A communication network intelligent operation and maintenance and fault emergency disposal system and method

Abstract

This invention discloses an intelligent operation and maintenance (O&M) and emergency response system and method for communication networks. The system comprises an intranet large-scale model platform module, a local API service module, a GNN topology inference module, a standardized database module, an interface adaptation and streaming transmission module, an AI agent autonomous execution module, a data integrity and self-consistency verification module, and an intelligent O&M and emergency expansion module. Based on intranet large-scale model and graph neural network technology, this system constructs an intelligent O&M system for communication networks, achieving rapid and accurate handling of all types of faults, real-time controllable O&M data, secure implementation of intranet AI, and automated closed-loop O&M. This improves network O&M efficiency and stability, ensures core data security, and is suitable for industrial-grade high-security and high-intensity O&M scenarios.

Description

Technical Field

[0001] This invention belongs to the field of intelligent operation and maintenance technology of communication networks, and specifically relates to an intelligent operation and maintenance and fault emergency response system and method for communication networks. Background Technology

[0002] In the current communication network full-domain operation and maintenance scenario, the traditional operation and maintenance model has core problems such as insufficient algorithm support, lack of data management and control, low level of intelligence, and lagging interface linkage, making it difficult to adapt to the unified handling needs of all types of network faults.

[0003] The existing operation and maintenance model has the following shortcomings: 1. Low efficiency in handling all types of faults and lack of universal intelligent algorithms: Existing technologies are mostly designed for single optical cable faults and cannot cover all scenarios of faults such as equipment downtime, power outages, and node anomalies. They still rely on manual troubleshooting, manual path planning, and preparation of emergency repair plans, which is time-consuming and prone to errors. Conventional algorithms cannot adapt to complex mesh topologies and do not combine link weights, service priorities, and fault types for optimized reasoning, making it difficult to quickly match the optimal handling solution and resulting in low decision-making accuracy. 2. Lagging emergency plan management and lack of dynamic simulation mechanism: Existing emergency plans are mostly static texts prepared manually and lack intelligent matching and simulation algorithms adapted to all types of faults. They cannot be dynamically adjusted with topology changes, resource updates, and fault levels. In the event of a sudden fault, the plan has poor adaptability, its feasibility cannot be verified, and the emergency response process is chaotic. 3. Technical barriers exist for internal network AI integration, with no standardized adaptation solutions: Public cloud large-scale models cannot be accessed via the internal network, posing a risk of core data leakage. The internal network AI platform and local operation and maintenance API lack supporting algorithms such as authentication adaptation, streaming transmission, and timeout control, easily leading to issues such as authentication incompatibility, binary stream parsing failure, cross-network segment request timeouts, and cumbersome configuration. Deep integration of AI capabilities and operation and maintenance business is difficult. 4. Operation and maintenance data is lagging and of uncontrollable quality, lacking a full-process management mechanism: Currently, only the northbound interfaces of each transmission system's network management are accessible, supporting only periodic batch updates of operation and maintenance data. There are no supporting data integrity and consistency verification mechanisms. After data updates, issues such as field conflicts, broken associations, contradictory states, and topology errors are highly likely to occur, resulting in extremely poor data availability. Furthermore, real-time data synchronization and southbound interface linkage are not implemented, making one-click configuration data distribution impossible. Network management operations rely on manual processes, resulting in extremely low automation levels. 5. Lack of autonomous operation and maintenance capabilities and no closed-loop execution system: There is no autonomous decision-making intelligent agent architecture and supporting task decision-making and API autonomous calling algorithms adapted to industrial-grade operation and maintenance scenarios. It is impossible to achieve a natural language-driven, unmanned, full-domain operation and maintenance closed loop. Complex tasks need to be manually broken down and executed. The overall level of intelligence and autonomy is difficult to meet the needs of industrial-grade operation and maintenance. Summary of the Invention

[0004] To address the shortcomings of existing technologies, one of the objectives of this invention is to provide an intelligent operation and maintenance and emergency response system for communication networks. Relying on large-scale intranet models and graph neural network technology, it constructs a secure, efficient, and fully autonomous intelligent operation and maintenance system for communication networks. This system enables rapid and accurate handling of all types of faults, real-time control of operation and maintenance data, secure implementation of intranet AI, and automated closed-loop operation and maintenance. It improves network operation and maintenance efficiency and stability, ensures the security of core data, and is suitable for industrial-grade high-security and high-intensity operation and maintenance scenarios.

