-time scenario was generated: By generating the final target handling scheme based on the general handling scheme and the root cause localization results, this invention ensures that the handling strategy not only draws on historical experience but also undergoes verification and optimization in the current network context. This directly solves the bottleneck of "the handling scheme being out of sync with the actual scenario" pointed out in the background technology, avoids the risk of ineffective handling or secondary failures caused by scheme mismatch, and greatly improves the effectiveness and security of the handling measures.

[0051] (3) Improve the overall efficiency and automation level of the operation and maintenance process: From receiving operation and maintenance issues in natural language form to automatically executing repair solutions and generating structured reports, this invention realizes end-to-end automation of the operation and maintenance process. This completely changes the situation of traditional operation and maintenance that relies heavily on manual troubleshooting and is inefficient, reduces the stringent requirements on the professional skills of operation and maintenance personnel, and can quickly and accurately respond to various sudden and complex fault scenarios.

[0052] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0053] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof. The above and other features and advantages will become more apparent to those skilled in the art from the detailed description of exemplary embodiments with reference to the accompanying drawings, in which:

[0054] Figure 1 A schematic diagram illustrating the operation and maintenance method of a private network in a campus provided in an embodiment of the present invention;

[0055] Figure 2 A schematic diagram of the operation and maintenance process of a private network in a campus provided in an embodiment of the present invention;

[0056] Figure 3 This is a schematic diagram illustrating the multi-user concurrency support capability provided in an embodiment of the present invention;

[0057] Figure 4 A flowchart of the data processing of the dynamic knowledge base engine provided in this embodiment of the invention;

[0058] Figure 5 This is a flowchart of the fault analysis matrix data processing provided in an embodiment of the present invention;

[0059] Figure 6 This is a schematic diagram of the operation and maintenance system for a private network in a campus provided in an embodiment of the present invention;

[0060] Figure 7 This is an architectural diagram of an electronic device provided in an embodiment of the present invention.

[0061] Figure reference numerals: 10, data interaction module; 20, semantic parsing module; 30, knowledge retrieval module; 40, root cause analysis module; 50, solution generation module; 60, solution execution module; 70, report generation module; 100, processor; 200, memory. Detailed Implementation

[0062] It is understood that the specific embodiments and accompanying drawings described herein are merely for explaining the invention and are not intended to limit the invention.

[0063] It is understood that, without conflict, the various embodiments and features in the embodiments of the present invention can be combined with each other.

[0064] It is understood that, for ease of description, only the parts related to the present invention are shown in the accompanying drawings, while the parts unrelated to the present invention are not shown in the drawings.

[0065] It is understood that each unit or module involved in the embodiments of the present invention may correspond to only one entity structure, or may be composed of multiple entity structures, or multiple units or modules may be integrated into one entity structure.

[0066] It is understood that, without conflict, the functions and steps marked in the flowcharts and block diagrams of this invention may occur in a different order than that marked in the accompanying drawings.

[0067] It is understood that the flowcharts and block diagrams of this invention illustrate the possible architecture, functions, and operations of systems, apparatuses, devices, and methods according to various embodiments of this invention. Each block in the flowchart or block diagram may represent a unit, module, program segment, or code, containing executable instructions for implementing the specified function. Furthermore, each block or combination of blocks in the block diagram and flowchart can be implemented using a hardware-based system to achieve the specified function, or using a combination of hardware and computer instructions.

[0068] It is understood that the units and modules involved in the embodiments of the present invention can be implemented by software or by hardware. For example, the units and modules can be located in a processor.

[0069] Example 1:

[0070] Given the current situation where private network operation and maintenance in industrial parks heavily relies on manual experience, resulting in low fault location efficiency and a lack of scenario adaptability in handling solutions, this embodiment provides an intelligent operation and maintenance method that integrates semantic understanding, knowledge retrieval, and causal reasoning. This method receives operation and maintenance questions in natural language from users and generates semantic vectors. On one hand, it retrieves matching historical cases from the operation and maintenance knowledge base to obtain general handling solutions. On the other hand, it combines real-time network operation data, using Bayesian networks and topology centrality algorithms to jointly locate the root cause and assess its impact, thereby generating a precisely adapted target handling solution. After performing the repair, it automatically records operation logs and results, generating a structured operation and maintenance report, forming a closed-loop feedback. This embodiment is particularly suitable for private network scenarios with high network reliability requirements and high operation and maintenance complexity, such as enterprise parks, universities, and industrial parks, effectively improving fault response speed, diagnostic accuracy, and knowledge reuse capabilities. Figure 1 As shown in the figure, this embodiment provides a method for the operation and maintenance of a campus private network, which includes the following steps S1 to S6.

