Method and device for identifying root alarm of optical transport network based on space-time graph network

Abstract

The application provides a kind of method and device for identifying root alarm of optical transmission network based on space-time graph network, the method comprises: constructing space-time alarm subgraph according to the space-time clustering result of multiple alarm data;Multiple alarm data includes root alarm data and derived alarm data, and the space-time alarm subgraph is used to represent the alarm relationship with space-time characteristics;Identify the space-time alarm subgraph based on the space-time graph network to obtain the target alarm identification result;The space-time graph network is trained by taking multiple sample alarm data as nodes, taking the correlation between multiple sample alarm data as the edge weight parameter between different node features, and updating the edge weight parameter by sampling attention mechanism.The method of the application fully considers the space-time characteristics of alarm, improves the accuracy of root alarm identification of optical transmission network, and enhances the explainability of alarm identification result.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of network security, in particular to a root alarm identification method and device for an optical transmission network based on a space-time graph network. BACKGROUND

[0002] In an optical transmission network, interruption of any service can lead to serious service interruption and huge economic losses. In the face of complex network structure and diversified fault modes, alarm information becomes an important means of fault management, which can extract key fault information from complex network environment. Real-time and accurate alarm analysis is crucial for effective fault management, which can not only reveal abnormal state in the network in a timely manner, but also predict potential fault points.

[0003] In a complex optical transmission network system, multiple entities are connected to each other, and a fault of a single device can affect related modules, thereby triggering a large number of derivative alarms, and even leading to an alarm storm. The root alarm is usually a direct indicator of the initial fault, but it is often covered by derivative alarms generated by other devices affected by the fault. The process of finding the root alarm in sequence among numerous alarms is not only inefficient, but also prolongs the recovery time and increases the risk. Therefore, quickly and accurately locating the root alarm that triggers the chain reaction is crucial for efficient fault management.

[0004] In the related art, time stamp and time interval are usually used as indicators to measure alarm correlation, or graph theory method is used to explore the propagation path of the alarm in a specific service line. The time and space characteristics of the alarm are not fully considered, which cannot effectively express the comprehensive influence of the propagation path of the logical fault and the fault time on the root cause positioning of the optical transmission network alarm, resulting in low accuracy of root alarm identification and positioning. SUMMARY

[0005] The application provides a root alarm identification method and device for an optical transmission network based on a space-time graph network, which solves the defect that the time and space characteristics of the alarm are not fully considered in the fault root cause positioning process of the optical transmission network in the prior art, resulting in low accuracy of root alarm identification and positioning, and improves the accuracy of root alarm identification of the optical transmission network.

[0006] The application provides a root alarm identification method for an optical transmission network based on a space-time graph network, comprising:

[0007] Constructing a space-time alarm subgraph according to the space-time clustering results of a plurality of alarm data; wherein the plurality of alarm data comprises root alarm data and derivative alarm data, and the space-time alarm subgraph is used to represent the alarm relationship with time and space characteristics;

[0008] The target alarm identification result is obtained by identifying the time-space alarm subgraph based on a time-space graph network, wherein the target alarm identification result includes at least one of an alarm type, an edge weight parameter between different alarm identification results, and alarm positioning information; and the time-space graph network is trained by taking a plurality of sample alarm data as nodes, taking the correlation between the plurality of sample alarm data as an edge weight parameter between different node features, and updating the edge weight parameter by sampling attention mechanism.

[0009] According to the optical transmission network root alarm identification method based on the time-space graph network, the time-space alarm subgraph is constructed according to the time-space clustering result of the plurality of alarm data, including:

[0010] The plurality of alarm data are clustered according to the generation time of each alarm data, to obtain a time clustering result.

[0011] The time clustering result is spatially clustered according to the physical topology relationship between the plurality of alarm data, to obtain the time-space clustering result, and the time-space alarm subgraph is constructed according to the time-space clustering result.

[0012] According to the optical transmission network root alarm identification method based on the time-space graph network, the target alarm identification result includes a plurality of root alarm identification results.

[0013] After the target alarm identification result is obtained, the method further includes:

[0014] The target alarm identification result is screened according to the edge weight parameter corresponding to each root alarm identification result, to obtain a first root alarm identification result.