[0005] The second objective of this invention is to provide a method for using the intelligent operation and maintenance and emergency response system for the communication network.

[0006] This invention provides an intelligent operation and maintenance and emergency response system for communication networks, including an intranet large model platform module, a local API service module, a GNN topology inference module, a standardized database module, an interface adaptation and streaming transmission module, an AI agent autonomous execution module, a data integrity and self-consistency verification module, and an intelligent operation and maintenance and emergency expansion module.

[0007] The intranet large model platform module is deployed in isolation on the intranet, equipped with natural language parsing, instruction generation, and fault type identification algorithms. It supports one-click import of API configurations via curl command, providing underlying inference support for AI agents. Core data does not leave the intranet and is compatible with fault instruction parsing and decision output for all categories of products.

[0008] The local API service module listens to a designated port on the intranet and is equipped with algorithms for permission verification, parameter parsing, streaming transmission scheduling, and multi-format result encapsulation. It uses Basic authentication + Base64 encoding verification to complete identity verification. It works in conjunction with the GNN topology inference module to complete fault location and path planning for all categories of faults. It outputs multiple types of results in binary streaming, including fault solutions, emergency plans, resource lists, and verification reports in Excel format, as well as optimal and backup route comparison diagrams, full network topology diagrams, fault impact range diagrams, and resource usage diagrams in PNG / SVG format. This solves the problems of traditional interface output being single, lacking visualization, and unstable transmission.

[0009] The GNN topology reasoning module is based on the GNN reasoning algorithm, which abstracts the entire network of facilities such as computer rooms, base stations, transmission equipment, and power nodes into a directed weighted topology graph. It is suitable for all types of fault location, path planning, risk prediction and contingency plan simulation. Before reasoning, it is linked to the data verification module to verify the validity of the data and ensure the accuracy of reasoning.

[0010] The standardized database module establishes a unified and standardized data table to store all data, including the entire network topology, device parameters, power information, fault logs, emergency plans, and operation and maintenance records. It synchronously connects to the northbound and southbound interfaces of the network management systems of various transmission systems, and has a built-in real-time data synchronization algorithm to realize automatic reading and real-time database updates of basic data on the network management side, ensuring that the data is fresh and accurate, and providing unified data support for the entire system.

[0011] The interface adapter and streaming module is equipped with authentication encapsulation, flow control, and timeout control algorithms. It is configured with a string unparsed receiving mode and a timeout threshold of 30 seconds, which solves the problem of connecting intranet AI with local API and ensures stable transmission of requests in all scenarios.

[0012] The AI ​​intelligent agent autonomous execution module is based on the autonomous decision-making intelligent agent framework and is equipped with full-scenario task decomposition, decision scheduling, API autonomous calling, and memory iteration algorithms. It can adapt to various fault handling logics and complete the entire process of operation and maintenance tasks without manual intervention.

[0013] The data integrity and self-consistency verification module serves as a prerequisite for system operation. It is equipped with integrity verification, self-consistency judgment, and bidirectional reverse lookup algorithms to verify the full-domain operation and maintenance data from all dimensions, prevent dirty data from entering the inference process, and support the synchronization of abnormal data to the network management system after repair, so as to achieve bidirectional data consistency.

[0014] The intelligent operation and maintenance and emergency expansion module is based on core algorithms to realize functions such as full-category fault plan management, early warning, resource scheduling, and work order dispatch. It has a built-in network management interface linkage unit, supports one-click configuration data distribution and remote network management automatic operation, replaces manual network management operation, realizes full automation of configuration process, and covers all domain operation and maintenance scenarios.

[0015] In the GNN topology reasoning module, a full-category fault location and path planning algorithm is constructed based on the GNN reasoning algorithm, specifically as follows:

[0016] Quantitative construction of the target region's global topology map;

[0017] Based on historical fault, topology, and operation and maintenance data of all product categories, a graph convolutional network is used for iterative training to learn node association, link connectivity, and fault propagation patterns, optimize model parameters, and obtain a fault localization and path reasoning model for all product categories.

[0018] Based on the full-category fault location and path reasoning model, the fault identifier is input to complete the fault type determination and location. Then, based on the optimized Dijkstra algorithm combined with topology weights, idle resources, backup links and backup equipment are searched to select the processing solution with the least business impact and the fastest recovery.