[0071] Step S1: Receive operation and maintenance questions input by users and collect network operation data of the park's private network.

[0072] As a specific implementation method, receiving user-inputted maintenance issues includes:

[0073] An embedded floating assistant is used to establish a user interaction channel to receive text input or voice input from the user.

[0074] The user interaction channel is implemented based on Kubernetes' multi-tenant containerized deployment method. When a user initiates an interaction through the user interaction channel, an independent session space with resource isolation is allocated to the user or the user's group, and the corresponding visual operation components are dynamically rendered according to the user's permissions.

[0075] Step S2: Perform semantic parsing on the operation and maintenance issues to generate corresponding semantic vectors.

[0076] As a specific implementation method, semantic parsing is performed on operation and maintenance issues to generate corresponding semantic vectors, specifically including:

[0077] When maintenance issues are addressed via voice input, a voice recognition model converts the voice input into text information.

[0078] The semantic vector model is used to vectorize textual operation and maintenance questions, generating corresponding semantic vectors.

[0079] Step S3: Based on semantic vectors, retrieve matching historical fault cases from the operation and maintenance knowledge base to obtain general handling solutions; at the same time, based on semantic vectors and network operation data, use Bayesian network to locate the root cause node, and combine the topology centrality algorithm to evaluate the impact of the fault and determine the root cause location result.

[0080] As a specific implementation method, based on semantic vectors, matching historical fault cases are retrieved from the operation and maintenance knowledge base to obtain general handling solutions, specifically including:

[0081] Calculate the similarity between the semantic vector and the vectors of each knowledge item in the operation and maintenance knowledge base;

[0082] Based on vector similarity, candidate knowledge items with similarity higher than a preset threshold are selected.

[0083] The candidate knowledge items are sorted according to their decay weights, and the processing scheme corresponding to the knowledge item with the highest weight is selected as the general processing scheme.

[0084] The operation and maintenance knowledge base is updated using an incremental decay weight algorithm. Newly added entries are assigned an initial weight, the weight increases when an entry is accessed, and the weight of all entries decays over time and according to a preset decay factor based on the number of times they are not accessed.

[0085] As a specific implementation method, based on semantic vectors and network operation data, a Bayesian network is used to locate the root cause node, and the impact of the fault is evaluated by combining the topological centrality algorithm to determine the root cause location result, specifically including:

[0086] The semantic vectors are matched with predefined fault modes in the fault analysis matrix to determine the initial fault type;

[0087] Based on the initial fault type, relevant performance indicators and log data are extracted from network operation data;

[0088] A Bayesian network model is constructed using performance metrics and log data as input. The posterior probability of each network node becoming a root cause node is calculated, and the node with the highest posterior probability is identified as the root cause node.

[0089] Based on the network topology, the topological centrality algorithm is applied to calculate the root cause nodes and identify key fault points.

[0090] Based on the network topology link information, and according to the root cause node, key fault point, and network operation data, the root cause localization result, including a heat map of the spread of fault impact, is generated.

[0091] Step S4: Based on the general handling plan and root cause identification results, generate a target handling plan that matches the operation and maintenance problem.

[0092] As a specific implementation method, based on the general handling plan and root cause localization results, a target handling plan matching the operational problem is generated, specifically including:

[0093] The consistency between the general treatment plan and the root cause localization results is verified.

[0094] When the verification result is a match, the general handling solution will be used as the target handling solution;

[0095] When the verification result is a mismatch or there are multiple root causes, the general handling solution is tailored, supplemented or reordered based on the causal path in the knowledge graph and the device context information to generate a target handling solution that is adapted to the current fault scenario.

