Multi-operator optical fiber network fault diagnosis method, device, equipment and medium

Abstract

The invention discloses a multi-operator optical fiber network fault diagnosis method and device, equipment and a medium, and relates to the technical field of deep neural networks, and the method comprises the steps: constructing a three-dimensional digital twin model, and carrying out the parameter correction of the three-dimensional digital twin model; acquiring open source library data and operation and maintenance data of a plurality of operators to generate a cross-operator network asset database; respectively deploying the corrected three-dimensional digital twin model to each operator so that the operator can calculate corresponding model training parameters; acquiring all model training parameters, training the graph neural network model based on the model training parameters, and judging whether the trained graph neural network model is converged or not; if convergence occurs, inputting the cross-operator network asset database into the trained graph neural network model to predict an optical fiber network fault point; the fault report and the repair scheme are generated based on the optical fiber network fault points, the accuracy of multi-operator optical fiber network fault diagnosis is improved, and the cross-operator cooperation efficiency and the user experience feeling are improved.

Description

Technical Field

[0001] The present invention relates to the field of deep neural network technology, and in particular to a multi-operator optical fiber network fault diagnosis method, device, equipment and medium. Background Art

[0002] Currently, fiber optic networks operated and maintained independently by multiple operators, with fault responsibility determination relying on manual negotiation. The average processing time exceeds 24 hours, and fault location accuracy is insufficient. Existing OTDRs (Optical Time-Domain Reflectometers) can only locate faults down to a hundred meters and cannot distinguish between mixed fibers from multiple operators. To ensure service reliability, operators generally reserve more than 30% of redundant fiber, resulting in a fiber resource waste rate of up to 40%. Furthermore, each operator only opens limited interfaces, lacking a global network view, making dynamic cross-operator resource scheduling difficult.

[0003] As can be seen from the above, how to improve the accuracy of multi-operator fiber optic network fault diagnosis and enhance cross-operator collaboration efficiency and user experience is an issue to be solved in this field. Summary of the Invention

[0004] In view of this, the purpose of the present invention is to provide a multi-operator fiber optic network fault diagnosis method, device, equipment and medium, which can improve the accuracy of multi-operator fiber optic network fault diagnosis, enhance cross-operator collaboration efficiency and user experience. The specific scheme is as follows:

[0005] In a first aspect, the present application discloses a multi-operator optical fiber network fault diagnosis method, comprising:

[0006] Constructing a three-dimensional digital twin model, and performing parameter correction on the three-dimensional digital twin model to obtain a corrected three-dimensional digital twin model;

[0007] Obtaining open source library data and operation and maintenance data of multiple operators, and generating a cross-operator network asset database based on the open source library data and the operation and maintenance data;

[0008] Deploy the revised three-dimensional digital twin model to each operator, so that the operator can calculate the model training parameters corresponding to the revised three-dimensional digital twin model based on its own local data;

[0009] Obtaining the model training parameters sent by all the operators, training a preset graph neural network model based on the model training parameters, and determining whether the trained graph neural network model has converged;

[0010] If the trained graph neural network model converges, the cross-operator network asset database is input into the trained graph neural network model for fault diagnosis to predict the optical fiber network fault point;

[0011] A fault report and a repair plan are generated based on the optical fiber network fault point, and the fault report and the repair plan are sent to the corresponding operator.

[0012] Optionally, the constructing of a three-dimensional digital twin model includes:

[0013] Build an initial 3D digital twin model using building information modeling and geographic information systems;

[0014] The community terrain data and fiber optic completion drawings collected by the drone are combined with the initial three-dimensional digital twin model to obtain a three-dimensional digital twin model including fiber optic routing, splitters, and fiber optic distribution frames.

[0015] Optionally, the performing parameter correction on the three-dimensional digital twin model includes:

[0016] Use fiber optic monitoring equipment to collect optical power attenuation data;

[0017] Based on the segmented loss model, the extended Kalman filter and the optical power attenuation data, the parameters of the three-dimensional digital twin model are corrected by using an iterative nearest point algorithm.

[0018] Optionally, obtaining open source library data and operation and maintenance data of multiple operators includes:

[0019] Obtaining point cloud data from an open source library for processing 3D point cloud data, and performing denoising, segmentation, and fiber routing extraction on the point cloud data to obtain open source library data;

[0020] The operation and maintenance data is collected from the operator's network element management system using a standardized interface; the operation and maintenance data includes the port status of the optical line terminal, the online rate of the optical network unit, and the service flow.

[0021] Optionally, deploying the modified three-dimensional digital twin model to each operator separately so that the operator can calculate model training parameters corresponding to the modified three-dimensional digital twin model based on its own local data includes:

[0022] The revised three-dimensional digital twin model is deployed to each operator respectively, so that the operator can calculate the initial model training parameters corresponding to the revised three-dimensional digital twin model based on its own optical power anomaly records and historical fault labels. If the initial model training parameters are gradients, differential privacy is used to perform noise injection on the initial model training parameters to obtain the model training parameters.

