A Fault Path Tracing Method and System for Distribution Communication Networks Based on Knowledge Graphs

Abstract

The present invention relates to the technical field of distribution communication networks, and discloses a method and system for tracing the fault path of a distribution communication network based on a knowledge graph. The method includes: obtaining the dependency relationship strength of each communication link according to the communication data of each communication link in the knowledge graph; identifying the fault propagation path in the set of directed edges based on the fault propagation rate calculated based on the dependency relationship strength; screening the fault propagation path graph based on a preset fault propagation termination condition to obtain a first fault propagation path sub-graph, and determining the degree of each hierarchical node in the first fault propagation path sub-graph; identifying key nodes in the first fault propagation path sub-graph according to the first node importance; and expanding the first fault propagation path sub-graph based on the out-edge link data to obtain a second fault propagation path sub-graph. The present invention can quickly locate key fault nodes and propagation paths.

Description

Technical Field

[0001] The present invention relates to the technical field of distribution communication networks, and particularly to a method and system for tracing the fault path of a distribution communication network based on a knowledge graph. Background Art

[0002] In the construction of smart grids, the stability of distribution communication networks is crucial. With the improvement of the intelligence level of the power grid, the number and types of devices in the distribution communication network are increasing continuously, and the topological relationship between devices is becoming more and more complex. By modeling the topological relationship between devices through a knowledge graph, the hierarchical structure and connection mode of devices can be intuitively displayed, thus facilitating the fault analysis of the distribution communication network. When a key node device fails, the fault information will spread according to the topological structure of the knowledge graph, and the propagation path is affected by the device connection relationship and dependence intensity, triggering a fault cascade effect. In this process, it is necessary to dynamically adjust the node importance weight and the graph inference depth. The node importance weight is determined based on the device topological position, the fault transmission path, and the degree of abnormal state, and the graph inference depth depends on the complexity of the fault propagation and the change of the device dependence relationship.

[0003] However, the existing fault tracing technologies have some defects in dynamically adjusting the node importance weight and the graph inference depth. First, they overemphasize the device information at a certain level while ignoring the potential fault sources at other levels; second, the insufficient graph inference depth leads to the truncation of the fault propagation path, thus unable to fully present the fault cascade effect, seriously affecting the accuracy of fault tracing.

[0004] Therefore, there is an urgent need to propose an efficient method for tracing the fault path of a distribution communication network to achieve the dynamic balance of the node importance weight and the graph inference depth. Summary of the Invention

[0005] In view of the defects existing in the prior art, the present invention provides a method and system for tracing the fault path of a distribution communication network based on a knowledge graph.

[0006] In a first aspect, an embodiment of the present invention provides a method for tracing the fault path of a distribution communication network based on a knowledge graph, including:

[0007] Constructing a corresponding knowledge graph based on the operation state and topological structure of the target distribution communication network, and obtaining the dependence relationship intensity corresponding to each communication link according to the communication data of each communication link in the knowledge graph;

[0008] Obtaining the set of directed edges of abnormal nodes in the knowledge graph, and identifying the fault propagation path in the set of directed edges based on the fault propagation rate calculated according to the dependence relationship intensity, where the set of directed edges includes the directed edges between the abnormal node and each adjacent node;

[0009] Traverse the fault propagation path to generate a fault propagation path graph, and filter the fault propagation path graph based on a preset fault propagation termination condition to obtain a first sub - graph of the fault propagation path, and determine the degree of each hierarchical node in the first sub - graph of the fault propagation path;

[0010] Perform fitting prediction based on the degree of each hierarchical node in the first sub - graph of the fault propagation path to obtain the first node importance of the corresponding node, and identify the key nodes in the first sub - graph of the fault propagation path according to the first node importance;

[0011] Obtain the out - edge link data of the key nodes in the first sub - graph of the fault propagation path, and expand the first sub - graph of the fault propagation path based on the out - edge link data to obtain a second sub - graph of the fault propagation path.

[0012] Preferably, obtaining the dependency strength of each communication link in the knowledge graph according to the communication data of each communication link in the knowledge graph includes:

[0013] Obtain the first operating state data of each communication link in the knowledge graph, and perform weighted calculation based on the first operating state data to obtain the first link score corresponding to the communication link, where the first operating state data includes communication delay, bandwidth utilization rate, and data packet loss rate;

[0014] Perform weighted correction on the first link score based on the physical distance of each communication link to obtain the second link score corresponding to the communication link;

[0015] Obtain the service traffic data of each communication link, and perform weighted calculation based on the service traffic data to obtain the service dependency of the corresponding communication link, where the service traffic data includes control instruction flow, measurement data flow, and alarm information flow;

[0016] Perform weighted fusion on the second link score and the service dependency of each communication link to obtain the dependency strength of the corresponding communication link.

[0017] Preferably, before obtaining the set of directed edges of abnormal nodes in the knowledge graph, it further includes:

[0018] Obtain the second operating state data of each node in the knowledge graph, and calculate the degree of abnormality of the corresponding node based on the second operating state data, where the second operating state data includes response delay time, load rate, and bandwidth utilization rate;

[0019] Judge whether the degree of abnormality is greater than a preset abnormality determination threshold, and if so, determine that the corresponding node is an abnormal node.

[0020] Preferably, the fault propagation rate calculated based on the dependency strength identifies the fault propagation path in the set of directed edges, including:

[0021] Obtain the dependency strength of the communication link corresponding to each directed edge in the set of directed edges, and perform a linear calculation on the dependency strength to obtain the initial fault propagation rate corresponding to the directed edge;

[0022] Obtain the performance index data of the communication link corresponding to each directed edge, and perform a weighted calculation based on the performance index data to obtain the edge weight coefficient corresponding to the directed edge, where the performance index data includes physical distance, network hop count, and communication delay;

[0023] Based on the edge weight coefficient of each directed edge, perform a weighted correction on the initial fault propagation rate to obtain the fault propagation rate corresponding to the directed edge;

[0024] Judge whether the fault propagation rate is greater than the preset propagation rate threshold. If so, mark the corresponding directed edge as the fault propagation path.