[0096] Step S5: Repair the faults in the park's private network according to the target handling plan, and record the corresponding operation logs and execution results.

[0097] Step S6: Based on the operation logs and execution results, generate a structured operation and maintenance report to realize the operation and maintenance of the park's private network.

[0098] As a specific implementation method, a structured operation and maintenance report is generated based on the operation logs and execution results, specifically including:

[0099] Collect operation logs and execution results;

[0100] Based on a preset report template, operation logs and execution results are integrated through a drag-and-drop canvas editing interface to generate a structured preliminary operation and maintenance report;

[0101] The preliminary operation and maintenance report is dynamically converted into Word or PPT format documents through a multi-format output engine, and visualized charts of key data nodes are rendered to generate a structured operation and maintenance report.

[0102] As a specific implementation method, after generating the structured operation and maintenance report, the method further includes:

[0103] Structured operation and maintenance reports are fed back to the operation and maintenance knowledge base for iterative updates.

[0104] The implementation process of this embodiment is shown in Figure 2. It adopts a three-layer intelligent architecture consisting of an intelligent interaction layer, a cognitive decision layer, and an execution output layer to realize closed-loop operation and maintenance management throughout the entire lifecycle from problem input to report output. The intelligent interaction layer is responsible for multimodal interaction with users, the cognitive decision layer is responsible for in-depth analysis and intelligent decision-making of input information, and the execution output layer is responsible for executing relevant decisions and generating the final visual report.

[0105] Specifically, at the intelligent interaction layer, the system provides dual-channel input of text and voice through an embedded floating assistant. Voice signals are converted into text information using an end-to-end speech recognition model, and all text requests are uniformly parsed using semantic vector models such as BERT to generate semantic vector information. Simultaneously, based on Kubernetes' multi-tenant containerization engine, it allocates resource-isolated independent session spaces for different users or groups and dynamically renders visual operation components according to user permissions, ensuring data security and personalized interactive experiences when multiple users use the system concurrently. At the cognitive decision-making layer, the system works collaboratively through three main modules: a dynamic knowledge engine, a fault analysis matrix, and a solution reasoning engine. The dynamic knowledge engine receives semantic vector information, periodically crawls equipment vendor operation and maintenance manuals in batches, and collects equipment logs and topology logs. It uses large models such as bge-ranker to refine and integrate the data. Based on semantic vector similarity, it retrieves historical fault cases through multi-hop inference chains to obtain general handling solutions, and uses an incremental decay weight algorithm (such as W=e^{-0.05t}) to achieve the autonomous evolution of the knowledge base. The fault analysis matrix takes semantic vectors, fault handling requests, and real-time alarm streams as inputs. It locates the root cause node through Bayesian network modeling, identifies key fault points by combining topology centrality algorithms (such as calculating node betweenness centrality), and integrates KPI mutation data and topology link information to generate a fault impact diffusion heatmap. The solution inference engine performs consistency verification between the general handling solution and the root cause location results. Based on the causal path of the knowledge graph and equipment context information, it trims, supplements, or reorders the solution to generate a final handling solution adapted to the current scenario. It then converts the solution into a pre-set script instruction or work order information containing alarms through complexity evaluation. At the output layer, the system employs an intelligent execution system for tiered handling. For simple faults, it automatically executes pre-set scripts; for complex faults, it triggers SMS alarms and assigns work orders, while simultaneously monitoring execution results in real time. The report generation hub captures full-timeline data on the fault (including fault symptoms, location and source tracing, handling methods, and repair verification). It allows for customization of Word / PPT templates through a drag-and-drop visual canvas editing interface, dynamically converts documents using a multi-format output engine, and renders topology diagrams and indicator charts. Finally, the handling results and visual reports are pushed to the front-end floating assistant in real time to complete the closed-loop feedback. This embodiment achieves autonomous evolution of the knowledge system through a dynamic knowledge engine, precise probabilistic location through a fault analysis matrix, and continuous optimization through a fully closed-loop iterative learning mechanism, ultimately significantly improving operational efficiency, accuracy, and automation levels.