[0023] Optionally, obtaining the model training parameters sent by all the operators, training a preset graph neural network model based on the model training parameters, and determining whether the trained graph neural network model has converged includes:

[0024] The model training parameters sent by all operators are aggregated using a federated learning algorithm and a preset resource collaboration strategy, and a preset graph neural network model is trained based on the model training parameters; the preset resource collaboration strategy includes a dynamic adjustment strategy for splitting resources, an optical path switching strategy, and a resource trading strategy;

[0025] Determining whether the trained graph neural network model has converged;

[0026] If the trained graph neural network model does not converge, the trained graph neural network model will be sent to each operator so that the operator can repeat the process of calculating the model training parameters until the trained graph neural network model converges.

[0027] Optionally, inputting the cross-operator network asset database into the trained graph neural network model for fault diagnosis includes:

[0028] Use micro optical power sensors to collect optical power data from optical fibers to achieve real-time monitoring of optical power across multiple operators.

[0029] The cross-operator network asset database is input into the trained graph neural network model for fault diagnosis; the graph neural network model is a model built based on the graph attention network and the graph convolutional neural network.

[0030] In a second aspect, the present application discloses a multi-operator optical fiber network fault diagnosis device, comprising:

[0031] A model construction and correction module is used to construct a three-dimensional digital twin model and perform parameter correction on the three-dimensional digital twin model to obtain a corrected three-dimensional digital twin model;

[0032] A database generation module, configured to obtain open source library data and operation and maintenance data of multiple operators, and generate a cross-operator network asset database based on the open source library data and the operation and maintenance data;

[0033] A model deployment module, configured to deploy the modified three-dimensional digital twin model to each operator, so that the operator can calculate model training parameters corresponding to the modified three-dimensional digital twin model based on its own local data;

[0034] A model training module, configured to obtain the model training parameters sent by all the operators, train a preset graph neural network model based on the model training parameters, and determine whether the trained graph neural network model has converged;

[0035] A fault diagnosis module is configured to input the cross-operator network asset database into the trained graph neural network model for fault diagnosis if the trained graph neural network model converges, so as to predict the optical fiber network fault point;

[0036] The report and solution sending module is used to generate a fault report and a repair solution based on the optical fiber network fault point, and send the fault report and the repair solution to the corresponding operator.

[0037] In a third aspect, the present application discloses an electronic device, comprising:

[0038] Memory, used to store computer programs;

[0039] A processor is used to execute the computer program to implement the aforementioned multi-operator optical fiber network fault diagnosis method.

[0040] In a fourth aspect, the present application discloses a computer storage medium for storing a computer program; wherein, when the computer program is executed by a processor, the steps of the multi-operator optical fiber network fault diagnosis method disclosed above are implemented.

[0041] It can be seen that the present application provides a multi-operator fiber optic network fault diagnosis method, including constructing a three-dimensional digital twin model, performing parameter correction on the three-dimensional digital twin model to obtain a corrected three-dimensional digital twin model; obtaining open source library data and operation and maintenance data of multiple operators, and generating a cross-operator network asset database based on the open source library data and the operation and maintenance data; deploying the corrected three-dimensional digital twin model to each operator respectively, so that the operator can calculate the model training parameters corresponding to the corrected three-dimensional digital twin model based on its own local data; obtaining the model training parameters sent by all the operators, training the preset graph neural network model based on the model training parameters, and judging whether the trained graph neural network model converges; if the trained graph neural network model converges, inputting the cross-operator network asset database into the trained graph neural network model for fault diagnosis to predict the fiber optic network fault point; generating a fault report and repair plan based on the fiber optic network fault point, and sending the fault report and repair plan to the corresponding operator. This application constructs a three-dimensional digital twin model that can intuitively display details such as optical fiber bending radius and fusion point loss, assist in project acceptance and maintenance, synchronize physical network status in real time, support virtual debugging, avoid actual operation risks, obtain open source library data and operation and maintenance data of multiple operators, and generate a cross-operator network asset database based on open source library data and operation and maintenance data to achieve unified modeling and collaborative operation and maintenance of cross-operator optical fiber networks, solving the problems of unclear responsibilities and waste of resources in traditional operation and maintenance. The revised three-dimensional digital twin model is deployed to each operator separately so that the operator can calculate the model training parameters corresponding to the revised three-dimensional digital twin model based on its own local data, and train the preset graph neural network model based on the model training parameters calculated by the operator. If the trained graph neural network model converges, the cross-operator network asset database is input into the trained graph neural network model for fault diagnosis, forming an edge-core hybrid decision-making mechanism, that is, the three-dimensional digital twin model is deployed to the operator, and the graph neural network model is trained and predicted, taking into account real-time and global optimization, improving the accuracy of multi-operator optical fiber network fault diagnosis, and improving cross-operator collaboration efficiency and user experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are merely embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0043] Figure 1 This is a flow chart of a multi-operator optical fiber network fault diagnosis method disclosed in this application;