[0019] By combining the data integrity and self-consistency verification module, abnormal correlation schemes are eliminated, and standardized inference results are output.

[0020] The global topology graph quantization construction includes the following steps:

[0021] Set all facilities as topology nodes, assign node feature vectors, and represent them using the following formula: ;in, Longitude Latitude; For facility type; For carrying capacity; Load rate; This refers to the node's running status;

[0022] Fiber optic links, device connections, and power supply lines are defined as topological edges, and edge feature vectors are assigned, represented by the following formula: ;in, For length; This is the loss value; This indicates the link's operational status. The edge weight; Prioritize business needs;

[0023] Edge weights are calculated using the following weighting formula: ;in, This is the first preset weighting coefficient; This is the second preset weighting coefficient; This is the third preset weighting coefficient; This is the fourth preset weighting coefficient.

[0024] The data integrity and self-consistency verification module employs quantitative judgment rules to ensure data compliance and constructs a comprehensive operational data quality control algorithm, specifically including:

[0025] Traverse the entire domain operation and maintenance data table and perform non-null, unique, and format validation on key fields including ID, parameters, and status;

[0026] Verify the consistency of data association logic and status, and check the matching of node-device-optical cable-power supply associations, the absence of resource status conflicts, and the absence of isolated nodes and broken links in the topology;

[0027] Establish a bidirectional index for the entire data domain to perform forward queries and reverse tracing;

[0028] Locate abnormal data, generate an anomaly report and push repair prompts; repair and review pass the inspection to proceed to the next step.

[0029] After the anomaly is repaired, it will be automatically synchronized to the network management of each transmission system to ensure bidirectional consistency between internal network data and network management data.

[0030] The process involves traversing the entire domain operation and maintenance data table and performing non-empty, unique, and format checks on key fields including ID, parameters, and status, using the following formula: ;in, n is the integrity indicator for a single data entry, with 1 for complete and 0 for missing; n is the total amount of data. This is an integrity metric. When the integrity metric is below a preset threshold, it is marked as an integrity anomaly.

[0031] The standardized database module is compatible with the northbound and southbound interface protocols of the network management system, completing interface authentication and link establishment, and supporting compatible integration with network management systems of multiple transmission systems. The standardized database module reads basic data from the network management side, including topology, devices, power supply, and resources, using a combination of timed and triggered methods. After automatic cleaning and verification, the data is stored in the database, replacing outdated data and ensuring real-time synchronization between system data and the network management side. Based on fault handling and maintenance needs, the standardized database module automatically encapsulates compliant configuration commands and sends them to the corresponding network management system with one click through the southbound interface to perform remote configuration operations, with full traceability.

[0032] The AI ​​agent autonomous execution module, based on the ReAct reasoning-action core paradigm, enables autonomous decomposition and closed-loop execution of operation and maintenance tasks across all scenarios. Specifically, it includes the following steps:

[0033] Receive natural language operation and maintenance instructions, identify task objectives, and extract key parameters including fault location, type, and level;

[0034] The divide-and-conquer algorithm is used to break down complex tasks into standardized subtasks and determine the execution order and dependencies.

[0035] Match the corresponding API with the network management interface, automatically complete the authentication encoding, encapsulate the request parameters, make asynchronous calls according to priority, monitor the execution status, and automatically retry when timeout;

[0036] Real-time tracking of execution results and automatic downgrade handling of anomalies; after task completion, the handling process and results are stored in the memory bank to optimize subsequent decision-making logic and achieve continuous iteration.

[0037] The present invention also provides a method for using the intelligent operation and maintenance and emergency response system for the communication network, comprising the following steps:

[0038] The latest operation and maintenance data is synchronized in real time via the network management interface;

[0039] Conduct comprehensive data integrity and self-consistency verification;

[0040] Analyze fault information and fault type;

[0041] Complete interface authentication and send the call request;

[0042] Optimal fault handling solution for GNN inference;

[0043] Generate Excel format reports and visual images.

[0044] When there are configuration changes in the fault handling plan, configuration commands can be issued with one click through the network management interface to complete the processing; the processing results are archived and stored.

[0045] This invention discloses an intelligent operation and maintenance and emergency response system and method for communication networks, which has the following beneficial effects:

[0046] 1. Breaking through the limitations of single optical cable faults, the general GNN algorithm enables second-level fault location and solution planning for all types of faults. The fully automated process reduces manual handling time from hours to seconds, significantly shortening business interruption time.