[0015] According to the optical transmission network root alarm identification method based on the time-space graph network, the target alarm identification result further includes a plurality of derived alarm identification results.

[0016] After the first root alarm identification result is obtained, the method further includes:

[0017] The target alarm identification result is screened according to the edge weight parameter corresponding to the first root alarm identification result and the edge weight parameter corresponding to each derived alarm identification result, to obtain a second root alarm identification result.

[0018] According to the optical transmission network root alarm identification method based on the time-space graph network, the target alarm identification result is an alarm identification subgraph; wherein the alarm identification subgraph includes an alarm type, an edge weight value between different alarm identification results, and a physical topology relationship of each alarm identification result.

[0019] The application further provides an optical transmission network root alarm identification device based on a time-space graph network, including:

[0020] a spatio-temporal clustering and composition module, configured to construct a spatio-temporal alarm subgraph according to a spatio-temporal clustering result of a plurality of alarm data, wherein the plurality of alarm data comprises root alarm data and derived alarm data, and the spatio-temporal alarm subgraph is used to represent an alarm relationship with spatio-temporal characteristics;

[0021] a recognition module, configured to recognize the spatio-temporal alarm subgraph based on a spatio-temporal graph network to obtain a target alarm recognition result, wherein the target alarm recognition result comprises at least one of an alarm type, an edge weight parameter between different alarm recognition results, and alarm positioning information, and the spatio-temporal graph network is obtained by training in a manner of taking a plurality of sample alarm data as nodes, taking a correlation between the plurality of sample alarm data as an edge weight parameter between different node features, and updating the edge weight parameter by sampling an attention mechanism.

[0022] According to the optical transmission network root alarm recognition method and device based on the spatio-temporal graph network, the target alarm recognition result comprises a plurality of root alarm recognition results.

[0023] The device further comprises:

[0024] a screening module, configured to, after obtaining the target alarm recognition result, screen the target alarm recognition result according to the edge weight parameter corresponding to each root alarm recognition result to obtain a first root alarm recognition result.

[0025] The application further provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the optical transmission network root alarm recognition method based on the spatio-temporal graph network as described above when executing the computer program.

[0026] The application further provides a non-transitory computer-readable storage medium having a computer program stored thereon, wherein the computer program is executable by a processor to implement the optical transmission network root alarm recognition method based on the spatio-temporal graph network as described above.

[0027] The application further provides a computer program product comprising a computer program, wherein the computer program is executable by a processor to implement the optical transmission network root alarm recognition method based on the spatio-temporal graph network as described above.

[0028] The optical transmission network root alarm recognition method and device based on the spatio-temporal graph network provided by the application obtain alarm types, edge weight parameters between different alarm recognition results, or alarm positioning information and other recognition results by sequentially performing time clustering and space clustering on a plurality of alarm data, fully consider the spatio-temporal characteristics of alarms, improve the accuracy of optical transmission network root alarm recognition, and enhance the explainability of alarm recognition results. BRIEF DESCRIPTION OF DRAWINGS

[0029] In order to more clearly illustrate the technical solutions of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings described below are only some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without creative effort.

[0030] Figure 1 Figure is one of the flow diagrams of the optical transmission network root alarm identification method based on the space-time graph network provided by the present application.

[0031] Figure 2 Figure is another of the flow diagrams of the optical transmission network root alarm identification method based on the space-time graph network provided by the present application.

[0032] Figure 3 Figure is a schematic diagram of the topological relationship between the network elements and the corresponding edges in the space-time alarm subgraph.

[0033] Figure 4 Figure is a schematic diagram of the structure of the optical transmission network root alarm identification device based on the space-time graph network provided by the present application.

[0034] Figure 5 Figure is a schematic diagram of the structure of the electronic device provided by the present application. DETAILED DESCRIPTION

[0035] In order to make the objects, technical solutions and advantages of the present application clearer, the technical solutions of the present application will be described clearly and completely below with reference to the drawings in the present application. Obviously, the described embodiments are only some of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present application.