[0106] like Figure 3As shown, this embodiment constructs a multi-tenant containerized session engine based on a Kubernetes-based containerized deployment architecture to achieve resource isolation and session management in scenarios with concurrent access by multiple users. After user requests enter the system through a unified request entry point, the engine intelligently routes requests from different users (user A, user B...user N) to the corresponding independent tenant container groups (container group A, container group B...container group N) through a concurrent request distribution module. Complete resource isolation is achieved between each container group, ensuring the security and independence of user session data. This engine completes the containerized deployment of applications based on Kubernetes cloud-native technology. It dynamically allocates and manages the session resources of each user through the k8s cloud resource orchestration mechanism, achieving isolated allocation of computing resources such as CPU, memory, and storage, and supporting elastic scaling of resource quotas based on user load. At the same time, it provides a unified administrator interface for operation and maintenance account management, supports one-click creation of new campus operation and maintenance user accounts, allows flexible configuration of the session cache quantity for a single user, and enables fine-grained user permission allocation and management. The engine automatically renders the corresponding visual operation components dynamically according to the user's operable permission level, achieving permission-based component access control and ensuring that users can only operate authorized functional modules. Furthermore, the system automatically generates corresponding operation interfaces based on user permission configurations, supporting differentiated interface displays for different types of users such as operation and maintenance administrators, ordinary operation and maintenance personnel, and visitors. It implements component-level functional permission control to ensure accurate matching of interface elements with user permissions and provides a one-click call function to simplify user operation processes. Through the above mechanisms, this embodiment achieves resource isolation, permission control, and session management when multiple users share the same operation and maintenance system, ensuring both efficient utilization of system resources and the security of user data and consistency of user experience.

[0107] In this embodiment, the operation and maintenance knowledge base is implemented through a dynamic knowledge base engine (e.g., Figure 4As shown), the engine's data processing flow forms a closed-loop process from multi-source data access, quality control, knowledge integration, intelligent reasoning to continuous optimization. It covers key stages such as multi-source data input, refined knowledge organization, dynamic quality detection, multi-modal knowledge base construction, multi-hop reasoning and intelligent operation and maintenance applications, feedback learning and optimization, specifically implemented as follows: In the multi-source data input stage, the system receives two types of data input: one is batch data, including various terminal device operation and maintenance manuals regularly crawled from mainstream manufacturers, as well as periodically collected device logs and cluster topology logs; the other is real-time data streams, i.e., real-time vectorized information from the intelligent interaction layer, which serves as one of the inputs for dynamic knowledge processing; the refined knowledge organization stage uses bge-ranker. The large model performs semantic vectorization on the aforementioned multi-source input data, achieving refined knowledge organization and integration, and providing a structured representation for subsequent processing. The dynamic quality inspection stage verifies the quality of the organized knowledge, mainly including conflict verification (identifying logical or factual conflicts between knowledge from different sources) and timeliness verification (ensuring the knowledge used meets time validity requirements). If verification fails, it enters the decay weight injection stage, where old knowledge is downgraded according to the time decay factor W=e^(-0.05t) and submitted to the manual review stage for further examination. The multimodal knowledge base construction stage integrates the quality-tested knowledge to build a multimodal knowledge base containing a Milvus vector library and a Neo4j graph. The Milvus vector library is used to store the semantically vectorized knowledge representation, and Neo4j... The knowledge graph organizes knowledge in a graph structure, supporting complex relationship retrieval and reasoning. The multi-hop reasoning engine, based on the constructed knowledge base, executes a multi-hop reasoning chain from fault symptoms to root causes and then to solutions. The system retrieves recent historical fault cases with high similarity and generates general fault handling solutions, which are then output to the solution reasoning engine. In the intelligent operation and maintenance application phase, the reasoning results are applied to actual operation and maintenance scenarios, mainly including generating detailed fault diagnosis analysis (diagnosis reports) and providing specific operational suggestions or solutions (handling solutions). In the feedback learning and optimization phase, users rate the system's output diagnosis reports and handling solutions. The feedback information is used to adjust knowledge weights and reasoning strategies based on the feedback, and also to incorporate validated new fault cases and their handling solutions into the knowledge base, continuously enriching the system's knowledge system. Through this complete process, an intelligent operation and maintenance knowledge system with self-learning and dynamic update capabilities is constructed.