[0025] Preferably, traversing the fault propagation path to generate a fault propagation path graph, and screening the fault propagation path graph based on the preset fault propagation termination condition to obtain the first fault propagation path subgraph, and determining the degree of each level node in the first fault propagation path subgraph, including:

[0026] Use the breadth-first search algorithm to traverse the fault propagation path to obtain the fault propagation path graph;

[0027] Use the depth-first search algorithm to obtain the number of levels of the fault propagation path graph;

[0028] Based on the total number of nodes and the maximum network diameter of the target distribution communication network, calculate the preset fault propagation depth reflecting the network scale of the target distribution communication network;

[0029] Based on the number of levels and the preset fault propagation depth, judge whether the fault propagation path graph meets the preset fault propagation termination condition. If so, screen the fault propagation path graph to obtain the first fault propagation path subgraph, otherwise characterize the fault propagation path graph as the first fault propagation path subgraph, where the preset fault propagation termination condition includes that the number of levels is greater than the preset fault propagation depth;

[0030] Count the in-degree and out-degree of each level node in the first fault propagation path subgraph.

[0031] Preferably, the first node importance corresponding to each node is obtained by fitting and predicting based on the degrees of nodes at each level in the first fault propagation path subgraph, and the key nodes in the first fault propagation path subgraph are identified according to the first node importance, including:

[0032] Perform weighted calculation on the degrees of nodes at each level in the first fault propagation path subgraph to obtain the node eigenvalue corresponding to each node;

[0033] Use the random forest algorithm to perform fitting and prediction on the node eigenvalues of nodes at each level in the first fault propagation path subgraph to obtain the first node importance corresponding to each node;

[0034] Evaluate the eigenvector centrality of nodes at each level in the first fault propagation path subgraph, and characterize the obtained network centrality as the second node importance corresponding to each node;

[0035] Perform weighted calculation on the first node importance and the second node importance to obtain the node importance corresponding to each node;

[0036] Judge whether the node importance is higher than the first preset threshold. If so, mark the corresponding node as the key node in the first fault propagation path subgraph.

[0037] Preferably, the first fault propagation path subgraph is expanded based on the out-edge link data to obtain the second fault propagation path subgraph, including:

[0038] Perform weighted calculation based on the out-edge link data to obtain the node coupling coefficient corresponding to each out-edge link, where the out-edge link data includes the communication delay, bandwidth utilization rate, and data packet loss rate of several out-edge links;

[0039] Use the random forest algorithm to perform fitting and prediction on the node coupling coefficient corresponding to each out-edge link to obtain the fault propagation influence intensity of the target node in each out-edge link;

[0040] Determine the fault propagation depth increment value based on the fault propagation influence intensity, and perform hierarchical expansion on the first fault propagation path subgraph according to the fault propagation depth increment value to obtain the third fault propagation path subgraph;

[0041] Obtain the communication link data between the key nodes and each newly added boundary node in the third fault propagation path subgraph, and calculate the fault propagation probability between the key nodes and each newly added boundary node in the third fault propagation path subgraph based on the communication link data;

[0042] Node labeling is performed on the third fault propagation path subgraph based on the fault propagation probability to obtain a second fault propagation path subgraph, where the node labeling includes labeling the new boundary nodes corresponding to the fault propagation probability greater than the second preset threshold in the third fault propagation path subgraph.

[0043] In a second aspect, an embodiment of the present invention provides a fault path tracing system for a distribution communication network based on a knowledge graph, including:

[0044] A dependency strength determination module, configured to construct a corresponding knowledge graph based on the operating state and topological structure of a target distribution communication network, and obtain the dependency strength corresponding to each communication link according to the communication data of each communication link in the knowledge graph;

[0045] A fault propagation path determination module, configured to obtain a set of directed edges of abnormal nodes in the knowledge graph, and identify the fault propagation paths in the set of directed edges based on the fault propagation rate calculated according to the dependency strength, where the set of directed edges includes the directed edges between the abnormal nodes and each adjacent node;

[0046] A node degree determination module, configured to traverse the fault propagation path to generate a fault propagation path graph, and filter the fault propagation path graph based on a preset fault propagation termination condition to obtain a first fault propagation path subgraph, and determine the degree of each level of nodes in the first fault propagation path subgraph;

[0047] A key node identification module, configured to perform fitting prediction based on the degree of each level of nodes in the first fault propagation path subgraph to obtain the first node importance of the corresponding nodes, and identify the key nodes in the first fault propagation path subgraph according to the first node importance;

[0048] A target fault propagation path determination module, configured to obtain the out-edge link data of the key nodes in the first fault propagation path subgraph, and expand the first fault propagation path subgraph based on the out-edge link data to obtain a second fault propagation path subgraph.

[0049] Preferably, the dependency strength determination module includes:

[0050] A first link score determination unit, configured to obtain the first operating state data of each communication link in the knowledge graph, and perform weighted calculation based on the first operating state data to obtain the first link score corresponding to the communication link, where the first operating state data includes communication delay, bandwidth utilization rate, and data packet loss rate;

[0051] The second link scoring determination unit is configured to perform weighted correction on the first link score based on the physical distance of each communication link to obtain the second link score corresponding to the communication link;

[0052] The service dependence determination unit is configured to obtain the service traffic data of each communication link and perform weighted calculation based on the service traffic data to obtain the service dependence corresponding to the communication link, where the service traffic data includes control instruction flow, measurement data flow, and alarm information flow;

[0053] The weighted fusion unit is configured to perform weighted fusion on the second link score and the service dependence of each communication link to obtain the dependence relationship strength corresponding to the communication link.

