Wireless network fault positioning method and device, electronic equipment and storage medium

Abstract

The invention provides a wireless network fault positioning method and device, electronic equipment and a storage medium. The method comprises the following steps: acquiring equipment attribute data, topological structure data and alarm time sequence data of a wireless network; processing the equipment attribute data, the topological structure data and the alarm time sequence data to obtain an embedded vector of each node of the wireless network; inputting the embedded vector of each node into a graph neural network model to obtain a fault root cause positioning result of the wireless network output by the graph neural network model; the graph neural network model is obtained by training based on embedded vector samples of node samples of the wireless network samples and fault root cause positioning result labels of the wireless network samples. According to the method, the normalized node embedding vector is obtained by fusing the multi-source heterogeneous data of the wireless network, and the fault root cause of the wireless network is positioned based on the node embedding vector by adopting the graph neural network model, so that the fault positioning efficiency and accuracy are improved.

Description

Technical Field

[0001] This invention relates to the field of network operation and maintenance technology, and in particular to a method, apparatus, electronic device and storage medium for locating wireless network faults. Background Technology

[0002] With the rapid development of wireless communication technology, the scale and complexity of wireless networks are constantly increasing. In such vast networks, faults are inevitable, and how to quickly and accurately locate these faults is a critical issue that urgently needs to be addressed. Traditional wireless network fault location methods have many limitations: on the one hand, most traditional methods rely on simple statistical analysis based on human experience and preset rules, making it difficult to deeply mine and effectively characterize complex network data; on the other hand, traditional methods only consider a single characteristic of the wireless network, resulting in low fault location accuracy and efficiency, failing to meet the current needs of wireless network operation and maintenance. Summary of the Invention

[0003] This invention provides a method, apparatus, electronic device, and storage medium for locating wireless network faults, in order to overcome the shortcomings of low efficiency and poor accuracy in existing wireless network fault location methods.

[0004] This invention provides a method for locating wireless network faults, comprising: Acquire device attribute data, topology data, and alarm timing data of the wireless network; The device attribute data, topology data, and alarm timing data are processed to obtain the embedding vector of each node in the wireless network. The embedding vectors of each node are input into a graph neural network model to obtain the root cause localization result of the wireless network fault output by the graph neural network model. The graph neural network model is trained based on the embedding vector samples of each node sample of the wireless network sample and the labels of the root cause localization result of the wireless network sample.

[0005] In some embodiments, processing the device attribute data, topology data, and alarm timing data to obtain the embedding vectors of each node in the wireless network includes: The device attribute data, the topology data, and the alarm timing data are input into the Embedding model to obtain the embedding vector of each node output by the Embedding model. The Embedding model is trained based on device attribute data samples, topology data samples, and alarm timing data samples of the wireless network samples, as well as the embedding vector labels of each node sample of the wireless network samples.

[0006] In some embodiments, the embedding model includes: An attribute embedding layer is used to encode the device attribute data to obtain the attribute embedding vector of each node; A structure embedding layer is used to encode the topology data to obtain the structure embedding vector of each node; An alarm embedding layer is used to encode the alarm timing data to obtain the alarm embedding vector for each node. The spatiotemporal fusion layer is used to fuse the attribute embedding vector, structure embedding vector and alarm embedding vector of each node based on the attention mechanism to obtain the embedding vector of each node.

[0007] In some embodiments, the graph neural network model includes: The graph attention layer is used to dynamically calculate the attention weights between each node and its corresponding neighboring nodes, and based on the attention weights, update the embedding vectors of each node to obtain the feature vectors of each node. A dilated temporal convolutional layer is used to locate the root cause of the fault in the wireless network based on the feature vectors of each node, and to obtain the root cause location result.

[0008] In some embodiments, after obtaining the root cause localization result of the wireless network output by the graph neural network model, the method further includes: The results of the fault root cause localization are displayed visually; The root cause location results of the fault are sent to the client.

[0009] In some embodiments, after obtaining the root cause localization result of the wireless network output by the graph neural network model, the method further includes: After performing fault handling on the wireless network based on the fault root cause location results, obtain fault handling feedback information; Determine the first performance evaluation result of the Embedding model and the second performance evaluation result of the graph neural network model; Based on the fault handling feedback information and the first performance evaluation result, the parameters of the Embedding model are updated, and based on the fault handling feedback information and the second performance evaluation result, the parameters of the graph neural network model are updated.