[0036] The following will describe the optical transmission network root alarm identification method and device based on the space-time graph network provided by the present application. Figures 1-4

[0037] Figure 1 Figure is one of the flow diagrams of the optical transmission network root alarm identification method based on the space-time graph network provided by the present application, as shown in the figure, the optical transmission network root alarm identification method based on the space-time graph network includes the following steps: Figure 1

[0038] Step 110, constructing a space-time alarm subgraph according to the space-time clustering results of a plurality of alarm data; wherein the plurality of alarm data includes root alarm data and derived alarm data, and the space-time alarm subgraph is used to represent the alarm relationship with space-time characteristics. ​​

[0039] In this step, the alarm data includes one or more of device alarms, service alarms, OTU alarms, ODU alarms, network configuration errors, and system upgrade and maintenance data.

[0040] In this embodiment, by sequentially performing time clustering and space clustering on multiple alarm data, alarms that are truly related and suggest the same root fault are clustered together to construct a corresponding spatiotemporal alarm subgraph.

[0041] In this embodiment, for time clustering, a timestamp or a time interval can be used as an index for measuring alarm correlation, for example, alarms occurring within a m (the value of m can be set according to user requirements) minute window are classified into the same class.

[0042] In this embodiment, for space clustering, as long as two alarm events are physically topologically connected, they can be classified into the same class.

[0043] In this embodiment, multiple alarm data can be sequentially clustered by a two-stage clustering based on time and space (such as the topological relationship between multiple alarm events) to construct a corresponding spatiotemporal alarm subgraph; specifically, alarms and network elements are regarded as graph nodes, network elements are connected according to physical topology, and alarms are associated with the network element nodes that reported the alarms.

[0044] Step 120, identifying the spatiotemporal alarm subgraph based on the spatiotemporal graph network to obtain a target alarm identification result; wherein the target alarm identification result includes at least one of an alarm type, an edge weight parameter between different alarm identification results, and alarm positioning information; the spatiotemporal graph network is trained by taking multiple sample alarm data as nodes, taking the correlation between multiple sample alarm data as an edge weight parameter between different node features, and updating the edge weight parameter by sampling attention mechanism.

[0045] In this step, the spatiotemporal graph network is the same as the Graph Attention Networks (GAT), which learns the weight between nodes by introducing an attention mechanism, thereby achieving a more effective representation of graph data; this spatiotemporal graph network can assign different attention weights to different nodes in the graph to reflect the importance of different node features.

[0046] In this step, the spatiotemporal alarm subgraph includes multiple alarm clustering results and their edge weight parameters, and the edge weight parameter includes a weight value corresponding to the edge between two nodes.

[0047] In this embodiment, the spatio-temporal graph network adopts a multi-head attention mechanism to capture different aspects of node features, each attention head learns a set of attention scores, and the final node representation is a combination or average of all head results, which enables the model to learn the representation of the node from different perspectives; for example, the spatio-temporal graph network learns multiple categories of alarm data, location information, and the relevance between different types of alarms using a multi-head attention mechanism.

[0048] It should be noted that the spatio-temporal graph network performs well in predicting edge weights because it intuitively adjusts the importance of node interaction without explicit information about the global graph structure

[0049] In this embodiment, the spatio-temporal graph network identifies the root alarm by updating the edge weights in the spatio-temporal alarm subgraph; specifically, the spatio-temporal graph network recalibrates the edge weights between alarm nodes during the identification process to reflect the relevance of the nodes, rather than just the time interval; and the spatio-temporal graph network uses attention to dynamically assess the importance of each node neighbor, thereby focusing the model on the most relevant connections during the feature aggregation process.

[0050] The method for identifying root alarms in an optical transmission network based on a spatio-temporal graph network provided in the embodiments fully considers the spatio-temporal characteristics of the alarms, improves the accuracy of root alarm identification in the optical transmission network, and enhances the interpretability of the alarm identification results.

[0051] In some embodiments, constructing a spatio-temporal alarm subgraph based on the spatio-temporal clustering results of the multiple alarm data includes: clustering the multiple alarm data based on the generation times of the alarm data to obtain time clustering results; and performing spatial clustering on the time clustering results based on the physical topology relationships between the multiple alarm data to obtain spatio-temporal clustering results, and constructing the spatio-temporal alarm subgraph based on the spatio-temporal clustering results.