[0047] 2. The northbound / southbound interfaces of the network management system are linked to achieve real-time data synchronization and updates, ensuring data freshness and validity; a full data verification mechanism is provided to eliminate dirty data from the source; one-click configuration and automatic network management operations are supported, completely replacing manual configuration and eliminating operational errors.

[0048] 3. Deployed entirely within an intranet with no public network data interaction, solving the challenges of intranet integration and API adaptation for large models, while ensuring both intelligence and core data security;

[0049] 4. Relying on an autonomous decision-making intelligent agent, it can autonomously decompose tasks and autonomously call interfaces, operate 24 / 7 without human intervention, adapt to various fault handling, greatly free up manpower, and improve the consistency of operation and maintenance.

[0050] 5. Multiple safeguards, including pre-data verification, exception retry, and access control, reduce the probability of system failures and decision-making errors, ensuring stable adaptation to high-intensity industrial-grade operation and maintenance scenarios. Attached Figure Description

[0051] Figure 1 This is a schematic diagram of the system of the present invention;

[0052] Figure 2 This is a flowchart illustrating the method of the present invention. Detailed Implementation

[0053] This invention provides an intelligent operation and maintenance and emergency response system for communication networks, the structural diagram of which is shown below. Figure 1 As shown, it includes an intranet large model platform module, a local API service module, a GNN topology inference module, a standardized database module, an interface adaptation and streaming module, an AI intelligent agent autonomous execution module, a data integrity and self-consistency verification module, and an intelligent operation and maintenance and emergency expansion module.

[0054] The intranet large model platform module is deployed in isolation on the intranet, equipped with natural language parsing, instruction generation, and fault type identification algorithms. It supports one-click import of API configurations via curl command, providing underlying inference support for AI agents. Core data does not leave the intranet and is compatible with fault instruction parsing and decision output for all categories of products.

[0055] The local API service module listens to a designated port on the intranet and is equipped with algorithms for permission verification, parameter parsing, streaming transmission scheduling, and multi-format result encapsulation. It uses Basic authentication + Base64 encoding verification to complete identity verification. It works in conjunction with the GNN topology inference module to complete fault location and path planning for all categories of faults. It outputs multiple types of results in binary streaming, including fault solutions, emergency plans, resource lists, and verification reports in Excel format, as well as optimal and backup route comparison diagrams, full network topology diagrams, fault impact range diagrams, and resource usage diagrams in PNG / SVG format. This solves the problems of traditional interface output being single, lacking visualization, and unstable transmission.

[0056] The GNN topology reasoning module is based on the GNN reasoning algorithm, which abstracts the entire network of facilities such as computer rooms, base stations, transmission equipment, and power nodes into a directed weighted topology graph. It is suitable for all types of fault location, path planning, risk prediction and contingency plan simulation. Before reasoning, it is linked to the data verification module to verify the validity of the data and ensure the accuracy of reasoning.

[0057] The standardized database module establishes a unified and standardized data table to store all data, including the entire network topology, device parameters, power information, fault logs, emergency plans, and operation and maintenance records. It synchronously connects to the northbound and southbound interfaces of the network management systems of various transmission systems, and has a built-in real-time data synchronization algorithm to realize automatic reading and real-time database updates of basic data on the network management side, ensuring that the data is fresh and accurate, and providing unified data support for the entire system.

[0058] The interface adapter and streaming module is equipped with authentication encapsulation, flow control, and timeout control algorithms. It is configured with a string unparsed receiving mode and a timeout threshold of 30 seconds, which solves the problem of connecting intranet AI with local API and ensures stable transmission of requests in all scenarios.

[0059] The AI ​​intelligent agent autonomous execution module is based on the autonomous decision-making intelligent agent framework and is equipped with full-scenario task decomposition, decision scheduling, API autonomous calling, and memory iteration algorithms. It can adapt to various fault handling logics and complete the entire process of operation and maintenance tasks without manual intervention.

[0060] The data integrity and self-consistency verification module serves as a prerequisite for system operation. It is equipped with integrity verification, self-consistency judgment, and bidirectional reverse lookup algorithms to verify the full-domain operation and maintenance data from all dimensions, prevent dirty data from entering the inference process, and support the synchronization of abnormal data to the network management system after repair, so as to achieve bidirectional data consistency.