[0054] Preferably, the fault propagation path determination module includes:

[0055] The linear calculation unit is configured to obtain the dependence relationship strength of each directed edge corresponding to the communication link in the directed edge set, and perform linear calculation on the dependence relationship strength to obtain the initial fault propagation rate corresponding to the directed edge;

[0056] The weighted calculation unit is configured to obtain the performance index data of each directed edge corresponding to the communication link, and perform weighted calculation based on the performance index data to obtain the edge weight coefficient corresponding to the directed edge, where the performance index data includes physical distance, network hop count, and communication delay;

[0057] The weighted correction unit is configured to perform weighted correction on the initial fault propagation rate based on the edge weight coefficient of each directed edge to obtain the fault propagation rate corresponding to the directed edge;

[0058] The propagation rate judgment unit is configured to judge whether the fault propagation rate is greater than a preset propagation rate threshold, and if so, mark the corresponding directed edge as the fault propagation path.

[0059] Compared with the prior art, the method and system for tracing the fault path of a distribution communication network based on a knowledge graph according to an embodiment of the present invention have the beneficial effects that: by constructing a knowledge graph, the operation state and topological structure information of the distribution communication network can be comprehensively integrated; according to the multi-source data of the communication link, the dependence relationship strength can be accurately obtained, and the fault propagation path can be effectively identified; in the analysis of the fault propagation path graph, based on the preset fault propagation termination condition, the first fault propagation path sub-graph is screened out, and then the key nodes are accurately identified; by using the out-edge link data of the key nodes, the sub-graph is expanded. The present invention greatly improves the accuracy and comprehensiveness of fault analysis, can quickly locate the key fault nodes and propagation paths, and realizes the dynamic balance of the node importance and the reasoning depth of the knowledge graph. Description of the Drawings

[0060] Figure 1 It is a schematic flow chart of a method for tracing the fault path of a distribution communication network based on a knowledge graph according to an embodiment of the present invention;

[0061] Figure 2 It is a schematic flow chart of obtaining the strength of the dependency relationship according to an embodiment of the present invention;

[0062] Figure 3 It is a schematic structural diagram of a system for tracing the fault path of a distribution communication network based on a knowledge graph according to an embodiment of the present invention. Detailed implementation manners

[0063] The following will further describe in detail the specific implementation manners of the present invention in conjunction with the accompanying drawings and embodiments. The following embodiments are used to illustrate the present invention, but are not used to limit the scope of the present invention.

[0064] In the description of the present invention, it should be understood that the terms "first" and "second" etc. used in the present invention are used to distinguish different objects, rather than to describe a specific order.

[0065] In the description of the present invention, it should be noted that unless otherwise defined, all technical and scientific terms used in the present invention have the same meaning as commonly understood by those skilled in the art. The terms used in the specification of the present invention are only for the purpose of describing specific embodiments, and are not intended to limit the present invention. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific situations.

[0066] As Figure 1 shown, an embodiment of the present invention provides a method for tracing the fault path of a distribution communication network based on a knowledge graph, including the steps:

[0067] S1. Construct a corresponding knowledge graph based on the operating state and topological structure of the target distribution communication network, and obtain the strength of the dependency relationship of each communication link according to the communication data of each communication link in the knowledge graph;

[0068] To construct the knowledge graph of the target distribution communication network, it is first necessary to determine the topological structure of the target distribution communication network, then obtain the relevant operation state data, use technologies such as knowledge extraction and knowledge fusion to convert the relevant data into knowledge, and finally display it in an intuitive graphical manner. The communication data of each communication link in the knowledge graph includes first operating state data, physical distance, and service traffic data.

[0069] Specifically, as Figure 2 shown, step S1 includes:

[0070] S101. Obtain the first running state data of each communication link in the knowledge graph, and perform weighted calculation based on the first running state data to obtain the first link score corresponding to the communication link;

[0071] The first running state data includes communication delay, bandwidth utilization rate, and data packet loss rate. Weighted calculation is performed on the communication delay, bandwidth utilization rate, and data packet loss rate to obtain the first link score corresponding to the communication link. This score reflects the running quality of the communication link. It should be noted that weights are assigned to the communication delay, bandwidth utilization rate, and data packet loss rate respectively based on the service requirements of the target distribution communication network. Considering that there is a differential protection service with high real-time requirements in the target distribution communication network, in this embodiment, the weight coefficient of communication delay is set to 0.4, the weight coefficient of bandwidth utilization rate is set to 0.3, and the weight coefficient of data packet loss rate is set to 0.3.

[0072] S102. Perform weighted correction on the first link score based on the physical distance of each communication link to obtain the second link score corresponding to the communication link;

[0073] Considering that the physical distances of each communication link vary greatly, the physical distance is converted into a normalization coefficient between 0 and 1, and the first link score is weighted and corrected according to the normalized physical distance to obtain the second link score corresponding to the communication link. By correcting the first link score through the physical distance, it can be found that the farther the link is, the more obvious the correction effect on its score is, which reflects the influence of physical distance on the reliability of the communication link.

[0074] S103. Obtain the service traffic data of each communication link, and perform weighted calculation based on the service traffic data to obtain the service dependence degree corresponding to the communication link;

[0075] The service traffic data includes control instruction flow, measurement data flow, and alarm information flow. Specifically, the control instruction flow includes key instructions such as start-stop control of distributed generation equipment and distribution network topology switching control, the measurement data flow includes real-time measurement values such as voltage, current, active power, and reactive power, and the alarm information flow includes operation events such as equipment fault alarms and over-limit alarms.

[0076] Weighted calculation is performed on the control instruction flow, measurement data flow, and alarm information flow to obtain the service dependence degree corresponding to the communication link. It should be noted that considering the importance of different types of service traffic data to the safe and stable operation of the target distribution communication network, in this embodiment, the weight coefficient of the control instruction flow is set to 0.4, the weight coefficient of the measurement data flow is set to 0.3, and the weight coefficient of the alarm information flow is set to 0.3.