[0010] In some embodiments, the graph neural network model is trained based on the following steps: Obtain device attribute data samples, topology data samples, and alarm timing data samples of wireless network samples, and determine the fault root cause location result label of the wireless network samples; The device attribute data sample, topology data sample, and alarm timing data sample are processed to obtain the embedding vector sample of each node sample of the wireless network sample. The embedding vector samples of each node sample are input into the initial graph neural network model to obtain the predicted fault root cause localization result of the wireless network sample output by the initial graph neural network model. Based on the predicted root cause localization result and the label of the root cause localization result, a loss function value is calculated. Based on the loss function value, the parameters of the initial graph neural network model are iteratively optimized to obtain the graph neural network model.

[0011] The present invention also provides a wireless network fault location device, comprising: The acquisition unit is used to acquire device attribute data, topology data, and alarm timing data of the wireless network. The processing unit is used to process the device attribute data, topology data and alarm timing data to obtain the embedding vector of each node of the wireless network. The fault location unit is used to input the embedding vectors of each node into the graph neural network model to obtain the fault root cause location result of the wireless network output by the graph neural network model; the graph neural network model is trained based on the embedding vector samples of each node sample of the wireless network sample and the fault root cause location result label of the wireless network sample.

[0012] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement any of the wireless network fault location methods described above.

[0013] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the wireless network fault location method as described above.

[0014] The wireless network fault location method, apparatus, electronic device, and storage medium provided by this invention acquire device attribute data, topology data, and alarm timing data of the wireless network; process the device attribute data, topology data, and alarm timing data to obtain the embedding vector of each node in the wireless network; input the embedding vector of each node into a graph neural network model to obtain the root cause location result of the wireless network fault output by the graph neural network model, thereby improving the efficiency and accuracy of fault location. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0016] Figure 1 This is a flowchart illustrating the wireless network fault location method provided in an embodiment of the present invention.

[0017] Figure 2 This is a flowchart illustrating the training process of the graph neural network model provided in an embodiment of the present invention.

[0018] Figure 3 This is a schematic diagram of the wireless network fault location device provided in an embodiment of the present invention.

[0019] Figure 4 This is a schematic diagram of the structure of the electronic device provided in an embodiment of the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0021] The terms "first," "second," etc., used in this invention are used to distinguish similar objects, not to describe a specific order or sequence. It should be understood that such terms can be used interchangeably where appropriate so that embodiments of the invention can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first" and "second" are generally of the same class, not limited in number; for example, the first object can be one or more. Furthermore, in this invention, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0022] Figure 1 This is a flowchart illustrating a wireless network fault location method provided in an embodiment of the present invention. Figure 1 As shown, a method for locating wireless network faults is provided, including the following steps: step 110, step 120, and step 130. These method steps are merely one possible implementation of the present invention.

[0023] Step 110: Obtain device attribute data, topology data, and alarm timing data of the wireless network.

[0024] Optionally, a wireless network may include at least equipment such as base stations, terminals, and core networks.

[0025] Optionally, the device attribute data includes at least the identification information, hardware attributes, configuration attributes, and performance specifications of each device in the wireless network; the topology data includes at least the relationships between the devices, which can be topology graph data, with each device as a node and the relationships between the devices as edges; and the alarm timing data includes at least the timestamp, alarm device, alarm type, and alarm details.

[0026] Step 120: Process the device attribute data, topology data, and alarm timing data to obtain the embedding vectors of each node in the wireless network.

[0027] Optionally, the device attribute data, topology data, and alarm time series data are cleaned, normalized, and feature extracted to obtain the embedding vector of each node.

[0028] In some embodiments, device attribute data, topology data, and alarm timing data are processed to obtain the embedding vectors of each node in the wireless network, including: Input device attribute data, topology data, and alarm timing data into the Embedding model to obtain the embedding vector of each node output by the Embedding model; The Embedding model is trained based on device attribute data samples, topology data samples, and alarm timing data samples of wireless network samples, as well as the embedding vector labels of each node sample of the wireless network samples.

[0029] Optionally, the embedding model includes a Long Short-Term Memory Network (LSTM) and a Graph Sample and Aggregate (GraphSAGE) network.