[0052] Figure 2 is a flowchart of the method for identifying root alarms in an optical transmission network based on a spatio-temporal graph network provided by the present application, and Figure 2 In the embodiment shown in the figure, the spatio-temporal alarm subgraph is constructed by the following steps:

[0053] (1) Perform two-stage alarm clustering based on topology and time.

[0054] Specifically, first, perform time-based clustering to ensure that only alarms occurring within a m=15-minute window are grouped into the same class; then perform topology-based clustering to group alarms into a cluster as long as the alarms in the time clustering results are physically topologically connected.

[0055] (2) Constructing a spatiotemporal alarm subgraph according to the final clustering result to fully consider the spatiotemporal characteristics of the alarm.

[0056] In this embodiment, for the repeated alarm events, the spatiotemporal alarm subgraph does not increase new alarm nodes, but increases the number of connected edges.

[0057] In this embodiment, the time feature embedding in the spatiotemporal alarm subgraph is embodied through the edge weight of the graph: the alarms are associated with the network elements in the time sequence, and all the alarms are connected through the weight represented by the time interval at which they occur.

[0058] In this embodiment, since the identification of the root alarm is highly dependent on the time feature, using time as the initial weight helps to accelerate the convergence of the model.

[0059] Figure 3 is a schematic diagram of the topological relationship between the network element and the corresponding edge in the spatiotemporal alarm subgraph provided by the present application, in which Figure 3 In the embodiment shown in the figure, the spatiotemporal alarm subgraph includes 16 network element topological interconnections (NE1-NE16), generates 45 alarm events, and fully represents the time feature by embedding the space attribute of the network element and the alarm node and taking the time interval as the edge weight; the edge value between the alarms includes the alarm occurrence time difference.

[0060] The optical transmission network root alarm identification method based on the spatiotemporal graph network provided by the embodiment of the present application clusters a plurality of alarm data through the generation time of each alarm data, then spatially clusters the time clustering result according to the physical topological relationship between the plurality of alarm data, constructs the obtained clustering result into a spatiotemporal alarm subgraph, so as to be able to depict the spatiotemporal correlation between each alarm event in the form of a graph, support a plurality of alarm analysis tasks, and further enhance the explainability of the alarm identification result.

[0061] In some embodiments, the target alarm identification result includes a plurality of root alarm identification results; after obtaining the target alarm identification result, the optical transmission network root alarm identification method based on the spatiotemporal graph network further includes: screening the target alarm identification result according to the edge weight parameter corresponding to each root alarm identification result to obtain a first root alarm identification result.

[0062] In this embodiment, the target alarm identification includes a plurality of root alarm identification results and an edge weight value corresponding to each root alarm identification result, and the larger the edge weight value, the greater the possibility that the good scene data is the root cause of the fault.

[0063] In Figure 2 In the embodiment shown in the figure, the graph attention network model (3) (i.e., the spatiotemporal graph network) constructs a spatiotemporal alarm subgraph according to the prediction function f (x edge weight prediction is performed on the edge features between the node features in the spatio-temporal alarm subgraph, a corresponding root alarm (R_LOS) node is identified from N alarm clustering results, and a plurality of analysis tasks are implemented according to each root alarm node: task 1, root alarm identification, task 2, alarm correlation, and task 3, alarm positioning; wherein the constructed spatio-temporal alarm subgraph (2) includes network elements NE1 and NE2, includes alarms A1, A2 and A3, the network elements are connected through a physical topology, the connection relationship between the network elements and the alarms indicates that the alarm is reported to the network element, and the reporting order is 1, 2, …, n; the connection relationship between the alarm A1 and the alarm A2 indicates that A1 occurs 6 seconds before A2, and so on; the alarm subgraph (4) after the graph attention network updates the connection relationship between each alarm, and the corresponding edge value is represented by a1, a2 and a3.

[0064] Table 1 Weight ordering table in alarm subgraph after learning of graph attention network model

[0065]

[0066] In the above table 1, the weights of all edges in the spatio-temporal alarm subgraph are sorted by the spatio-temporal graph network, which can explain whether the root alarm identification is successful; wherein, for two root alarms A12 and A1, since the edge weight value of the A12 alarm is the largest, it can be determined that this alarm is the root alarm, which means that the larger the weight value, the greater the possibility that the alarm is the root cause.