[0061] The intelligent operation and maintenance and emergency expansion module is based on core algorithms to realize functions such as full-category fault plan management, early warning, resource scheduling, and work order dispatch. It has a built-in network management interface linkage unit, supports one-click configuration data distribution and remote network management automatic operation, replaces manual network management operation, realizes full automation of configuration process, and covers all domain operation and maintenance scenarios.

[0062] In the GNN topology reasoning module, a full-category fault location and path planning algorithm is constructed based on the GNN reasoning algorithm, specifically as follows:

[0063] Quantitative construction of the target region's global topology map;

[0064] Based on historical fault, topology, and operation and maintenance data of all product categories, a graph convolutional network is used for iterative training to learn node association, link connectivity, and fault propagation patterns, optimize model parameters, and obtain a fault localization and path reasoning model for all product categories.

[0065] Based on the full-category fault location and path reasoning model, the fault identifier is input to complete the fault type determination and location. Then, based on the optimized Dijkstra algorithm combined with topology weights, idle resources, backup links and backup equipment are searched to select the processing solution with the least business impact and the fastest recovery.

[0066] By combining the data integrity and self-consistency verification module, abnormal correlation schemes are eliminated, and standardized inference results are output.

[0067] The global topology graph quantization construction includes the following steps:

[0068] Set all facilities as topology nodes, assign node feature vectors, and represent them using the following formula: ;in, Longitude Latitude; For facility type; For carrying capacity; Load rate; This refers to the node's running status;

[0069] Fiber optic links, device connections, and power supply lines are defined as topological edges, and edge feature vectors are assigned, represented by the following formula: ;in, For length; This is the loss value; This indicates the link's operational status. The edge weight; Prioritize business needs;

[0070] Edge weights are calculated using the following weighting formula: ;in, This is the first preset weighting coefficient; This is the second preset weighting coefficient; This is the third preset weighting coefficient; This is the fourth preset weighting coefficient.

[0071] The data integrity and self-consistency verification module employs quantitative judgment rules to ensure data compliance and constructs a comprehensive operational data quality control algorithm, specifically including:

[0072] Traverse the entire domain operation and maintenance data table and perform non-null, unique, and format validation on key fields including ID, parameters, and status;

[0073] Verify the consistency of data association logic and status, and check the matching of node-device-optical cable-power supply associations, the absence of resource status conflicts, and the absence of isolated nodes and broken links in the topology;

[0074] Establish a bidirectional index for the entire data domain to perform forward queries and reverse tracing;

[0075] Locate abnormal data, generate an anomaly report and push repair prompts; repair and review pass the inspection to proceed to the next step.

[0076] After the anomaly is repaired, it will be automatically synchronized to the network management of each transmission system to ensure bidirectional consistency between internal network data and network management data.

[0077] The process involves traversing the entire domain operation and maintenance data table and performing non-empty, unique, and format checks on key fields including ID, parameters, and status, using the following formula: ;in, n is the integrity identifier for a single data entry, where 1 represents complete data and 0 represents missing data; n is the total amount of data. This is an integrity metric. When the integrity metric is below a preset threshold, it is marked as an integrity anomaly.

[0078] The standardized database module is compatible with the northbound and southbound interface protocols of the network management system, completing interface authentication and link establishment, and supporting compatible integration with network management systems of multiple transmission systems. The standardized database module reads basic data from the network management side, including topology, devices, power supply, and resources, using a combination of timed and triggered methods. After automatic cleaning and verification, the data is stored in the database, replacing outdated data and ensuring real-time synchronization between system data and the network management side. Based on fault handling and maintenance needs, the standardized database module automatically encapsulates compliant configuration commands and sends them to the corresponding network management system with one click through the southbound interface to perform remote configuration operations, with full traceability.

[0079] The AI ​​agent autonomous execution module, based on the ReAct reasoning-action core paradigm, enables autonomous decomposition and closed-loop execution of operation and maintenance tasks across all scenarios. Specifically, it includes the following steps:

[0080] Receive natural language operation and maintenance instructions, identify task objectives, and extract key parameters including fault location, type, and level;

[0081] The divide-and-conquer algorithm is used to break down complex tasks into standardized subtasks and determine the execution order and dependencies.

[0082] Match the corresponding API with the network management interface, automatically complete the authentication encoding, encapsulate the request parameters, make asynchronous calls according to priority, monitor the execution status, and automatically retry when timeout;

[0083] Real-time tracking of execution results and automatic downgrade handling of anomalies; after task completion, the handling process and results are stored in the memory bank to optimize subsequent decision-making logic and achieve continuous iteration.