[0030] Optionally, a device attribute matrix is ​​constructed based on device attribute data, an adjacency matrix is ​​constructed based on topology data, and a time-series alarm sequence is constructed based on alarm time-series data.

[0031] Optionally, the expression for the embedding vector of each node is as follows: ; in, This represents the embedding vector of each node. This represents the activation function. Represents the device attribute matrix. Represents the adjacency matrix. Indicates the timing alarm sequence. .

[0032] In some embodiments, the embedding model includes: The attribute embedding layer is used to encode device attribute data to obtain the attribute embedding vector of each node. The structure embedding layer is used to encode the topological data to obtain the structure embedding vector of each node; The alarm embedding layer is used to encode alarm time-series data to obtain the alarm embedding vector of each node. The spatiotemporal fusion layer is used to fuse the attribute embedding vectors, structure embedding vectors, and alarm embedding vectors of each node based on the attention mechanism to obtain the embedding vector of each node.

[0033] Optionally, the calculation formula for the attribute embedding layer is as follows: ; in, Represents a node v The attribute embedding vector, Represents a node v Numerical attributes, Represents a node v Category attributes, Indicates splicing, Indicates quantile encoding, Represents the embedding matrix. This represents the weight matrix.

[0034] Optionally, the calculation formula for the structural embedding layer is as follows: ; ; in, Represents a node v With nodes u Attention coefficient Represents a node v The structural embedding vector, It is a node u The attribute embedding vector, It is a node v The attribute embedding vector, Represents a non-linear activation function. Represents the attention vector. This indicates a shared linear transformation weight matrix. This represents a non-linear activation function.

[0035] Optionally, the calculation formula for the spatiotemporal fusion layer is as follows: ; in, Represents a node v Embedded vector, Represents a node v The alarm embedding vector, Representation layer normalization, Represents a node v Alarm timing data.

[0036] Optionally, during the training of the embedding model, the total loss function value is calculated based on the total loss function, and the parameters of the initial embedding model are iteratively optimized based on the total loss function value. The formula for calculating the total loss function value is as follows: ; in, This represents the total loss function value. This represents the value of the autoencoder reconstruction loss function. This represents the value of the topological similarity loss function. The weights for the values ​​of the autoencoder reconstruction loss function. The weights are the values ​​of the topological similarity loss function.

[0037] Step 130: Input the embedding vectors of each node into the graph neural network model to obtain the root cause localization result of the wireless network fault output by the graph neural network model; the graph neural network model is trained based on the embedding vector samples of each node sample of the wireless network sample and the labels of the root cause localization result of the wireless network sample.

[0038] Optionally, the root cause localization results should include at least: basic information about the fault, possible root causes, scope of impact, and repair recommendations.

[0039] Optionally, the graph neural network model includes at least a spatial layer, a temporal layer, and an output layer.

[0040] Optionally, the spatial layer uses a Graph Attention Network (GAT) to learn the fault impact weights between nodes, calculated as follows: ; in, Represents a node i With nodes j The attention weights between them, i.e., the fault impact weights, Represents a node i Embedded vector, Represents a node jEmbedded vector, This represents the normalized exponential function.

[0041] Optionally, the temporal layer captures long-period alarm patterns by employing a dilated convolutional neural network (Dilated CNN).

[0042] In some embodiments, the graph neural network model includes: The graph attention layer is used to dynamically calculate the attention weights between each node and its corresponding neighboring nodes. Based on the attention weights, the embedding vectors of each node are updated to obtain the feature vectors of each node. The dilated temporal convolutional layer is used to locate the root cause of faults in wireless networks based on the feature vectors of each node, and obtain the root cause location results.

[0043] Optionally, a spatiotemporal graph G is constructed based on a graph attention layer. E is the edge set, and W is the attention weight.

[0044] Optionally, the dilated temporal convolutional layer includes at least a spatial graph convolutional layer and a temporal convolutional layer, calculated as follows: ; in, These are the temporal convolution kernels, corresponding to the input and output, respectively. G is the spatial graph convolution kernel, and G is the graph convolution operation. This is a temporal convolution operation.

[0045] In this embodiment of the invention, device attribute data, topology data, and alarm timing data of the wireless network are acquired; the device attribute data, topology data, and alarm timing data are processed to obtain the embedding vector of each node in the wireless network; the embedding vector of each node is input into a graph neural network model to obtain the root cause localization result of the wireless network fault output by the graph neural network model, thereby improving the efficiency, accuracy, and interpretability of fault localization.