[0077] S104. Perform weighted fusion on the second link score and the service dependence degree of each communication link to obtain the dependence relationship strength corresponding to the communication link.

[0078] After weighted fusion, the dependency strength of the corresponding communication link can be obtained, which can be used to evaluate the importance of the corresponding communication link in the entire target distribution communication network. This quantitative evaluation method provides an important reference for the optimal configuration and fault prevention of the target distribution communication network. It should be noted that in the weighted fusion process, the weight coefficients assigned to the second link score and the service dependency are the same.

[0079] S2. Obtain the set of directed edges of the abnormal nodes in the knowledge graph, and identify the fault propagation paths in the set of directed edges based on the fault propagation rate calculated according to the dependency strength;

[0080] Before obtaining the set of directed edges of the abnormal nodes in the knowledge graph, it further includes the steps of:

[0081] 1) Obtain the second operating state data of each node in the knowledge graph, and calculate the abnormality degree of the corresponding node based on the second operating state data;

[0082] Compare the second operating state data with the preset normal operating state data, and calculate the deviation ratio between the two to obtain the abnormality degree of the corresponding node. Specifically, the second operating state data includes response delay time, load rate, and bandwidth utilization rate, and the preset normal operating state data includes a response delay time of 5 milliseconds, a load rate of 40%, and a bandwidth utilization rate of 30%.

[0083] 2) Determine whether the abnormality degree is greater than the preset abnormality determination threshold. If so, determine that the corresponding node is an abnormal node.

[0084] After identifying the abnormal nodes, use the graph traversal algorithm to find the adjacent nodes of the abnormal nodes in the knowledge graph, so as to obtain the set of directed edges. The set of directed edges includes the directed edges between the abnormal nodes and each adjacent node. Specifically, step S2 includes:

[0085] 1) Obtain the dependency strength of each directed edge corresponding to the communication link in the set of directed edges, and perform a linear calculation on the dependency strength to obtain the initial fault propagation rate of the corresponding directed edge;

[0086] There is a certain positive correlation between the dependency strength and the initial fault propagation rate. The greater the dependency strength, the greater the initial fault propagation rate, because a tight dependency relationship makes it easier for abnormalities to spread between nodes.

[0087] Specifically, use a preset linear function to calculate the initial fault propagation path rate of the corresponding directed edge. Further, the parameters in the preset linear function can be estimated by regression analysis using the data such as the dependency strength and the fault propagation rate collected in the target distribution communication network.

[0088] 2) Obtain the performance index data of the communication link corresponding to each directed edge, and perform weighted calculation based on the performance index data to obtain the edge weight coefficient of the corresponding directed edge.

[0089] The performance index data includes physical distance, network hop count, and communication delay. Standardize the physical distance, network hop count, and communication delay of the communication link corresponding to each directed edge respectively, and perform weighted calculation on the standardized physical distance, network hop count, and communication delay to obtain the edge weight coefficient of the corresponding directed edge. The edge weight coefficient can comprehensively reflect the performance and quality of the communication link, providing a reference basis for network resource allocation. It should be noted that the respective weight coefficients of the physical distance, network hop count, and communication delay are determined according to the actual situation of the target distribution communication network. Considering that there are differential protection services with high real-time requirements in the target distribution communication network, the weight coefficient of the communication delay can be set larger.

[0090] 3) Based on the edge weight coefficient of each directed edge, perform weighted correction on the initial fault propagation rate to obtain the fault propagation rate of the corresponding directed edge.

[0091] 4) Determine whether the fault propagation rate is greater than the preset propagation rate threshold. If so, mark the corresponding directed edge as the fault propagation path.

[0092] In this embodiment, the preset propagation rate threshold is 0.06. If there is a fault propagation rate greater than 0.06, mark the corresponding directed edge as the fault propagation path. It should be noted that the preset propagation rate threshold in this embodiment can be adjusted according to the actual situation. If the marking requirement for the fault propagation path is high, the preset propagation rate threshold can be appropriately increased, and vice versa, it can be appropriately decreased.

[0093] S3. Traverse the fault propagation path to generate a fault propagation path graph, and screen the fault propagation path graph based on the preset fault propagation termination condition to obtain the first fault propagation path subgraph, and determine the degree of each hierarchical node in the first fault propagation path subgraph.

[0094] Specifically, step S3 includes:

[0095] 1) Use the breadth-first search algorithm to traverse the fault propagation path to obtain the fault propagation path graph.

[0096] The fault propagation path graph includes but is not limited to three levels: the control layer, the bay layer, and the execution layer. It intuitively shows the diffusion trend of the abnormal state in the target distribution communication network, highlighting the high-risk propagation paths, and providing an important basis for fault isolation and risk control.

[0097] 2) Use the depth-first search algorithm to obtain the number of levels of the fault propagation path graph.

[0098] The fault propagation path diagram includes but is not limited to three levels: the control layer, the bay layer, and the execution layer.

[0099] 3) Based on the total number of nodes and the maximum network diameter of the target distribution communication network, calculate the preset fault propagation depth reflecting the network scale of the target distribution communication network;

[0100] The total number of nodes reflects the overall scale of the target distribution communication network, while the maximum network diameter reflects the maximum distance between nodes in the target distribution communication network, that is, the maximum range where faults may spread.

[0101] Specifically, after normalizing the total number of nodes and the maximum network diameter respectively, the following formula is used to calculate the preset fault propagation depth:

[0102]

[0103] Among them, represents the preset fault propagation depth, represents the proportionality constant, which is used to adjust the final calculation result to conform to the actual network situation, and represent the weight coefficients, represents the normalized total number of nodes, represents the normalized maximum network diameter.