[0084] The present invention also provides a method for using the intelligent operation and maintenance and emergency response system for the communication network, the flowchart of which is shown below. Figure 2 As shown, it includes the following steps:

[0085] The latest operation and maintenance data is synchronized in real time via the network management interface;

[0086] Conduct comprehensive data integrity and self-consistency verification;

[0087] Analyze fault information and fault type;

[0088] Complete interface authentication and send the call request;

[0089] Optimal fault handling solution for GNN inference;

[0090] Generate Excel format reports and visual images.

[0091] When there are configuration changes in the fault handling plan, configuration commands can be issued with one click through the network management interface to complete the processing; the processing results are archived and stored.

[0092] The system of the present invention will be further described below with reference to embodiments:

[0093] Standard manual triggering of full network fault handling: Operation and maintenance personnel enter full network fault information, covering various scenarios such as fiber optic cable interruption, equipment downtime, and power supply abnormality;

[0094] The system reads the latest operation and maintenance data of the corresponding area in real time and completes synchronous updates through the northbound and southbound interfaces of the transmission network management system;

[0095] Initiate dual verification of the integrity and self-consistency of operation and maintenance data across the entire domain;

[0096] After verification, the intranet big data model parses the fault parameters and encapsulates standardized API requests;

[0097] The GNN topology reasoning module completes the accurate location of fault points, deduction of the optimal handling solution, and comparison of the advantages and disadvantages of routes.

[0098] The local API service module outputs Excel-formatted fault recovery plans, resource scheduling lists, and visualizations such as optimal and backup route comparison diagrams, fault topology location maps, and fault impact range maps.

[0099] The corresponding configuration command can be issued with one click through the southbound interface of the network management system to automatically complete the remote network management configuration operation.

[0100] The system automatically generates and dispatches maintenance work orders;

[0101] After the fault is resolved, data from the entire process is collected, a debriefing report is generated and archived, and the entire process is closed-loop with traceable data. The visualized results can effectively assist maintenance personnel in making quick decisions.

[0102] AI-powered intelligent agents autonomously handle network-wide faults: Operations and maintenance personnel issue natural language operations and maintenance commands through intranet office tools;

[0103] The autonomous decision-making AI agent receives instructions, analyzes the maintenance intent, and determines the fault type;

[0104] Trigger real-time data synchronization of the network management interface, automatically pull the latest global operation and maintenance data and complete the pre-qualification quality check;

[0105] The intelligent agent breaks down complex operation and maintenance tasks and autonomously calls the corresponding API interfaces and network management interfaces;

[0106] The GNN model completes fault location, disposal plan planning, and feasibility comparison;

[0107] The local API module outputs handling plans, emergency plans, and visualizations such as route comparison diagrams, full network topology diagrams, and fault impact range diagrams in binary stream Excel format.

[0108] The intelligent agent automatically encapsulates compliant configuration commands and issues them with one click through the network management interface to complete automated network management operations;

[0109] The intelligent agent dispatches maintenance work orders and monitors the execution progress throughout the process.

[0110] After the fault is repaired, the intelligent agent memory library, emergency plan and operation and maintenance resource data are updated synchronously. No manual intervention is required throughout the process, realizing fully autonomous operation and maintenance of all types of network faults.

[0111] Implementation of a comprehensive operational data integrity and self-consistency verification project: For routine data quality control, a separate verification process is initiated to solidify the data foundation for handling faults across all product categories. The steps are as follows:

[0112] Manually triggered on a timed basis, or automatically triggered by the system before fault handling or contingency plan updates, selecting the entire verification scope including backbone network, equipment, power supply, topology, etc.

[0113] Traverse the data table to check the compliance of fields, detect anomalies such as data integrity failure, resource status conflicts, and orphaned nodes, and locate the location and root cause of the anomalies;

[0114] By using two-way reverse lookup to locate the cause of the anomaly, a repair prompt is pushed, and the operation and maintenance personnel complete the correction and re-verify until the data is fully compliant.

[0115] Once the verification is successful, the data is automatically synchronized to the network management system, unlocking full-process operation and maintenance functions, retaining verification and repair logs, and meeting compliance audit requirements.