[0044] Figure 2 This is a diagram of the horizontal federated learning technology architecture disclosed in this application;

[0045] Figure 3 This is a schematic structural diagram of a multi-operator optical fiber network fault diagnosis device disclosed in this application;

[0046] Figure 4 This is a structural diagram of an electronic device provided in this application. DETAILED DESCRIPTION

[0047] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0048] Currently, multi-operator fiber optic networks are independently operated and maintained, with fault responsibility determination relying on manual negotiation. The average processing time exceeds 24 hours, and fault location accuracy is insufficient. Existing OTDRs can only locate faults down to a hundred meters and cannot distinguish between mixed fibers from multiple operators. To ensure service reliability, operators generally reserve more than 30% of redundant fiber, resulting in a fiber resource waste rate as high as 40%. Furthermore, each operator only provides limited interfaces, lacking a global network view, making dynamic cross-operator resource scheduling difficult. As can be seen from the above, improving the accuracy of multi-operator fiber optic network fault diagnosis, enhancing cross-operator collaboration efficiency, and improving user experience are unresolved challenges in this field.

[0049] See also Figure 1 As shown, the embodiment of the present invention discloses a multi-operator optical fiber network fault diagnosis method, which may specifically include:

[0050] Step S11: construct a three-dimensional digital twin model, and perform parameter correction on the three-dimensional digital twin model to obtain a corrected three-dimensional digital twin model.

[0051] In this embodiment, an initial three-dimensional digital twin model is constructed using a building information model and a geographic information system; the community terrain data and optical fiber completion drawings collected by drones are combined with the initial three-dimensional digital twin model to obtain a three-dimensional digital twin model including optical fiber routing, splitters, and optical fiber distribution frames; optical fiber monitoring equipment is used to collect optical power attenuation data; based on a segmented loss model, an extended Kalman filter, and the optical power attenuation data, an iterative nearest point algorithm is used to perform parameter correction on the three-dimensional digital twin model.

[0052] Specifically, this application integrates BIM (Building Information Modeling) and GIS (Geographic Information System) to construct an initial three-dimensional digital twin model including optical fiber routing, splitters, and ODF (Optical Distribution Frame), and integrates the network topology data of each operator; obtains community terrain data through drone LiDAR (Light Detection And Ranging) scanning, and combines it with the optical fiber completion drawings provided by the operator to construct a three-dimensional digital twin model including optical fiber routing, splitter location, and fusion point coordinates; deploys optical fiber monitoring equipment, such as OBR (Optical Frequency Domain Reflectometer), to collect optical power attenuation data in real time, and correct the optical fiber loss parameters in the three-dimensional digital twin model to ensure that the model error is <5%.

[0053] The physical layer mapping of model calibration includes the construction of a segmented loss model and a dynamic calibration mechanism:

[0054] (1) Constructing a segment loss model: Based on the optical fiber transmission theory, a segment loss model is established:

[0055] ;

[0056] in: is the attenuation coefficient of the i-th section of optical fiber, dB / km, ranging from 0.15 to 0.35 dB / km, is the length of the optical fiber segment, km, is the loss of the jth splice point, dB, with a typical value of 0.05-0.3dB, is the insertion loss of the kth optical splitter, dB, obtained by looking up the splitting ratio table;

[0057] (2) Dynamic calibration mechanism: Deploy OBR optical frequency domain reflectometer for real-time sampling and use EKF (Extended Kalman Filter) to update model parameters:

[0058] ;

[0059] in, , observed values The measured loss of OBR is used, and the model error is reduced to <3% through iterative correction.

[0060] To ensure that the physical device state (such as location and operating parameters) is accurately mapped to the digital twin model in terms of time series and spatial dimensions, a spatiotemporal calibration algorithm can be used:

[0061] To address the spatial deviation between LiDAR data and as-built drawings, the ICP (Iterative Closest Point) algorithm is used for registration:

[0062] ;

[0063] in, is the LiDAR point cloud coordinate, is the theoretical coordinate of the drawing, T is the rotation and translation matrix, and the optimal transformation parameters are solved by SVD (Singular value decomposition) to ensure that the spatial error is less than 0.5m. is the optimal transformation matrix.

[0064] In addition, it is also possible to develop a real-time data acquisition interface to synchronize physical layer indicators such as OLT (Optical Line Terminal) optical power and ONU (Optical Network Unit) online rate, achieving second-level synchronization between the digital twin model and the physical network.

[0065] Step S12: Obtain open source library data and operation and maintenance data of multiple operators, and generate a cross-operator network asset database based on the open source library data and the operation and maintenance data.

[0066] In this embodiment, point cloud data is obtained from an open source library for processing 3D point cloud data, and the point cloud data is denoised, segmented, and the fiber routing direction is extracted to obtain open source library data; operation and maintenance data is collected from the operator's network element management system using a standardized interface; the operation and maintenance data includes the port status of the optical line terminal, the online rate of the optical network unit, and the service traffic.