[0104] 4) Based on the number of levels and the preset fault propagation depth, determine whether the fault propagation path diagram meets the preset fault propagation termination condition. If so, screen the fault propagation path diagram to obtain the first fault propagation path sub-diagram; otherwise, characterize the fault propagation path diagram as the first fault propagation path sub-diagram;

[0105] The preset fault propagation termination condition includes that the number of levels of the fault propagation path diagram is greater than the preset fault propagation depth. When the fault propagation path diagram meets the preset fault propagation termination condition, the propagation path is truncated at the level equal to the preset fault propagation depth in the fault propagation path diagram to obtain the first fault propagation path sub-diagram; when the fault propagation path diagram does not meet the preset fault propagation termination condition, the fault propagation path diagram is characterized as the first fault propagation path sub-diagram.

[0106] 5) Count the in-degree and out-degree of the nodes at each level in the first fault propagation path sub-diagram.

[0107] Count the number of links pointing to the nodes at each level in the first fault propagation path sub-diagram as the in-degree of the corresponding level nodes, and count the number of links starting from the nodes at each level in the first fault propagation path sub-diagram as the out-degree of the corresponding level nodes.

[0108] S4. Perform fitting prediction based on the degrees of nodes at each level in the first fault propagation path sub-graph to obtain the first node importance of the corresponding nodes, and identify the key nodes in the first fault propagation path sub-graph according to the first node importance;

[0109] Specifically, step S4 includes:

[0110] 1) Perform weighted calculation on the degrees of nodes at each level in the first fault propagation path sub-graph to obtain the node feature values of the corresponding nodes;

[0111] Assign weight coefficients to the in-degree and out-degree of nodes at each level in the first fault propagation path sub-graph respectively, and perform weighted calculation on the in-degree and out-degree based on the weight coefficients to obtain the node feature values of the corresponding nodes. Considering that the weight coefficients of the in-degree and out-degree are the same, in this embodiment, the weight coefficient of the in-degree is set to 0.3, and the weight coefficient of the out-degree is set to 0.3.

[0112] 2) Use the random forest algorithm to perform fitting prediction on the node feature values of nodes at each level in the first fault propagation path sub-graph to obtain the first node importance of the corresponding nodes;

[0113] Use the random forest algorithm to realize the fitting of the relationship between the node feature values and the node importance, and predict the node feature values of nodes at each level in the first fault propagation path sub-graph according to this fitting relationship to obtain the first node importance of the corresponding nodes.

[0114] 3) Evaluate the eigenvector centrality of nodes at each level in the first fault propagation path sub-graph, and characterize the obtained network centrality as the second node importance of the corresponding nodes;

[0115] Use the eigenvector centrality algorithm to calculate the network centrality of nodes at each level in the first fault propagation path sub-graph, and characterize the network centrality as the second node importance of the corresponding nodes.

[0116] 4) Perform weighted calculation on the first node importance and the second node importance to obtain the node importance of the corresponding nodes;

[0117] In the process of weighted calculation, the weight coefficients assigned to the first node importance and the second node importance are the same.

[0118] 5) Judge whether the node importance is higher than the first preset threshold. If so, mark the corresponding node as the key node in the first fault propagation path sub-graph.

[0119] The key nodes are the nodes in the first fault propagation path subgraph whose node importance is higher than the first preset threshold. In this embodiment, the first preset threshold is 0.75. If there are node importances higher than 0.75, the corresponding nodes are marked as key nodes. It should be noted that the first preset threshold in this embodiment can be adjusted according to the actual situation. If the marking requirements for key nodes are relatively high, the first preset threshold can be appropriately increased; otherwise, it can be appropriately decreased.

[0120] S5. Obtain the out-edge link data of the key nodes in the first fault propagation path subgraph, and expand the first fault propagation path subgraph based on the out-edge link data to obtain the second fault propagation path subgraph.

[0121] The out-edge link data includes the communication delay, bandwidth utilization rate, and data packet loss rate of several out-edge links. Specifically, after obtaining the out-edge link data, step S5 includes:

[0122] 1) Perform weighted calculation based on the out-edge link data to obtain the node coupling coefficient corresponding to each out-edge link;

[0123] Perform standardization processing on the communication delay of each out-edge link in the out-edge link data, and assign weights to the standardized communication delay, bandwidth utilization rate, and data packet loss rate respectively. Finally, perform weighted calculation on the standardized communication delay, bandwidth utilization rate, and data packet loss rate based on the weights to obtain the node coupling coefficient corresponding to each out-edge link.

[0124] Specifically, the following formula is used to calculate the node coupling coefficient:

[0125]

[0126] Among them, represents the node coupling coefficient, , and represent weights, represents the bandwidth utilization rate, represents the standardized communication delay, represents the data packet loss rate. It should be noted that the higher the bandwidth utilization rate, the more frequent the communication between nodes, and the higher the coupling degree; the smaller the communication delay, the more timely the interaction between nodes, and the higher the coupling degree. Therefore, is used to represent the positive impact of the communication delay on the coupling coefficient; the lower the data packet loss rate, the higher the reliability of data transmission, and the higher the coupling degree between nodes. Therefore, is used to reflect the role of the data packet loss rate on the coupling coefficient.

[0127] 2) The random forest algorithm is used to fit and predict the coupling coefficient between nodes corresponding to each outgoing edge link, and the fault propagation influence intensity of the target node in each outgoing edge link is obtained.

[0128] The random forest algorithm is used to fit the relationship between the coupling coefficient between nodes and the fault propagation influence intensity, and based on this fitting relationship, the coupling coefficient between nodes corresponding to each outgoing edge link is predicted to obtain the fault propagation influence intensity of the corresponding target node.

[0129] 3) Based on the fault propagation influence intensity, the fault propagation depth increment value is determined, and the first fault propagation path subgraph is hierarchically expanded according to the fault propagation depth increment value to obtain the third fault propagation path subgraph.

[0130] The number of target nodes corresponding to the fault propagation influence intensity greater than the third preset threshold is counted and characterized as the fault propagation depth increment value. The first fault propagation path subgraph is hierarchically expanded by depth-first search according to the fault propagation depth increment value to obtain the third fault propagation path subgraph.