[0108] Figure 5The diagram illustrates the data processing flow of the fault analysis matrix. Starting with multi-source data input, this flow, through Bayesian network modeling, topological centrality analysis, KPI mutation fusion and heatmap generation, fault diagnosis report output, solution inference engine, and fault handling, achieves accurate fault category judgment and impact range analysis, forming a closed-loop processing from fault input and analysis location to solution generation and execution. Specifically: In the multi-source data input stage, the system receives various input data, including fault handling requests initiated by users or the system (fault handling requests and maintenance issues), real-time alarm information collected periodically by the system (real-time alarm stream), and vectorized information generated from the intelligent interaction layer as one of the inputs for analysis; the Bayesian network modeling stage utilizes Bayesian networks to... The alarm propagation path is modeled to construct a propagation path matrix, thereby achieving root cause node localization (identifying the root cause node in the fault propagation path). For nodes with low confidence, they are output as low-confidence nodes and corrected data processing is performed to finally determine the high-probability root cause node, which serves as the basis for subsequent analysis. The topology centrality analysis step uses graph theory centrality algorithms (such as betweenness centrality) to analyze the network topology, identify key nodes and their associated topological links, detect abnormal topology interruptions, and locate key fault points. For uncertain parts of the analysis results, a manual review step is introduced for confirmation. KPIs The mutation fusion and heatmap generation stage integrates key node information with real-time collected KPI (Key Performance Indicator) mutation data. The system quickly generates an impact range heatmap, intuitively displaying the area and extent of the fault's impact. This heatmap, as an important component of the fault diagnosis report, helps maintenance personnel quickly grasp the scope of the fault's impact. The fault diagnosis report output stage synthesizes the analysis results from the above stages, generating a fault diagnosis report including root cause location, key fault points, and impact range, and outputting it to the solution inference engine. In the solution inference engine and fault handling process, the solution inference engine receives general fault handling solutions from the dynamic knowledge base engine, as well as key fault point information located by the fault analysis matrix. On the one hand, it combines the two types of input and output to accurately locate the root cause and the corresponding fault handling solution; on the other hand, it links with the intelligent execution system to automatically repair network faults according to the handling solution, specifically adopting a tiered handling strategy: for simple faults, a pre-set script is automatically executed to complete the repair; for complex faults, an SMS alarm is triggered and a work order is assigned for maintenance personnel to intervene. This process has efficient and accurate fault diagnosis and handling capabilities, effectively improving the level of maintenance automation and emergency response efficiency.

[0109] The operation and maintenance method for the campus private network provided in this embodiment achieves automated operation and maintenance throughout the entire lifecycle, from problem input to report output, by integrating natural language interaction, semantic understanding, Bayesian inference, topology analysis and closed-loop feedback mechanism. This method adopts a three-layer architecture: intelligent interaction layer, cognitive decision layer, and execution output layer. In the intelligent interaction layer, a floating assistant embedded in the network management interface supports dual-channel input of text and voice. Based on Kubernetes multi-tenant containerization deployment, it allocates resource-isolated independent session spaces for different users and dynamically renders visualization components according to permissions. In the cognitive decision layer, after semantically vectorizing user questions, it retrieves general handling solutions from historical cases in the dynamic knowledge base. On the other hand, it combines real-time network operation data, uses Bayesian networks to locate root cause nodes, topological centrality algorithms to identify key fault points, and integrates them to generate root cause location results including an impact heatmap. Then, through consistency verification and knowledge graph reasoning, it generates a target handling solution adapted to the current scenario. In the execution output layer, the system automatically executes scripts or assigns work orders according to the complexity of the solution. After repairing the fault, it records operation logs and execution results, generates a structured operation and maintenance report (supporting Word / PPT format and visualization charts), and feeds it back to the knowledge base, driving knowledge entries to continuously iterate and update through an incremental decay weight algorithm. This embodiment is particularly suitable for private network scenarios with high reliability requirements, such as enterprise parks, universities, and industrial parks. It significantly improves fault response speed, diagnostic accuracy, handling adaptability, and knowledge reuse capabilities, achieving efficient, secure, and self-evolving intelligent operation and maintenance.