[0067] This application develops a standardized interface to access EMS (Network Element Management System) data from various operators, including OLT (Optical Line Terminal) port status, ONU (Optical Network Unit) online rate, service traffic, etc., to form a cross-operator network asset database. ETL (Extract-Transform-Load, data warehouse technology) pipelines are used to integrate multi-source data, including:

[0068] (1) Spatial data: UAV LiDAR point cloud data (accuracy 0.1m) is denoised and segmented using PCL (Point Cloud Library, open source library) to extract the fiber routing direction; the BIM model is imported into Revit API (Revit Application Programming Interface, Revit software application programming interface) to analyze the physical path of the fiber and the coordinates of the fusion point.

[0069] (2) Operation and maintenance data: Access to each operator's EMS system through a standardized RESTful (Internet software architecture style) interface, and use Kafka (message queue) to synchronize cross-network data. The data fields include: fiber segment ID (Identity document, identification number), length, bending radius, welding loss (dB); number of splitter cascade layers, port optical power (mW); OLT / ONU equipment model, operating temperature, and online time.

[0070] Step S13: Deploy the revised three-dimensional digital twin model to each operator respectively, so that the operator can calculate the model training parameters corresponding to the revised three-dimensional digital twin model based on its own local data.

[0071] In this embodiment, the corrected three-dimensional digital twin model is deployed to each operator respectively, so that the operator can calculate the initial model training parameters corresponding to the corrected three-dimensional digital twin model based on its own optical power anomaly records and historical fault labels. If the initial model training parameters are gradients, differential privacy is used to perform noise injection operation on the initial model training parameters to obtain the model training parameters.

[0072] To address data privacy barriers, lack of trust, and automated execution challenges under traditional models, a federated learning framework and blockchain smart contracts are employed in cross-carrier collaboration mechanisms. Each carrier maintains independent fiber network data (such as optical power, fault records, and user complaints), but this data is commercially sensitive and cannot be directly shared. Federated learning, through a "data fixed, model moving" model, jointly trains graph neural network models (such as cross-carrier load forecasting and fault location models) while preserving privacy. Because the network topologies of the various carriers are similar (including fiber segments, splitters, OLTs, and other entities), but user distribution and service data differ, Horizontal Federated Learning (FL) is suitable.

[0073] Horizontal federated learning technology architecture Figure 2As shown in the figure, the specific steps are as follows: The revised 3D digital twin model, such as the graph structure parameters of the fault location model, is distributed to each operator. Operators train the revised 3D digital twin model based on local data, such as optical power anomaly records and historical fault labels for their network. Only gradients or parameter updates, such as weight changes, are calculated, and the original data is not transmitted. Differential privacy technology is used to inject noise, such as Laplace noise, into the gradients to further protect data privacy.

[0074] Step S14: Obtain the model training parameters sent by all the operators, train the preset graph neural network model based on the model training parameters, and determine whether the trained graph neural network model converges.

[0075] In this embodiment, a federated learning algorithm and a preset resource collaboration strategy are used to aggregate the model training parameters sent by all operators, and the preset graph neural network model is trained based on the model training parameters; the preset resource collaboration strategy includes a dynamic adjustment strategy for optical resources, an optical path switching strategy, and a resource trading strategy; it is determined whether the trained graph neural network model converges; if the trained graph neural network model does not converge, the trained graph neural network model is sent to each operator so that the operator can repeat the process of calculating the model training parameters until the trained graph neural network model converges.

[0076] The graph neural network model server in this application aggregates the updates of each operator through FedAvg (federated learning algorithm):

[0077] ;

[0078] in, is the data volume of the ith operator, For their local model parameters, operators with large data volumes can contribute more to the graph neural network model.

[0079] The trained graph neural network model is returned to each operator, and the training-aggregation process is repeated until the model converges, for example, the fault location accuracy no longer improves.

[0080] The preset resource coordination strategies in this application include dynamic adjustment strategy of optical splitting resources, optical path switching strategy, and resource trading strategy:

[0081] Dynamic adjustment of optical splitting resources: Edge nodes adjust the splitting ratio based on local traffic forecasts, and the core network coordinates cross-cell resource pools for flexible allocation. For example, if a carrier's bandwidth utilization is less than 20% at night, 50% of the optical splitting resources will be automatically released to other carriers.

[0082] Optical path switching strategy: When a fiber segment fails, the SDN (Software Defined Network) controller automatically calculates alternative optical paths, prioritizes idle fibers from other operators, and reconfigures the optical paths through intelligent optical splitters, achieving 50ms-level service switching.

[0083] Resource trading mechanism: Develop blockchain smart contracts to define cross-operator fiber resource leasing rules (e.g., 100Mbps / hour = 0.5 yuan) and realize automated settlement of resource transactions.

[0084] Step S15: If the trained graph neural network model converges, the cross-operator network asset database is input into the trained graph neural network model for fault diagnosis to predict the fiber optic network fault point.