[0131] 4) The communication link data between the key nodes and each newly added boundary node in the third fault propagation path subgraph is obtained, and the fault propagation probability between the key nodes and each newly added boundary node in the third fault propagation path subgraph is calculated based on the communication link data.

[0132] Based on the communication link data, the association rule mining algorithm is used to calculate the fault propagation probability between the key nodes and each newly added boundary node in the third fault propagation path subgraph.

[0133] 5) Based on the fault propagation probability, the nodes in the third fault propagation path subgraph are marked to obtain the second fault propagation path subgraph.

[0134] The newly added boundary nodes corresponding to the fault propagation probability greater than the second preset threshold in the third fault propagation path subgraph are marked to obtain the second fault propagation path subgraph.

[0135] In an embodiment of the present invention, a method for tracing the fault path of a distribution communication network based on a knowledge graph can comprehensively integrate the operation status and topological structure information of the distribution communication network by constructing a knowledge graph; according to the multi-source data of communication links, the dependence strength can be accurately obtained, and the fault propagation path can be effectively identified; in the analysis of the fault propagation path graph, the first fault propagation path subgraph is screened based on the preset fault propagation termination condition, and then the key nodes are accurately identified; by using the data of the outgoing edge links of the key nodes, the subgraph is expanded. The present invention greatly improves the accuracy and comprehensiveness of fault analysis, can quickly locate the key fault nodes and propagation paths, and realizes the dynamic balance of the node importance and the inference depth of the knowledge graph.

[0136] Based on the above method for tracing the fault path of a distribution communication network based on a knowledge graph, as Figure 3 shown, an embodiment of the present invention provides a system for tracing the fault path of a distribution communication network based on a knowledge graph, including:

[0137] A dependency strength determination module 1, configured to construct a corresponding knowledge graph based on the operating state and topological structure of the target distribution communication network, and obtain the dependency strength of each communication link in the knowledge graph according to the communication data of each communication link;

[0138] Specifically, the dependency strength determination module includes:

[0139] A first link score determination unit, configured to obtain the first operating state data of each communication link in the knowledge graph, and perform weighted calculation based on the first operating state data to obtain the first link score of the corresponding communication link, where the first operating state data includes communication delay, bandwidth utilization rate, and data packet loss rate;

[0140] A second link score determination unit, configured to perform weighted correction on the first link score based on the physical distance of each communication link to obtain the second link score of the corresponding communication link;

[0141] A service dependency determination unit, configured to obtain the service traffic data of each communication link, and perform weighted calculation based on the service traffic data to obtain the service dependency of the corresponding communication link, where the service traffic data includes control instruction flow, measurement data flow, and alarm information flow;

[0142] A weighted fusion unit, configured to perform weighted fusion on the second link score and service dependency of each communication link to obtain the dependency strength of the corresponding communication link.

[0143] A fault propagation path determination module 2, configured to obtain the set of directed edges of abnormal nodes in the knowledge graph, and identify the fault propagation path in the set of directed edges based on the fault propagation rate calculated according to the dependency strength, where the set of directed edges includes the directed edges between abnormal nodes and each adjacent node;

[0144] Specifically, the fault propagation path determination module includes:

[0145] A linear calculation unit, configured to obtain the dependency strength of each communication link corresponding to the directed edges in the set of directed edges, and perform linear calculation on the dependency strength to obtain the initial fault propagation rate of the corresponding directed edges;

[0146] A weighted calculation unit, configured to obtain the performance index data of each communication link corresponding to the directed edges, and perform weighted calculation based on the performance index data to obtain the edge weight coefficient of the corresponding directed edges, where the performance index data includes physical distance, network hop count, and communication delay;

[0147] A weighted correction unit, configured to perform weighted correction on the initial fault propagation rate based on the edge weight coefficient of each directed edge to obtain the fault propagation rate of the corresponding directed edge;

[0148] A propagation rate judgment unit, configured to judge whether the fault propagation rate is greater than a preset propagation rate threshold, and if so, mark the corresponding directed edge as a fault propagation path.

[0149] A node degree determination module 3, configured to traverse the fault propagation path to generate a fault propagation path graph, and screen the fault propagation path graph based on a preset fault propagation termination condition to obtain a first fault propagation path sub-graph, and determine the degree of each level of nodes in the first fault propagation path sub-graph;

[0150] A key node identification module 4, configured to perform fitting prediction based on the degree of each level of nodes in the first fault propagation path sub-graph to obtain the first node importance of the corresponding node, and identify the key nodes in the first fault propagation path sub-graph according to the first node importance;

[0151] A target fault propagation path determination module 5, configured to obtain the out-edge link data of the key nodes in the first fault propagation path sub-graph, and expand the first fault propagation path sub-graph based on the out-edge link data to obtain a second fault propagation path sub-graph.

[0152] It should be noted that each module in the above-mentioned fault path tracing system for a distribution communication network based on a knowledge graph can be implemented in whole or in part by software, hardware, and their combination. The above-mentioned modules can be embedded in the processor of a computer device in hardware form or be independent of it, or can be stored in the memory of the computer device in software form, so that the processor can call and execute the operations corresponding to the above-mentioned modules. For the specific limitations on a fault path tracing system for a distribution communication network based on a knowledge graph, refer to the limitations on a fault path tracing method for a distribution communication network based on a knowledge graph in the above text. The two have the same functions and effects, and will not be elaborated here.

[0153] In summary, in the embodiment of the present invention, a fault path tracing method and system for a distribution communication network based on a knowledge graph can comprehensively integrate the operation status and topological structure information of the distribution communication network by constructing a knowledge graph; according to the multi-source data of communication links, the strength of the dependence relationship can be accurately obtained, and the fault propagation path can be effectively identified; in the analysis of the fault propagation path graph, a first fault propagation path sub-graph is screened based on a preset fault propagation termination condition, and then the key nodes are accurately identified; by using the out-edge link data of the key nodes, the sub-graph expansion is realized. The present invention greatly improves the accuracy and comprehensiveness of fault analysis, can quickly locate the key fault nodes and propagation paths, and realizes the dynamic balance of the node importance and the reasoning depth of the knowledge graph.