[0046] In some embodiments, after obtaining the root cause localization results of the wireless network faults output by the graph neural network model, the method further includes: Visualize the results of fault root cause localization. Send the root cause analysis results to the client.

[0047] Optionally, the root cause localization results can be presented to users in the form of intuitive charts, maps, topology diagrams, etc., which can help users quickly understand the overall picture and details of network faults and improve user experience.

[0048] Optionally, the root cause analysis results can be output to a specified storage location or sent to operations and maintenance personnel to provide a basis for subsequent fault handling and network optimization.

[0049] In some embodiments, after obtaining the root cause localization results of the wireless network faults output by the graph neural network model, the method further includes: After handling the wireless network based on the root cause location results, obtain the fault handling feedback information; Determine the first performance evaluation result of the Embedding model and the second performance evaluation result of the graph neural network model; Based on the fault handling feedback information and the first performance evaluation results, the parameters of the Embedding model are updated, and based on the fault handling feedback information and the second performance evaluation results, the parameters of the graph neural network model are updated.

[0050] Optionally, the fault handling feedback information shall include at least the repair time, repair effect, and repair evaluation.

[0051] Optionally, the first performance index data of the Embedding model is obtained, and a first performance evaluation result is obtained based on the first performance index data; the second performance index data of the graph neural network model is obtained, and a second performance evaluation result is obtained based on the second performance index data.

[0052] Understandably, updating the parameters of the Embedding model based on fault handling feedback information and the first performance evaluation results, and updating the parameters of the graph neural network model based on fault handling feedback information and the second performance evaluation results, can continuously improve the fault location accuracy and adaptability of the system, and better cope with various fault problems in wireless networks.

[0053] Figure 2 This is a schematic flowchart illustrating the training process of a graph neural network model provided in an embodiment of the present invention. Figure 2 As shown, in some embodiments, the graph neural network model is trained based on the following steps: Step 210: Obtain device attribute data samples, topology data samples, and alarm timing data samples of the wireless network sample, and determine the fault root cause location result label of the wireless network sample.

[0054] Step 220: Process the device attribute data sample, topology data sample, and alarm timing data sample to obtain the embedding vector sample of each node sample of the wireless network sample.

[0055] Optionally, the device attribute data sample, topology data sample, and alarm timing data sample are input into the Embedding model to obtain the embedding vector samples of each node sample output by the Embedding model.

[0056] Step 230: Input the embedding vector samples of each node sample into the initial graph neural network model to obtain the predicted fault root cause localization results of the wireless network samples output by the initial graph neural network model.

[0057] Step 240: Based on the predicted root cause localization results and the labels of the root cause localization results, calculate the loss function value. Based on the loss function value, iteratively optimize the parameters of the initial graph neural network model to obtain the graph neural network model.

[0058] Optionally, the initial graph neural network model includes: The initial graph attention layer is used to dynamically calculate the predicted attention weights between each node sample and its corresponding neighboring node samples. Based on the predicted attention weights, the embedding vector samples of each node sample are updated to obtain the predicted feature vectors of each node sample. An initial dilated temporal convolutional layer is used to locate the root cause of faults in wireless network samples based on the predicted feature vectors of each node sample, and obtain the predicted root cause location result.

[0059] Optionally, the Sigmoid cross-entropy loss function is used to calculate the loss function value, and the calculation formula is as follows: ; in, This represents the value of the loss function. This is the predicted value of the root cause probability of the failure. This represents the actual value of the root cause probability of the fault.

[0060] Optionally, a knowledge graph can be used to store historical fault patterns for online incremental training of the model.

[0061] The wireless network fault location device provided in the embodiments of the present invention is described below. The wireless network fault location device described below can be referred to in correspondence with the wireless network fault location method described above.

[0062] Figure 3 This is a schematic diagram of the structure of the wireless network fault location device provided in an embodiment of the present invention, as shown below. Figure 3 As shown, the wireless network fault location device 300 includes: The acquisition unit 310 is used to acquire device attribute data, topology data and alarm timing data of the wireless network. The processing unit 320 is used to process device attribute data, topology data and alarm timing data to obtain the embedding vector of each node in the wireless network. The fault location unit 330 is used to input the embedding vectors of each node into the graph neural network model to obtain the fault root cause location result of the wireless network output by the graph neural network model. The graph neural network model is trained based on the embedding vector samples of each node sample of the wireless network sample and the fault root cause location result label of the wireless network sample.