[0085] If the trained graph neural network model converges, a micro optical power sensor is used to collect optical power data of the optical fiber to achieve real-time monitoring of the optical power of multiple operators' optical fibers; the cross-operator network asset database is input into the trained graph neural network model for fault diagnosis; the graph neural network model is a model constructed based on a graph attention network and a graph convolutional neural network.

[0086] This application deploys miniature optical power sensors on optical splitters and optical fiber optic fiber (ODF) racks to collect real-time optical power data for each fiber. The abnormality threshold is dynamically adjusted, for example, to ±3dB. A graph neural network model is constructed using graph data structures, with nodes representing fiber segments, optical splitters, and OLT / ONU devices, and edges representing physical connections. The model takes optical power anomaly data, user complaint tickets, and historical fault records as input, and uses the trained GNN (Graphical Neural Network) model to predict the fault probability distribution and output the three most likely fault points (with >90% accuracy).

[0087] The GNN network architecture is an improved GAT-GCN hybrid network, which adopts a two-layer architecture of GAT (Graph Attention Network) + GCN (Graph Convolutional Networks):

[0088] GAT layer: ;

[0089] GCN layer: ;

[0090] GAT captures the attention weights between heterogeneous nodes (such as the impact of user complaints on adjacent fiber segments); GCN aggregates spatial neighborhood features and outputs a fault probability distribution. .

[0091] Step S16: Generate a fault report and a repair plan based on the optical fiber network fault point, and send the fault report and the repair plan to the corresponding operator.

[0092] In this embodiment, the model analyzes the fault point, the operator's optical fiber, and the fault report of the affected area, and automatically generates a fault report. Different fault scenarios (such as optical fiber breakage and splitter aging) are simulated in a virtual environment to evaluate the impact on the services of multiple operators. The optimal repair plan, such as the emergency repair path and spare parts list, is generated, and the fault report and repair plan are pushed to the corresponding operator's operation and maintenance system.

[0093] In addition, this application can also combine the LSTM (Long Short-Term Memory) model to predict user traffic and weather data, dynamically adjust the OLT port sleep strategy, and put 30% of the ports into sleep mode during the off-peak period at night, reducing energy consumption by 25%.

[0094] This application realizes unified modeling and collaborative operation and maintenance of optical fiber networks across operators, solving the problems of unclear responsibilities and waste of resources in traditional operation and maintenance. It adopts federated learning technology to share operation and maintenance data and train graph neural network models under the premise of protecting data privacy. The three-dimensional digital twin model has an accuracy of optical fiber segment level (meter level), which can intuitively display details such as optical fiber bending radius and fusion point loss, assist in project acceptance and maintenance, synchronize physical network status in real time, support virtual debugging (such as simulating the impact of optical path adjustment on business), and avoid actual operation risks. The GNN fault location model integrates multi-dimensional information such as optical power, user complaints, and historical data, and its positioning accuracy is 10 times higher than that of traditional OTDR (from 100 meters to meters). The automatic switching mechanism achieves fault response time of <50ms, meeting the requirements of 5G (5th Generation Mobile Communication) Technology, the fifth generation of mobile communication technology), industrial Internet and other low-latency business needs; operators calculate the model training parameters corresponding to the revised three-dimensional digital twin model based on their own local data. The graph neural network model server is responsible for global resource scheduling and training the graph neural network model to form an edge-core hybrid decision-making mechanism, taking into account both real-time and global optimization; SDN technology is introduced to achieve programmable configuration of splitters and OLT ports, supporting resource adjustment in seconds; dynamic sleep strategy combined with renewable energy management reduces equipment energy consumption by more than 25%, in line with the "dual carbon" policy orientation; through digital twin simulation, optical fiber routing is optimized, the length of new optical fiber is reduced by 15%, and material consumption and construction costs are reduced.

[0095] In this embodiment, a three-dimensional digital twin model is constructed, and parameters of the three-dimensional digital twin model are corrected to obtain a corrected three-dimensional digital twin model; open source library data and operation and maintenance data of multiple operators are obtained, and a cross-operator network asset database is generated based on the open source library data and the operation and maintenance data; the corrected three-dimensional digital twin model is deployed to each operator respectively, so that the operator can calculate the model training parameters corresponding to the corrected three-dimensional digital twin model based on its own local data; the model training parameters sent by all the operators are obtained, and the preset graph neural network model is trained based on the model training parameters to determine whether the trained graph neural network model converges; if the trained graph neural network model converges, the cross-operator network asset database is input into the trained graph neural network model for fault diagnosis to predict the fiber optic network fault point; a fault report and a repair plan are generated based on the fiber optic network fault point, and the fault report and the repair plan are sent to the corresponding operator. This application constructs a three-dimensional digital twin model that can intuitively display details such as optical fiber bending radius and fusion point loss, assist in project acceptance and maintenance, synchronize physical network status in real time, support virtual debugging, avoid actual operation risks, obtain open source library data and operation and maintenance data of multiple operators, and generate a cross-operator network asset database based on open source library data and operation and maintenance data to achieve unified modeling and collaborative operation and maintenance of cross-operator optical fiber networks, solving the problems of unclear responsibilities and waste of resources in traditional operation and maintenance. The revised three-dimensional digital twin model is deployed to each operator separately so that the operator can calculate the model training parameters corresponding to the revised three-dimensional digital twin model based on its own local data, and train the preset graph neural network model based on the model training parameters calculated by the operator. If the trained graph neural network model converges, the cross-operator network asset database is input into the trained graph neural network model for fault diagnosis, forming an edge-core hybrid decision-making mechanism, that is, the three-dimensional digital twin model is deployed to the operator, and the graph neural network model is trained and predicted, taking into account real-time and global optimization, improving the accuracy of multi-operator optical fiber network fault diagnosis, and improving cross-operator collaboration efficiency and user experience.