[0110] Example 2:

[0111] like Figure 6 As shown, this embodiment provides an operation and maintenance system for a campus private network, which includes:

[0112] The data interaction module 10 is used to collect network operation data of the park's private network and also to receive operation and maintenance questions input by users.

[0113] The semantic parsing module 20, connected to the data interaction module 10, is used to perform semantic parsing on operation and maintenance issues and generate corresponding semantic vectors.

[0114] The knowledge retrieval module 30 is connected to the semantic parsing module 20 and is used to retrieve matching historical fault cases from the operation and maintenance knowledge base based on semantic vectors to obtain general handling solutions.

[0115] The root cause analysis module 40 is connected to the data interaction module 10 and the semantic parsing module 20 respectively. It is used to locate the root cause node based on semantic vectors and network operation data, using a Bayesian network, and to evaluate the impact of the fault by combining the topological centrality algorithm to determine the root cause location result.

[0116] The solution generation module 50 is connected to the knowledge retrieval module 30 and the root cause analysis module 40 respectively, and is used to generate a target solution that matches the operation and maintenance problem based on the general solution and the root cause location result.

[0117] The solution execution module 60, connected to the solution generation module 50, is used to repair faults in the park's private network according to the target disposal solution, and record the corresponding operation logs and execution results.

[0118] The report generation module 70 is connected to the solution execution module 60 and is used to generate structured operation and maintenance reports based on operation logs and execution results.

[0119] The system in this embodiment is capable of executing the method in Embodiment 1.

[0120] Example 3:

[0121] like Figure 7 As shown, this embodiment provides an electronic device, which includes a memory 200 and a processor 100. The memory 200 stores a computer program. When the processor 100 runs the computer program stored in the memory 200, the processor 100 executes the operation and maintenance method of the campus private network according to Embodiment 1.

[0122] It is understood that the above embodiments are merely exemplary implementations used to illustrate the principles of the present invention, and the present invention is not limited thereto. For those skilled in the art, various modifications and improvements can be made without departing from the spirit and essence of the present invention, and these modifications and improvements are also considered to be within the scope of protection of the present invention.

Claims

1. A method for operation and maintenance of a campus private network, characterized in that, The method comprises the following steps: receiving a user inputted operation and maintenance problem and collecting network operation data of the park private network; performing semantic analysis on the operation and maintenance problem to generate a corresponding semantic vector; based on the semantic vector, retrieving a matching historical fault case from an operation and maintenance knowledge base to obtain a general handling scheme; simultaneously, based on the semantic vector and the network operation data, using a Bayesian network to locate a root cause node and combining a topological centrality algorithm to evaluate fault impact to determine a root cause positioning result; generating a target handling scheme matching the operation and maintenance problem according to the general handling scheme and the root cause positioning result; repairing a fault in the park private network according to the target handling scheme and recording corresponding operation logs and execution results; generating a structured operation and maintenance report based on the operation logs and the execution results to realize park private network operation and maintenance.

2. The operation and maintenance method of the park private network according to claim 1, characterized in that receiving a user inputted operation and maintenance problem specifically comprises: establishing a user interaction channel through an embedded floating assistant embedded in a network management interface to receive user inputted text or receive user voice input; wherein the user interaction channel is realized based on Kubernetes multi-tenant containerized deployment; when a user initiates interaction through the user interaction channel, an independent session space isolated by resources is allocated for the user or a group to which the user belongs, and corresponding visual operation components are dynamically rendered according to the user's authority.

3. The operation and maintenance method of the park private network according to claim 1, characterized in that performing semantic analysis on the operation and maintenance problem to generate a corresponding semantic vector specifically comprises: when the operation and maintenance problem is inputted by voice, converting the voice input into text information through a voice recognition model; using a semantic vector model to vectorize process the operation and maintenance problem in text form to generate a corresponding semantic vector.