[0154] Each embodiment in this specification is described in a progressive manner. For parts that are the same or similar in each embodiment, reference can be made to each other. Each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple. For related parts, reference can be made to the partial description of the method embodiment. It should be noted that the technical features of the above embodiments can be combined arbitrarily. For the sake of concise description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope described in this specification.

[0155] The above is only the preferred embodiment of the present invention. It should be pointed out that for those of ordinary skill in the art, without departing from the technical principle of the present invention, several improvements and replacements can be made, and these improvements and replacements should also be regarded as the protection scope of the present invention.

Claims

1. A fault path tracing method for a distribution communication network based on a knowledge graph, characterized in that Including: Construct a corresponding knowledge graph based on the operating status and topological structure of the target distribution communication network, and obtain the dependency strength corresponding to each communication link in the knowledge graph according to the communication data of each communication link; Obtain the set of directed edges of abnormal nodes in the knowledge graph, and identify the fault propagation paths in the set of directed edges based on the fault propagation rate calculated based on the dependency strength, where the set of directed edges includes the directed edges between the abnormal nodes and each adjacent node; Traverse the fault propagation paths to generate a fault propagation path graph, and filter the fault propagation path graph based on a preset fault propagation termination condition to obtain a first sub-graph of fault propagation paths, and determine the degree of each hierarchical node in the first sub-graph of fault propagation paths; Perform fitting prediction based on the degree of each hierarchical node in the first sub-graph of fault propagation paths to obtain the first node importance corresponding to the nodes, and identify the key nodes in the first sub-graph of fault propagation paths according to the first node importance; Obtain the out-edge link data of the key nodes in the first sub-graph of fault propagation paths, and expand the first sub-graph of fault propagation paths based on the out-edge link data to obtain a second sub-graph of fault propagation paths; The expanding the first sub-graph of fault propagation paths based on the out-edge link data to obtain a second sub-graph of fault propagation paths includes: Perform weighted calculation based on the out-edge link data to obtain the node coupling coefficient corresponding to each out-edge link, where the out-edge link data includes the communication delay, bandwidth utilization rate, and data packet loss rate of several out-edge links; Use the random forest algorithm to perform fitting prediction on the node coupling coefficient corresponding to each out-edge link to obtain the fault propagation influence strength of the target nodes in each out-edge link; Determine the fault propagation depth increment value based on the fault propagation influence strength, and perform hierarchical expansion on the first sub-graph of fault propagation paths according to the fault propagation depth increment value to obtain a third sub-graph of fault propagation paths; Obtain the communication link data between the key nodes and each newly added boundary node in the third sub-graph of fault propagation paths, and calculate the fault propagation probability between the key nodes and each newly added boundary node in the third sub-graph of fault propagation paths based on the communication link data; Perform node marking on the third sub-graph of fault propagation paths based on the fault propagation probability to obtain a second sub-graph of fault propagation paths, where the node marking includes marking the newly added boundary nodes corresponding to the fault propagation probability greater than the second preset threshold in the third sub-graph of fault propagation paths.

2. The fault path tracing method for a distribution communication network according to claim 1, wherein The obtaining the dependency strength corresponding to each communication link according to the communication data of each communication link in the knowledge graph includes: Obtain the first operating status data of each communication link in the knowledge graph, and perform weighted calculation based on the first operating status data to obtain the first link score corresponding to the communication link, where the first operating status data includes communication delay, bandwidth utilization rate, and data packet loss rate; Weightedly correct the first link score based on the physical distance of each communication link to obtain a second link score corresponding to the communication link; Obtain the service traffic data of each communication link, and perform weighted calculation based on the service traffic data to obtain the service dependency corresponding to the communication link, where the service traffic data includes control instruction flow, measurement data flow, and alarm information flow; Weightedly fuse the second link score and the service dependency of each communication link to obtain the dependency relationship strength corresponding to the communication link.

3. The method for tracing the fault path of a distribution communication network according to claim 1, wherein Before obtaining the set of directed edges of abnormal nodes in the knowledge graph, it further includes: Obtain the second operating state data of each node in the knowledge graph, and calculate the degree of abnormality corresponding to the node based on the second operating state data, where the second operating state data includes response delay time, load rate, and bandwidth utilization rate; Judge whether the degree of abnormality is greater than a preset abnormality determination threshold, and if so, determine that the corresponding node is an abnormal node.

4. The fault path tracing method for a distribution communication network according to claim 1, wherein, Identifying the fault propagation path in the set of directed edges based on the fault propagation rate calculated from the dependency relationship strength includes: Obtain the dependency relationship strength of the communication link corresponding to each directed edge in the set of directed edges, and perform linear calculation on the dependency relationship strength to obtain the initial fault propagation rate corresponding to the directed edge; Obtain the performance index data of the communication link corresponding to each directed edge, and perform weighted calculation based on the performance index data to obtain the edge weight coefficient corresponding to the directed edge, where the performance index data includes physical distance, network hop count, and communication delay; Weightedly correct the initial fault propagation rate based on the edge weight coefficient of each directed edge to obtain the fault propagation rate corresponding to the directed edge; Judge whether the fault propagation rate is greater than a preset propagation rate threshold, and if so, mark the corresponding directed edge as a fault propagation path.