[0096] See also Figure 3 As shown, the embodiment of the present invention discloses a multi-operator optical fiber network fault diagnosis device, which may specifically include:

[0097] The model construction and correction module 11 is used to construct a three-dimensional digital twin model and perform parameter correction on the three-dimensional digital twin model to obtain a corrected three-dimensional digital twin model;

[0098] A database generation module 12 is configured to obtain open source library data and operation and maintenance data of multiple operators, and generate a cross-operator network asset database based on the open source library data and the operation and maintenance data;

[0099] A model deployment module 13 is configured to deploy the modified three-dimensional digital twin model to each operator, so that the operator can calculate model training parameters corresponding to the modified three-dimensional digital twin model based on its own local data;

[0100] A model training module 14 is configured to obtain the model training parameters sent by all the operators, train a preset graph neural network model based on the model training parameters, and determine whether the trained graph neural network model has converged;

[0101] A fault diagnosis module 15 is configured to input the cross-operator network asset database into the trained graph neural network model for fault diagnosis if the trained graph neural network model converges, so as to predict the optical fiber network fault point;

[0102] The report and solution sending module 16 is configured to generate a fault report and a repair solution based on the optical fiber network fault point, and send the fault report and the repair solution to a corresponding operator.

[0103] In this embodiment, a three-dimensional digital twin model is constructed, and parameters of the three-dimensional digital twin model are corrected to obtain a corrected three-dimensional digital twin model; open source library data and operation and maintenance data of multiple operators are obtained, and a cross-operator network asset database is generated based on the open source library data and the operation and maintenance data; the corrected three-dimensional digital twin model is deployed to each operator respectively, so that the operator can calculate the model training parameters corresponding to the corrected three-dimensional digital twin model based on its own local data; the model training parameters sent by all the operators are obtained, and the preset graph neural network model is trained based on the model training parameters to determine whether the trained graph neural network model converges; if the trained graph neural network model converges, the cross-operator network asset database is input into the trained graph neural network model for fault diagnosis to predict the fiber optic network fault point; a fault report and a repair plan are generated based on the fiber optic network fault point, and the fault report and the repair plan are sent to the corresponding operator. This application constructs a three-dimensional digital twin model that can intuitively display details such as optical fiber bending radius and fusion point loss, assist in project acceptance and maintenance, synchronize physical network status in real time, support virtual debugging, avoid actual operation risks, obtain open source library data and operation and maintenance data of multiple operators, and generate a cross-operator network asset database based on open source library data and operation and maintenance data to achieve unified modeling and collaborative operation and maintenance of cross-operator optical fiber networks, solving the problems of unclear responsibilities and waste of resources in traditional operation and maintenance. The revised three-dimensional digital twin model is deployed to each operator separately so that the operator can calculate the model training parameters corresponding to the revised three-dimensional digital twin model based on its own local data, and train the preset graph neural network model based on the model training parameters calculated by the operator. If the trained graph neural network model converges, the cross-operator network asset database is input into the trained graph neural network model for fault diagnosis, forming an edge-core hybrid decision-making mechanism, that is, the three-dimensional digital twin model is deployed to the operator, and the graph neural network model is trained and predicted, taking into account real-time and global optimization, improving the accuracy of multi-operator optical fiber network fault diagnosis, and improving cross-operator collaboration efficiency and user experience.

[0104] In some specific embodiments, the model building and correction module 11 may specifically include:

[0105] An initial 3D digital twin model construction module, used to construct an initial 3D digital twin model using a building information model and a geographic information system;

[0106] The three-dimensional digital twin model generation module is used to combine the community terrain data and optical fiber completion drawings collected by the drone with the initial three-dimensional digital twin model to obtain a three-dimensional digital twin model including optical fiber routing, splitters, and optical fiber distribution frames.

[0107] In some specific embodiments, the model building and correction module 11 may specifically include:

[0108] Optical power attenuation data acquisition module, used to collect optical power attenuation data using optical fiber monitoring equipment;

[0109] A parameter correction module is used to correct the parameters of the three-dimensional digital twin model based on the segmented loss model, the extended Kalman filter and the optical power attenuation data, and using an iterative nearest point algorithm.