[0067] The optical transmission network root alarm identification method based on the spatio-temporal graph network provided in the embodiment can filter the target alarm identification result through the edge weight parameter corresponding to each root alarm identification result, thereby improving the reliability of obtaining the root alarm identification result.

[0068] Further, the target alarm identification result also includes a plurality of derived alarm identification results; after obtaining the first root alarm identification result, the optical transmission network root alarm identification method based on the spatio-temporal graph network further includes: filtering the target alarm identification result according to the edge weight parameter corresponding to the first root alarm identification result and the edge weight parameter corresponding to each derived alarm identification result, to obtain a second root alarm identification result.

[0069] In this embodiment, the target identification result includes one or more alarm identification results and their corresponding edge weight values, and also includes each derived alarm identification result and its corresponding edge weight value.

[0070] In Table 1, through the continuous screening of the topological relationship of the alarm A1, it is found that the edge weight of the non-root alarm A9 (i.e. the derived alarm) is 0.08 higher than that of the root alarm A1, which indicates that A9 is more strongly connected with other alarms and is more likely to be a root alarm. Therefore, the attempt of identifying A1 as a root alarm can fail in this embodiment. In summary, the target alarm identification result corresponding to the spatiotemporal alarm subgraph can be used as an interpretable graph supporting A12 as the root cause of the alarm.

[0071] In Figure 2 In the embodiment shown in FIG. 4, the alarm subgraph (4) after the spatiotemporal graph network contains the root alarm identification result, and the output edge weight represents the correlation between the alarm events, including the spatiotemporal relationship and the potential derived interaction. The greater the edge weight corresponding to an alarm in the subgraph, the higher the possibility that the alarm is the root cause, indicating the origin of the fault. If a subgraph can contain multiple root alarms, the root alarm can be successfully identified if the weight of the root alarm is higher than that of the non-root alarm in the same subgraph.

[0072] The method for identifying a root alarm of an optical transmission network based on a spatiotemporal graph network provided in the embodiments of the present application further improves the reliability of obtaining the root alarm identification result by screening the target alarm identification result according to the edge weight parameters corresponding to the first root alarm identification result and the edge weight parameters corresponding to each derived alarm identification result.

[0073] In some embodiments, the target alarm identification result is an alarm identification subgraph; wherein the alarm identification subgraph includes alarm types, edge weight values between different alarm identification results, and physical topological relationships of each alarm identification result.

[0074] In this embodiment, by screening and analyzing the alarm identification subgraph, the position of the root alarm and the association between the root alarm and other alarms can be quickly obtained, the interpretability of the successful or failed identification of the root alarm is enhanced, and the subsequent maintenance and management of the optical transmission network fault are facilitated.

[0075] The device for identifying a root alarm of an optical transmission network based on a spatiotemporal graph network provided in the present application will be described below. The device for identifying a root alarm of an optical transmission network based on a spatiotemporal graph network described below can be correspondingly referred to the method for identifying a root alarm of an optical transmission network based on a spatiotemporal graph network described above.

[0076] Figure 4 FIG. 4 is a structural schematic diagram of the device for identifying a root alarm of an optical transmission network based on a spatiotemporal graph network provided in the present application, as shown in the figure, the device for identifying a root alarm of an optical transmission network based on a spatiotemporal graph network includes a clustering module 410 and an identification module 420. Figure 4

[0077] ​The space-time clustering and composition module 410 is configured to construct a space-time alarm subgraph according to a space-time clustering result of the plurality of alarm data, wherein the plurality of alarm data comprises root alarm data and derived alarm data, and the space-time alarm subgraph is used to represent an alarm relationship with space-time characteristics.

[0078] The recognition module 420 is configured to recognize the space-time alarm subgraph based on a space-time graph network to obtain a target alarm recognition result, wherein the target alarm recognition result comprises at least one of an alarm type, an edge weight parameter between different alarm recognition results, and alarm positioning information; and the space-time graph network is obtained by training in a manner of taking a plurality of sample alarm data as nodes, taking a correlation between the plurality of sample alarm data as an edge weight parameter between different node features, and updating the edge weight parameter by sampling an attention mechanism.