[0063] Optionally, the device attribute data, topology data, and alarm timing data are processed to obtain the embedding vectors of each node in the wireless network, including: Input device attribute data, topology data, and alarm timing data into the Embedding model to obtain the embedding vector of each node output by the Embedding model; The Embedding model is trained based on device attribute data samples, topology data samples, and alarm timing data samples of wireless network samples, as well as the embedding vector labels of each node sample of the wireless network samples.

[0064] Optionally, the embedding model includes: The attribute embedding layer is used to encode device attribute data to obtain the attribute embedding vector of each node. The structure embedding layer is used to encode the topological data to obtain the structure embedding vector of each node; The alarm embedding layer is used to encode alarm time-series data to obtain the alarm embedding vector of each node. The spatiotemporal fusion layer is used to fuse the attribute embedding vectors, structure embedding vectors, and alarm embedding vectors of each node based on the attention mechanism to obtain the embedding vector of each node.

[0065] Optionally, the graph neural network model includes: The graph attention layer is used to dynamically calculate the attention weights between each node and its corresponding neighboring nodes. Based on the attention weights, the embedding vectors of each node are updated to obtain the feature vectors of each node. The dilated temporal convolutional layer is used to locate the root cause of faults in wireless networks based on the feature vectors of each node, and obtain the root cause location results.

[0066] Optionally, the wireless network fault location device further includes: The display unit is used to visualize the results of fault root cause localization. The sending unit is used to send the root cause location results of the fault to the client.

[0067] Optionally, the wireless network fault location device further includes: The feedback unit is used to obtain fault handling feedback information after the wireless network is fault-handling based on the fault root cause location results. The determination unit is used to determine the first performance evaluation result of the Embedding model and the second performance evaluation result of the graph neural network model. The model parameter update unit is used to update the parameters of the Embedding model based on the fault handling feedback information and the first performance evaluation result, and to update the parameters of the graph neural network model based on the fault handling feedback information and the second performance evaluation result.

[0068] Optionally, the graph neural network model is trained based on the following steps: Obtain device attribute data samples, topology data samples, and alarm timing data samples of wireless network samples, and determine the fault root cause location result labels of wireless network samples; The device attribute data sample, topology data sample, and alarm timing data sample are processed to obtain the embedding vector sample of each node sample of the wireless network sample. The embedding vector samples of each node are input into the initial graph neural network model to obtain the predicted fault root cause localization results of the wireless network samples output by the initial graph neural network model. Based on the predicted root cause localization results and the labels of the root cause localization results, the loss function value is calculated. Based on the loss function value, the parameters of the initial graph neural network model are iteratively optimized to obtain the graph neural network model.

[0069] It should be noted that the wireless network fault location device provided in this embodiment of the invention can implement all the method steps implemented in the above-described wireless network fault location method embodiment and can achieve the same technical effect. Here, the parts that are the same as those in the method embodiment and the beneficial effects will not be described in detail.

[0070] Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention, such as... Figure 4As shown, the electronic device may include: a processor 410, a communications interface 420, a memory 430, and a communication bus 440. The processor 410, communications interface 420, and memory 430 communicate with each other via the communication bus 440. The processor 410 can call logical instructions in the memory 430 to execute a wireless network fault location method. This method includes: acquiring device attribute data, topology data, and alarm timing data of the wireless network; processing the device attribute data, topology data, and alarm timing data to obtain the embedding vectors of each node in the wireless network; inputting the embedding vectors of each node into a graph neural network model to obtain the root cause location result of the wireless network fault output by the graph neural network model; the graph neural network model is trained based on the embedding vector samples of each node sample of the wireless network sample and the labels of the root cause location result of the wireless network sample.

[0071] Furthermore, the logical instructions in the aforementioned memory 430 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0072] In another aspect, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the wireless network fault location method provided by the above methods. The method includes: acquiring device attribute data, topology data, and alarm timing data of the wireless network; processing the device attribute data, topology data, and alarm timing data to obtain the embedding vectors of each node of the wireless network; inputting the embedding vectors of each node into a graph neural network model to obtain the fault root cause location result of the wireless network output by the graph neural network model; the graph neural network model is trained based on the embedding vector samples of each node sample of the wireless network sample and the fault root cause location result labels of the wireless network sample.