[0110] In some specific embodiments, the database generation module 12 may specifically include:

[0111] An open source library data acquisition module is used to obtain point cloud data from an open source library for processing 3D point cloud data, and to perform denoising, segmentation, and fiber routing extraction on the point cloud data to obtain open source library data;

[0112] The operation and maintenance data collection module is used to collect operation and maintenance data from the operator's network element management system using a standardized interface; the operation and maintenance data includes the port status of the optical line terminal, the online rate of the optical network unit, and the service flow.

[0113] In some specific embodiments, the model deployment module 13 may specifically include:

[0114] The deployment module is used to deploy the revised three-dimensional digital twin model to each operator so that the operator can calculate the initial model training parameters corresponding to the revised three-dimensional digital twin model based on its own optical power anomaly records and historical fault labels. If the initial model training parameters are gradients, differential privacy is used to perform noise injection on the initial model training parameters to obtain the model training parameters.

[0115] In some specific embodiments, the model training module 14 may specifically include:

[0116] An aggregation module is configured to aggregate the model training parameters sent by all operators using a federated learning algorithm and a preset resource collaboration strategy, and train a preset graph neural network model based on the model training parameters; the preset resource collaboration strategy includes a dynamic adjustment strategy for splitting resources, an optical path switching strategy, and a resource trading strategy;

[0117] A judgment module, used to judge whether the graph neural network model has converged after training;

[0118] The model convergence module is used to send the trained graph neural network model to each operator if it does not converge, so that the operator can repeat the process of calculating the model training parameters until the trained graph neural network model converges.

[0119] In some specific embodiments, the fault diagnosis module 15 may specifically include:

[0120] The optical power real-time monitoring module is used to collect optical power data of optical fibers using a micro optical power sensor to achieve real-time monitoring of optical power of multiple operators' optical fibers;

[0121] A fault diagnosis module is used to input the cross-operator network asset database into a trained graph neural network model for fault diagnosis; the graph neural network model is a model built based on a graph attention network and a graph convolutional neural network.

[0122] Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. The electronic device 20 may specifically include: at least one processor 21, at least one memory 22, a power supply 23, a communication interface 24, an input / output interface 25, and a communication bus 26. The memory 22 is used to store a computer program, which is loaded and executed by the processor 21 to implement the relevant steps of the multi-operator optical fiber network fault diagnosis method performed by the electronic device as disclosed in any of the aforementioned embodiments.

[0123] In this embodiment, the power supply 23 is used to provide operating voltage for each hardware device on the electronic device 20; the communication interface 24 can create a data transmission channel between the electronic device 20 and the external device. The communication protocol it follows is any communication protocol that can be applied to the technical solution of this application and is not specifically limited here; the input and output interface 25 is used to obtain external input data or output data to the outside world. Its specific interface type can be selected according to specific application needs and is not specifically limited here.

[0124] In addition, the memory 22, as a carrier for resource storage, can be a read-only memory, random access memory, disk or CD, etc. The resources stored thereon include an operating system 221, a computer program 222 and data 223, etc. The storage method can be temporary storage or permanent storage.

[0125] The operating system 221 is used to manage and control the hardware devices and computer program 222 on the electronic device 20, so as to enable the processor 21 to calculate and process the data 223 in the memory 22. The operating system 221 can be Windows, Unix, Linux, etc. In addition to including computer programs capable of performing the multi-operator fiber optic network fault diagnosis method performed by the electronic device 20 as disclosed in any of the aforementioned embodiments, the computer program 222 can further include computer programs capable of performing other specific tasks. In addition to including data transmitted from external devices and received by the multi-operator fiber optic network fault diagnosis device, the data 223 can also include data collected by its own input and output interface 25.

[0126] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein may be implemented directly using hardware, a software module executed by a processor, or a combination of the two. The software module may be placed in random access memory (RAM), internal memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.

[0127] Furthermore, an embodiment of the present application also discloses a computer-readable storage medium, in which a computer program is stored. When the computer program is loaded and executed by a processor, the steps of the multi-operator optical fiber network fault diagnosis method disclosed in any of the aforementioned embodiments are implemented.

[0128] Finally, it should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or device. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of additional identical elements in the process, method, article, or device comprising the element.

[0129] The above is a detailed introduction to the multi-operator fiber optic network fault diagnosis method, device, equipment and storage medium provided by the present invention. Specific examples are used in this article to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core ideas. At the same time, for those skilled in the art, according to the ideas of the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. A multi-operator optical fiber network fault diagnosis method, characterized in that: include: Constructing a three-dimensional digital twin model, and performing parameter correction on the three-dimensional digital twin model to obtain a corrected three-dimensional digital twin model; Obtaining open source library data and operation and maintenance data of multiple operators, and generating a cross-operator network asset database based on the open source library data and the operation and maintenance data; Deploy the revised three-dimensional digital twin model to each operator, so that the operator can calculate the model training parameters corresponding to the revised three-dimensional digital twin model based on its own local data; Obtaining the model training parameters sent by all the operators, training a preset graph neural network model based on the model training parameters, and determining whether the trained graph neural network model has converged; If the trained graph neural network model converges, the cross-operator network asset database is input into the trained graph neural network model for fault diagnosis to predict the optical fiber network fault point; A fault report and a repair plan are generated based on the optical fiber network fault point, and the fault report and the repair plan are sent to the corresponding operator.