[0116] This embodiment fully highlights the pre-emptive guarantee role of the data verification module and the network management interface linkage. Through the dual control mechanism of real-time data synchronization on the network management side and two-layer data quality verification, it eliminates problems such as GNN inference failure, interface call anomalies, and fault handling errors caused by dirty data from the source. At the same time, it ensures that all subsequent output of various solution reports and visualization images are generated based on compliant and valid operation and maintenance data. It also provides reliable data support for subsequent one-click configuration data distribution and automated network management operations, ensuring the stable, accurate, and efficient operation of the full-domain intelligent operation and maintenance system in all aspects.

Claims

1. A communication network intelligent operation and maintenance and fault emergency response system, characterized in that, It includes an intranet large model platform module, a local API service module, a GNN topology inference module, a standardized database module, an interface adaptation and streaming module, an AI agent autonomous execution module, a data integrity and self-consistency verification module, and an intelligent operation and maintenance and emergency expansion module. The intranet large model platform module is deployed in isolation on the intranet, equipped with natural language parsing, instruction generation, and fault type identification algorithms. It supports one-click import of API configurations via curl command, providing underlying inference support for AI agents. Core data does not leave the intranet and is compatible with fault instruction parsing and decision output for all categories of products. The local API service module listens to a designated port on the intranet and is equipped with algorithms for permission verification, parameter parsing, streaming transmission scheduling, and multi-format result encapsulation. It uses Basic authentication + Base64 encoding verification to complete identity verification. It works in conjunction with the GNN topology inference module to complete fault location and path planning for all categories of faults and outputs multiple types of results in binary streaming. The results include fault solutions, emergency plans, resource lists, and verification reports in Excel format, as well as optimal and backup route comparison diagrams, full network topology diagrams, fault impact range diagrams, and resource usage diagrams in PNG / SVG format. The GNN topology reasoning module is based on the GNN reasoning algorithm, which abstracts the entire facility, including the computer room, base station, transmission equipment, and power nodes, into a directed weighted topology graph. It is suitable for all types of fault location, path planning, risk prediction and contingency plan simulation. Before reasoning, it is linked to the data verification module to verify the validity of the data and ensure the accuracy of reasoning. The standardized database module establishes a unified and standardized data table to store full data including the entire network topology, device parameters, power information, fault logs, emergency plans, and operation and maintenance records; it synchronously connects to the northbound and southbound interfaces of the network management systems of various transmission systems, and has a built-in real-time data synchronization algorithm to realize automatic reading and real-time database updates of basic data on the network management side, ensuring data freshness and accuracy, and providing unified data support for the entire system; The interface adapter and streaming module is equipped with authentication encapsulation, flow control, and timeout control algorithms. It is configured with a string unparsed receiving mode and a timeout threshold of 30 seconds, which solves the problem of connecting intranet AI with local API and ensures stable transmission of requests in all scenarios. The AI ​​agent autonomous execution module is based on the autonomous decision-making agent framework and is equipped with full-scenario task decomposition, decision scheduling, API autonomous calling, memory iteration algorithm, and adaptive fault handling logic. It can complete the entire process of operation and maintenance tasks without manual intervention. The data integrity and self-consistency verification module serves as a prerequisite for system operation. It is equipped with integrity verification, self-consistency judgment, and bidirectional reverse lookup algorithms to verify the full-domain operation and maintenance data from all dimensions, prevent dirty data from entering the inference process, and support the synchronization of abnormal data to the network management system after repair, so as to achieve bidirectional data consistency. The intelligent operation and maintenance and emergency expansion module is based on core algorithms to realize functions including full-category fault plan management, early warning, resource scheduling, and work order dispatch. It has a built-in network management interface linkage unit, supports one-click configuration data distribution and remote network management automatic operation, replaces manual network management operation, realizes full automation of configuration process, and covers all domain operation and maintenance scenarios.

2. The intelligent operation and maintenance and emergency response system for communication networks according to claim 1, characterized in that, In the GNN topology reasoning module, a full-category fault location and path planning algorithm is constructed based on the GNN reasoning algorithm, specifically as follows: Quantitative construction of the target region's global topology map; Based on historical fault, topology, and operation and maintenance data of all product categories, a graph convolutional network is used for iterative training to learn node association, link connectivity, and fault propagation patterns, optimize model parameters, and obtain a fault localization and path reasoning model for all product categories. Based on the full-category fault location and path reasoning model, the fault identifier is input to complete the fault type determination and location. Then, based on the optimized Dijkstra algorithm combined with topology weights, idle resources, backup links and backup equipment are searched to select the processing solution with the least business impact and the fastest recovery. By combining the data integrity and self-consistency verification module, abnormal correlation schemes are eliminated, and standardized inference results are output.