[0073] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0074] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0075] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for locating faults in a wireless network, characterized in that, include: Acquire device attribute data, topology data, and alarm timing data of the wireless network; The device attribute data, topology data, and alarm timing data are processed to obtain the embedding vector of each node in the wireless network. The embedding vectors of each node are input into a graph neural network model to obtain the root cause localization result of the wireless network fault output by the graph neural network model. The graph neural network model is trained based on the embedding vector samples of each node sample of the wireless network sample and the labels of the root cause localization result of the wireless network sample.

2. The wireless network fault location method according to claim 1, characterized in that, The process of processing the device attribute data, topology data, and alarm timing data to obtain the embedding vectors of each node in the wireless network includes: The device attribute data, the topology data, and the alarm timing data are input into the Embedding model to obtain the embedding vector of each node output by the Embedding model. The Embedding model is trained based on device attribute data samples, topology data samples, and alarm timing data samples of the wireless network samples, as well as the embedding vector labels of each node sample of the wireless network samples.

3. The wireless network fault location method according to claim 2, characterized in that, The Embedding model includes: An attribute embedding layer is used to encode the device attribute data to obtain the attribute embedding vector of each node; A structure embedding layer is used to encode the topology data to obtain the structure embedding vector of each node; An alarm embedding layer is used to encode the alarm timing data to obtain the alarm embedding vector for each node. The spatiotemporal fusion layer is used to fuse the attribute embedding vector, structure embedding vector and alarm embedding vector of each node based on the attention mechanism to obtain the embedding vector of each node.

4. The wireless network fault location method according to claim 1, characterized in that, The graph neural network model includes: The graph attention layer is used to dynamically calculate the attention weights between each node and its corresponding neighboring nodes, and based on the attention weights, update the embedding vectors of each node to obtain the feature vectors of each node. A dilated temporal convolutional layer is used to locate the root cause of the fault in the wireless network based on the feature vectors of each node, and to obtain the root cause location result.

5. The wireless network fault location method according to claim 1, characterized in that, After obtaining the root cause localization result of the wireless network output by the graph neural network model, the method further includes: The results of the fault root cause localization are displayed visually; The root cause location results of the fault are sent to the client.

6. The wireless network fault location method according to claim 2, characterized in that, After obtaining the root cause localization result of the wireless network output by the graph neural network model, the method further includes: After performing fault handling on the wireless network based on the fault root cause location results, obtain fault handling feedback information; Determine the first performance evaluation result of the Embedding model and the second performance evaluation result of the graph neural network model; Based on the fault handling feedback information and the first performance evaluation result, the parameters of the Embedding model are updated, and based on the fault handling feedback information and the second performance evaluation result, the parameters of the graph neural network model are updated.

7. The wireless network fault location method according to claim 1, characterized in that, The graph neural network model is trained based on the following steps: Obtain device attribute data samples, topology data samples, and alarm timing data samples of wireless network samples, and determine the fault root cause location result label of the wireless network samples; The device attribute data sample, topology data sample, and alarm timing data sample are processed to obtain the embedding vector sample of each node sample of the wireless network sample. The embedding vector samples of each node sample are input into the initial graph neural network model to obtain the predicted fault root cause localization result of the wireless network sample output by the initial graph neural network model. Based on the predicted root cause localization result and the label of the root cause localization result, a loss function value is calculated. Based on the loss function value, the parameters of the initial graph neural network model are iteratively optimized to obtain the graph neural network model.

8. A wireless network fault location device, characterized in that, include: The acquisition unit is used to acquire device attribute data, topology data, and alarm timing data of the wireless network. The processing unit is used to process the device attribute data, topology data and alarm timing data to obtain the embedding vector of each node of the wireless network. The fault location unit is used to input the embedding vectors of each node into the graph neural network model to obtain the fault root cause location result of the wireless network output by the graph neural network model; the graph neural network model is trained based on the embedding vector samples of each node sample of the wireless network sample and the fault root cause location result label of the wireless network sample.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the wireless network fault location method as described in any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the wireless network fault location method as described in any one of claims 1 to 7.

Images