4. The park private network operation and maintenance method according to claim 1, characterized in that, The general handling scheme is obtained by retrieving a matching historical fault case from an operation and maintenance knowledge base based on the semantic vector, specifically comprising: calculating the vector similarity between the semantic vector and each knowledge item in the operation and maintenance knowledge base; based on the vector similarity, filtering out candidate knowledge items with a similarity higher than a preset threshold; according to the decay weight corresponding to the candidate knowledge item, sorting to select the handling scheme corresponding to the knowledge item with the highest weight as the general handling scheme; wherein the operation and maintenance knowledge base is updated using an incremental decay weight algorithm, newly added items are assigned an initial weight, the weight increases when the item is accessed, and the weight of all items decays over time and the number of times not accessed according to a preset decay factor.

5. The park private network operation and maintenance method according to claim 1, wherein, Based on the semantic vector and the network operation data, the root cause node is located using a Bayesian network, and the fault impact is evaluated combining a topological centrality algorithm to determine the root cause positioning result, specifically comprising: matching the semantic vector with the pre-defined fault mode in the fault analysis matrix to determine the preliminary fault type; based on the preliminary fault type, extracting relevant performance indicators and log data from the network operation data; input the performance indicators and the log data into a constructed Bayesian network model, calculate posterior probabilities of each network node being a root cause node, and locate the node with the highest posterior probability as the root cause node; based on the network topology structure, applying a topological centrality algorithm to calculate the root cause node, and identifying a key fault point; based on topological link information of the network topology structure, and according to the root cause node, the key fault point, and the network operation data, generating a root cause positioning result including a fault influence diffusion thermodynamic diagram.

6. The park private network operation and maintenance method according to claim 1, wherein, The target treatment scheme matched with the operation and maintenance problem is generated according to the general treatment scheme and the root cause positioning result, specifically including: performing consistency checking on the general treatment scheme and the root cause positioning result; when the checking result is matching, taking the general treatment scheme as the target treatment scheme; when the checking result is not matching or there are multiple concurrent root causes, based on the causal path in the knowledge graph and the device context information, performing step clipping, supplementing or reordering on the general treatment scheme to generate a target treatment scheme adapted to the current fault scenario.

7. The park private network operation and maintenance method according to claim 1, characterized in that, The structured operation and maintenance report is generated based on the operation log and the execution result, specifically including: collecting the operation log and the execution result; based on a preset report template, integrating the operation log and the execution result through a drag-and-drop canvas editing interface to generate a structured preliminary operation and maintenance report; through a multi-format output engine, dynamically converting the preliminary operation and maintenance report into a Word or PPT format document, and associating and rendering a visual chart of a key data node, to realize generation of the structured operation and maintenance report.

8. The operation and maintenance method of the park private network according to any one of claims 1 to 7, characterized in that, after the structured operation and maintenance report is generated, the method further includes: feeding back the structured operation and maintenance report to the operation and maintenance knowledge base to iteratively update the operation and maintenance knowledge base.

9. An operation and maintenance system of a campus private network, characterized in that, The system includes: a data interaction module configured to collect network operation data of the park private network, and further configured to receive an operation and maintenance problem input by a user; a semantic analysis module connected with the data interaction module, configured to perform semantic analysis on the operation and maintenance problem to generate a corresponding semantic vector; a knowledge retrieval module connected with the semantic analysis module, configured to retrieve a matching historical fault case from an operation and maintenance knowledge base based on the semantic vector to obtain a general treatment scheme; a root cause analysis module connected with the data interaction module and the semantic analysis module respectively, configured to locate a root cause node based on the semantic vector and the network operation data by using a Bayesian network, and determine a root cause positioning result by combining a topological centrality algorithm to evaluate fault influence; a scheme generation module connected with the knowledge retrieval module and the root cause analysis module respectively, configured to generate a target treatment scheme matched with the operation and maintenance problem according to the general treatment scheme and the root cause positioning result; a scheme execution module connected with the scheme generation module, configured to repair a fault in the park private network according to the target treatment scheme, and record a corresponding operation log and execution result; A report generation module, connected with the scheme execution module, is configured to generate a structured operation and maintenance report based on the operation log and the execution result.

10. An electronic device comprising a memory and a processor, characterized in that, The memory stores a computer program, and when the processor executes the computer program stored in the memory, the processor executes the operation and maintenance method of the park private network according to any one of claims 1-8.

Images