5. The fault path tracing method for a power distribution communication network according to claim 1, wherein Traversing the fault propagation path to generate a fault propagation path graph, and screening the fault propagation path graph based on a preset fault propagation termination condition to obtain a first fault propagation path sub-graph, and determining the degree of each level node in the first fault propagation path sub-graph includes: Use the breadth-first search algorithm to traverse the fault propagation path to obtain a fault propagation path graph; Use the depth-first search algorithm to obtain the number of levels of the fault propagation path graph; Based on the total number of nodes and the network maximum diameter of the target distribution communication network, calculate a preset fault propagation depth reflecting the network scale of the target distribution communication network; Judge whether the fault propagation path graph meets the preset fault propagation termination condition based on the number of levels and the preset fault propagation depth. If so, screen the fault propagation path graph to obtain a first fault propagation path sub-graph, otherwise characterize the fault propagation path graph as the first fault propagation path sub-graph, where the preset fault propagation termination condition includes that the number of levels is greater than the preset fault propagation depth; Count the in-degree and out-degree of each level node in the first fault propagation path sub-graph.

6. The fault path tracing method for a distribution communication network according to claim 1, wherein Performing fitting prediction based on the degrees of nodes at each level in the first fault propagation path sub-graph to obtain the first node importance of corresponding nodes, and identifying key nodes in the first fault propagation path sub-graph according to the first node importance, including: Performing weighted calculation on the degrees of nodes at each level in the first fault propagation path sub-graph to obtain the node feature values of corresponding nodes; Using the random forest algorithm to perform fitting prediction on the node feature values of nodes at each level in the first fault propagation path sub-graph to obtain the first node importance of corresponding nodes; Performing eigenvector centrality evaluation on nodes at each level in the first fault propagation path sub-graph, and characterizing the obtained network centrality as the second node importance of corresponding nodes; Performing weighted calculation on the first node importance and the second node importance to obtain the node importance of corresponding nodes; Judging whether the node importance is higher than a first preset threshold, and if so, marking the corresponding node as a key node in the first fault propagation path sub-graph.

7. A fault path tracing system for a distribution communication network based on a knowledge graph, characterized in that, Including: A dependency strength determination module, configured to construct a corresponding knowledge graph based on the operating state and topological structure of the target distribution communication network, and obtain the dependency strength of each communication link according to the communication data of each communication link in the knowledge graph; A fault propagation path determination module, configured to obtain the set of directed edges of abnormal nodes in the knowledge graph, and identify the fault propagation paths in the set of directed edges based on the fault propagation rate calculated according to the dependency strength, where the set of directed edges includes the directed edges between the abnormal nodes and each adjacent node; A node degree determination module, configured to traverse the fault propagation paths to generate a fault propagation path graph, and screen the fault propagation path graph based on a preset fault propagation termination condition to obtain a first fault propagation path sub-graph, and determine the degrees of nodes at each level in the first fault propagation path sub-graph; A key node identification module, configured to perform fitting prediction based on the degrees of nodes at each level in the first fault propagation path sub-graph to obtain the first node importance of corresponding nodes, and identify key nodes in the first fault propagation path sub-graph according to the first node importance; A target fault propagation path determination module, configured to obtain the out-edge link data of the key nodes in the first fault propagation path sub-graph, and expand the first fault propagation path sub-graph based on the out-edge link data to obtain a second fault propagation path sub-graph; The expanding the first fault propagation path sub-graph based on the out-edge link data to obtain a second fault propagation path sub-graph includes: Performing weighted calculation based on the out-edge link data to obtain the node coupling coefficient corresponding to each out-edge link, where the out-edge link data includes the communication delay, bandwidth utilization rate, and data packet loss rate of a plurality of out-edge links; Using the random forest algorithm to perform fitting prediction on the node coupling coefficient corresponding to each out-edge link to obtain the fault propagation influence strength of the target node in each out-edge link; Determine the fault propagation depth increment value based on the fault propagation influence intensity, and hierarchically expand the first fault propagation path subgraph according to the fault propagation depth increment value to obtain a third fault propagation path subgraph; Obtain the communication link data between the key node and each newly added boundary node in the third fault propagation path subgraph, and calculate the fault propagation probability between the key node and each newly added boundary node in the third fault propagation path subgraph based on the communication link data; Perform node marking on the third fault propagation path subgraph based on the fault propagation probability to obtain a second fault propagation path subgraph, where the node marking includes marking the newly added boundary nodes corresponding to the fault propagation probability greater than a second preset threshold in the third fault propagation path subgraph.

8. The fault path tracing system for a distribution communication network according to claim 7, wherein The dependency strength determination module includes: The first link score determination unit is used to obtain the first operating state data of each communication link in the knowledge graph, and perform weighted calculation based on the first operating state data to obtain the first link score corresponding to the communication link, where the first operating state data includes communication delay, bandwidth utilization rate, and data packet loss rate; The second link score determination unit is used to perform weighted correction on the first link score based on the physical distance of each communication link to obtain the second link score corresponding to the communication link; The service dependency determination unit is used to obtain the service traffic data of each communication link, and perform weighted calculation based on the service traffic data to obtain the service dependency corresponding to the communication link, where the service traffic data includes control instruction flow, measurement data flow, and alarm information flow; The weighted fusion unit is used to perform weighted fusion on the second link score and the service dependency of each communication link to obtain the dependency strength corresponding to the communication link.

9. The fault path tracing system for a power distribution communication network according to claim 7, wherein, The fault propagation path determination module includes: The linear calculation unit is used to obtain the dependency strength of each directed edge corresponding communication link in the directed edge set, and perform linear calculation on the dependency strength to obtain the initial fault propagation rate corresponding to the directed edge; The weighted calculation unit is used to obtain the performance index data of each directed edge corresponding communication link, and perform weighted calculation based on the performance index data to obtain the edge weight coefficient corresponding to the directed edge, where the performance index data includes physical distance, network hop count, and communication delay; The weighted correction unit is used to perform weighted correction on the initial fault propagation rate based on the edge weight coefficient of each directed edge to obtain the fault propagation rate corresponding to the directed edge; The propagation rate judgment unit is used to judge whether the fault propagation rate is greater than a preset propagation rate threshold, and if so, mark the corresponding directed edge as a fault propagation path.

Images