3. The intelligent operation and maintenance and emergency response system for communication networks according to claim 2, characterized in that, The global topology graph quantization construction includes the following steps: Set all facilities as topology nodes, assign node feature vectors, and represent them using the following formula: ;in, Longitude Latitude; For facility type; For carrying capacity; Load rate; This refers to the node's running status; Fiber optic links, device connections, and power supply lines are defined as topological edges, and edge feature vectors are assigned, represented by the following formula: ;in, For length; This is the loss value; This indicates the link's operational status. The edge weight; Prioritize business needs; Edge weights are calculated using the following weighting formula: ;in, This is the first preset weighting coefficient; This is the second preset weighting coefficient; This is the third preset weighting coefficient; This is the fourth preset weighting coefficient.

4. The intelligent operation and maintenance and emergency response system for communication networks according to claim 1, characterized in that, The data integrity and self-consistency verification module employs quantitative judgment rules to ensure data compliance and constructs a comprehensive operational data quality control algorithm, specifically including: Traverse the entire domain operation and maintenance data table and perform non-null, unique, and format validation on key fields including ID, parameters, and status; Verify the consistency of data association logic and status, and check the matching of node-device-optical cable-power supply associations, the absence of resource status conflicts, and the absence of isolated nodes and broken links in the topology; Establish a bidirectional index for the entire data domain to perform forward queries and reverse tracing; Locate abnormal data, generate an anomaly report and push repair prompts; repair and review pass the inspection to proceed to the next step. After the anomaly is repaired, it will be automatically synchronized to the network management of each transmission system to ensure bidirectional consistency between internal network data and network management data.

5. The intelligent operation and maintenance and emergency response system for communication networks according to claim 4, characterized in that, The process involves traversing the entire domain operation and maintenance data table and performing non-empty, unique, and format checks on key fields including ID, parameters, and status, using the following formula: ;in, n is the integrity identifier for a single data entry, where 1 represents complete data and 0 represents missing data; n is the total amount of data. This is an integrity metric. When the integrity metric is below a preset threshold, it is marked as an integrity anomaly.

6. The intelligent operation and maintenance and emergency response system for communication networks according to claim 1, characterized in that, The standardized database module is compatible with the northbound and southbound interface protocols of the network management system, completing interface authentication and link establishment, and supporting compatible integration with network management systems of multiple transmission systems. The standardized database module reads basic data from the network management side, including topology, devices, power supply, and resources, using a combination of timed and triggered methods. After automatic cleaning and verification, the data is stored in the database, replacing outdated data and ensuring real-time synchronization between system data and the network management side. Based on fault handling and maintenance needs, the standardized database module automatically encapsulates compliant configuration commands and sends them to the corresponding network management system with one click through the southbound interface to execute remote configuration operations, with full traceability for subsequent tracking needs.

7. The intelligent operation and maintenance and emergency response system for communication networks according to claim 1, characterized in that, The AI ​​agent autonomous execution module, based on the ReAct reasoning-action core paradigm, enables autonomous decomposition and closed-loop execution of operation and maintenance tasks across all scenarios. Specifically, it includes the following steps: Receive natural language operation and maintenance instructions, identify task objectives, and extract key parameters including fault location, type, and level; The divide-and-conquer algorithm is used to break down complex tasks into standardized subtasks and determine the execution order and dependencies. Match the corresponding API with the network management interface, automatically complete the authentication encoding, encapsulate the request parameters, make asynchronous calls according to priority, monitor the execution status, and automatically retry when timeout; Real-time tracking of execution results and automatic downgrade handling of anomalies; after task completion, the handling process and results are stored in the memory bank to optimize subsequent decision-making logic and achieve continuous iteration.

8. A method for using the intelligent operation and maintenance and emergency response system for communication networks according to any one of claims 1 to 7, characterized in that, Includes the following steps: The latest operation and maintenance data is synchronized in real time via the network management interface; Conduct comprehensive data integrity and self-consistency verification; Analyze fault information and fault type; Complete interface authentication and send the call request; Optimal fault handling solution for GNN inference; Generate Excel format reports and visual images.

9. The method according to claim 8, characterized in that, When there are configuration changes in the fault handling plan, configuration commands can be issued with one click through the network management interface to complete the processing; the processing results are archived and stored.

Images