[0079] The space-time graph network-based optical transmission network root alarm recognition device provided by the embodiment of the present application can obtain alarm types, edge weight parameters between different alarm recognition results, alarm positioning information, and other recognition results by sequentially performing time clustering and space clustering on a plurality of alarm data, fully considers the space-time characteristics of alarms, improves the accuracy of optical transmission network root alarm recognition, and enhances the explainability of alarm recognition results.

[0080] In some embodiments, the target alarm recognition result comprises a plurality of root alarm recognition results; and the space-time graph network-based optical transmission network root alarm recognition device further comprises a screening module configured to, after obtaining the target alarm recognition result, screen the target alarm recognition result according to edge weight parameters corresponding to each root alarm recognition result to obtain a first root alarm recognition result.

[0081] The space-time graph network-based optical transmission network root alarm recognition device provided by the embodiment of the present application can screen the target alarm recognition result according to edge weight parameters corresponding to each root alarm recognition result, and improve the reliability of obtaining root alarm recognition results.

[0082] Figure 5 is a structural schematic diagram of an electronic device provided by the present application, as Figure 5As shown, the electronic device can include a processor 510, a communications interface 520, a memory 530, and a communications bus 540, wherein the processor 510, the communications interface 520, and the memory 530 complete mutual communication through the communications bus 540. The processor 510 can invoke a logical instruction in the memory 530 to execute a method for identifying a root alarm of an optical transmission network based on a space-time graph network, the method including: constructing a space-time alarm subgraph according to a space-time clustering result of a plurality of alarm data; wherein the plurality of alarm data includes root alarm data and derived alarm data, and the space-time alarm subgraph is used to represent an alarm relationship with space-time characteristics; identifying the space-time alarm subgraph based on a space-time graph network to obtain a target alarm identification result; wherein the target alarm identification result includes at least one of an alarm type, an edge weight parameter between different alarm identification results, and alarm positioning information; and the space-time graph network is obtained by training in a manner of taking a plurality of sample alarm data as nodes, taking a correlation between the plurality of sample alarm data as an edge weight parameter between different node features, and updating the edge weight parameter by sampling an attention mechanism.

[0083] In addition, the logical instructions in the memory 530 described above can be implemented in the form of a software function unit and sold or used as an independent product, and can be stored in a computer-readable storage medium. Based on this understanding, the technical solutions of the present application essentially or the part that contributes to the prior art or part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various program code storage media.

[0084] In another aspect, the present application also provides a computer program product comprising a computer program, which can be stored on a non-transitory computer-readable storage medium, and the computer program is executable by a processor to enable the computer to perform the method of identifying root alarm of optical transmission network based on a spatio-temporal graph network, which comprises: constructing a spatio-temporal alarm subgraph according to a spatio-temporal clustering result of a plurality of alarm data; wherein the plurality of alarm data comprises root alarm data and derived alarm data, and the spatio-temporal alarm subgraph is used to represent alarm relationships with spatio-temporal characteristics; identifying the spatio-temporal alarm subgraph based on the spatio-temporal graph network to obtain a target alarm identification result; wherein the target alarm identification result comprises at least one of an alarm type, an edge weight parameter between different alarm identification results, and alarm positioning information; and the spatio-temporal graph network is trained by taking a plurality of sample alarm data as nodes, taking correlations between the plurality of sample alarm data as edge weight parameters between different node features, and updating the edge weight parameters by using a sampling attention mechanism.

[0085] In another aspect, the present application also provides a non-transitory computer-readable storage medium having a computer program stored thereon, and the computer program is executable by a processor to implement the method of identifying root alarm of optical transmission network based on a spatio-temporal graph network, which comprises: constructing a spatio-temporal alarm subgraph according to a spatio-temporal clustering result of a plurality of alarm data; wherein the plurality of alarm data comprises root alarm data and derived alarm data, and the spatio-temporal alarm subgraph is used to represent alarm relationships with spatio-temporal characteristics; identifying the spatio-temporal alarm subgraph based on the spatio-temporal graph network to obtain a target alarm identification result; wherein the target alarm identification result comprises at least one of an alarm type, an edge weight parameter between different alarm identification results, and alarm positioning information; and the spatio-temporal graph network is trained by taking a plurality of sample alarm data as nodes, taking correlations between the plurality of sample alarm data as edge weight parameters between different node features, and updating the edge weight parameters by using a sampling attention mechanism.