2. The multi-operator optical fiber network fault diagnosis method according to claim 1, characterized in that: The construction of the three-dimensional digital twin model includes: Build an initial 3D digital twin model using building information modeling and geographic information systems; The community terrain data and fiber optic completion drawings collected by the drone are combined with the initial three-dimensional digital twin model to obtain a three-dimensional digital twin model including fiber optic routing, splitters, and fiber optic distribution frames.

3. The multi-operator optical fiber network fault diagnosis method according to claim 1, characterized in that: The parameter correction of the three-dimensional digital twin model includes: Use fiber optic monitoring equipment to collect optical power attenuation data; Based on the segmented loss model, the extended Kalman filter and the optical power attenuation data, the parameters of the three-dimensional digital twin model are corrected by using an iterative nearest point algorithm.

4. The multi-operator optical fiber network fault diagnosis method according to claim 1, characterized in that: The acquisition of open source library data and operation and maintenance data of multiple operators includes: Obtaining point cloud data from an open source library for processing 3D point cloud data, and performing denoising, segmentation, and fiber routing extraction on the point cloud data to obtain open source library data; The operation and maintenance data is collected from the operator's network element management system using a standardized interface; the operation and maintenance data includes the port status of the optical line terminal, the online rate of the optical network unit, and the service flow.

5. The multi-operator optical fiber network fault diagnosis method according to claim 1, characterized in that: The modified three-dimensional digital twin model is deployed to each operator respectively so that the operator can calculate the model training parameters corresponding to the modified three-dimensional digital twin model based on its own local data, including: The revised three-dimensional digital twin model is deployed to each operator respectively, so that the operator can calculate the initial model training parameters corresponding to the revised three-dimensional digital twin model based on its own optical power anomaly records and historical fault labels. If the initial model training parameters are gradients, differential privacy is used to perform noise injection on the initial model training parameters to obtain the model training parameters.

6. The multi-operator optical fiber network fault diagnosis method according to claim 1, characterized in that: The obtaining of the model training parameters sent by all the operators, training a preset graph neural network model based on the model training parameters, and determining whether the trained graph neural network model has converged includes: The model training parameters sent by all operators are aggregated using a federated learning algorithm and a preset resource collaboration strategy, and a preset graph neural network model is trained based on the model training parameters; the preset resource collaboration strategy includes a dynamic adjustment strategy for splitting resources, an optical path switching strategy, and a resource trading strategy; Determining whether the trained graph neural network model has converged; If the trained graph neural network model does not converge, the trained graph neural network model will be sent to each operator so that the operator can repeat the process of calculating the model training parameters until the trained graph neural network model converges.

7. The multi-operator optical fiber network fault diagnosis method according to any one of claims 1 to 6, characterized in that: Inputting the cross-operator network asset database into the trained graph neural network model for fault diagnosis includes: Use micro optical power sensors to collect optical power data from optical fibers to achieve real-time monitoring of optical power across multiple operators. The cross-operator network asset database is input into the trained graph neural network model for fault diagnosis; the graph neural network model is a model built based on the graph attention network and the graph convolutional neural network.

8. A multi-operator optical fiber network fault diagnosis device, characterized in that: include: A model construction and correction module is used to construct a three-dimensional digital twin model and perform parameter correction on the three-dimensional digital twin model to obtain a corrected three-dimensional digital twin model; A database generation module, configured to obtain open source library data and operation and maintenance data of multiple operators, and generate a cross-operator network asset database based on the open source library data and the operation and maintenance data; A model deployment module, configured to deploy the modified three-dimensional digital twin model to each operator, so that the operator can calculate model training parameters corresponding to the modified three-dimensional digital twin model based on its own local data; A model training module, configured to obtain the model training parameters sent by all the operators, train a preset graph neural network model based on the model training parameters, and determine whether the trained graph neural network model has converged; A fault diagnosis module is configured to input the cross-operator network asset database into the trained graph neural network model for fault diagnosis if the trained graph neural network model converges, so as to predict the optical fiber network fault point; The report and solution sending module is used to generate a fault report and a repair solution based on the optical fiber network fault point, and send the fault report and the repair solution to the corresponding operator.

9. An electronic device, characterized in that: include: Memory, used to store computer programs; A processor, configured to execute the computer program to implement the multi-operator optical fiber network fault diagnosis method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that Used to store a computer program; wherein, when the computer program is executed by a processor, the multi-operator optical fiber network fault diagnosis method according to any one of claims 1 to 7 is implemented.

Images