[0086] The device embodiments described above are only schematic, wherein the units shown as separate components can or can not be physically separate, and the components shown as units can or can not be physical units, i.e., can be located in one place, or can be distributed on a plurality of network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the present embodiment. Those skilled in the art can understand and implement without creative labor.

[0087] Those skilled in the art can clearly understand the technical solutions of the various embodiments from the above description of the embodiments, and the various embodiments can be implemented by means of software with the necessary general hardware platforms, and of course, can also be implemented by hardware. Based on such understanding, the above technical solutions, essentially or in other words, the part of the prior art that makes a contribution, can be embodied in the form of a software product, which can be stored in a computer readable storage medium, such as a ROM / RAM, a magnetic disk, an optical disk, and the like, and includes a number of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0088] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, rather than limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for some technical features therein; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for root cause identification of an optical transport network alarm based on a spatio-temporal graph network, characterized in that, The method comprises: constructing a time-space alarm subgraph according to a time-space clustering result of a plurality of alarm data; wherein the plurality of alarm data comprises root alarm data and derived alarm data, and the time-space alarm subgraph is used to represent alarm relationships with time-space characteristics; identifying the time-space alarm subgraph based on a time-space graph network to obtain a target alarm identification result; wherein the target alarm identification result comprises at least one of an alarm type, an edge weight parameter between different alarm identification results, and alarm positioning information; the time-space graph network is trained by taking a plurality of sample alarm data as nodes, taking the correlation between the plurality of sample alarm data as an edge weight parameter between different node features, and updating the edge weight parameter by using an attention mechanism; and the time-space alarm subgraph is constructed according to the time-space clustering result of the plurality of alarm data, which comprises: clustering the plurality of alarm data according to the generation time of each alarm data to obtain a time clustering result; spatially clustering the time clustering result according to the physical topology relationship between the plurality of alarm data to obtain the time-space clustering result, and constructing the time-space alarm subgraph according to the time-space clustering result; the target alarm identification result comprises a plurality of root alarm identification results; after obtaining the target alarm identification result, the method further comprises: screening the target alarm identification result according to the edge weight parameter corresponding to each root alarm identification result to obtain a first root alarm identification result; the target alarm identification result further comprises a plurality of derived alarm identification results; after obtaining the first root alarm identification result, the method further comprises: screening the target alarm identification result according to the edge weight parameter corresponding to the first root alarm identification result and the edge weight parameter corresponding to each derived alarm identification result to obtain a second root alarm identification result.

2. The method of claim 1, wherein the target alarm identification result is an alarm identification subgraph. The alarm identification subgraph comprises an alarm type, an edge weight value between different alarm identification results, and a physical topology relationship of each alarm identification result.

3. A device for identifying root alarms of an optical transport network based on a spatio-temporal graph network, applying the method for identifying root alarms of an optical transport network based on a spatio-temporal graph network according to claim 1, characterized in that, The method comprises: a time-space clustering and mapping module configured to construct a time-space alarm subgraph according to a time-space clustering result of a plurality of alarm data; wherein the plurality of alarm data comprises root alarm data and derived alarm data, and the time-space alarm subgraph is used to represent alarm relationships with time-space characteristics; an identification module configured to identify the time-space alarm subgraph based on a time-space graph network to obtain a target alarm identification result; wherein the target alarm identification result comprises at least one of an alarm type, an edge weight parameter between different alarm identification results, and alarm positioning information; and the time-space graph network is trained by taking a plurality of sample alarm data as nodes, taking the correlation between the plurality of sample alarm data as an edge weight parameter between different node features, and updating the edge weight parameter by using an attention mechanism.

4. An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor executes the computer program to implement the optical transmission network root alarm identification method based on the time-space graph network according to any one of claims 1 to 2.

5. A non-transitory computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the optical transmission network root alarm identification method based on the time-space graph network according to any one of claims 1 to 2.

6. A computer program product comprising a computer program, characterized in that, The computer program is executed by a processor to implement the optical transmission network root alarm identification method based on the space-time diagram network according to any one of claims 1 to 2.

Images