Electric power communication maintenance conflict identification method and device based on knowledge graph

Abstract

The invention discloses an electric power communication maintenance conflict identification method and device based on a knowledge graph, relates to the technical field of information, and mainly aims to improve the identification efficiency and the identification accuracy of electric power communication maintenance conflicts. The method comprises the following steps: acquiring electric power communication maintenance text information; entities are extracted from the electric power communication maintenance text information, entity relations among the entities are mined, the electric power communication maintenance knowledge graph is constructed based on the entities and the entity relations, the entities serve as nodes in the electric power communication maintenance knowledge graph, and the nodes are used as nodes in the electric power communication maintenance knowledge graph. The entity relationship is used as an edge between nodes in the electric power communication maintenance knowledge graph; and based on the electric power communication maintenance knowledge graph, identifying electric power communication maintenance conflicts. The method is suitable for scenes of conflict identification of electric power communication maintenance.

Description

Technical Field

[0001] The present invention relates to the field of information technology, and in particular to a method and device for identifying power communication maintenance conflicts based on a knowledge graph. Background Art

[0002] Power communication networks are a key component of the power system, crucial for power transmission and dispatch, and closely intertwined with people's daily lives. With rapid economic development, electricity demand is increasing. To meet the demands of daily power transmission and supply, power communication networks are expanding in scale and complexity. Consequently, maintenance of power communication networks is becoming increasingly difficult. Currently, maintenance of power communication networks is plagued by the frequent occurrence of various types of maintenance conflicts. Therefore, to ensure smooth maintenance, proactive identification of maintenance conflicts is necessary.

[0003] Currently, maintenance conflicts are typically identified through manual review of maintenance tickets. However, this manual review process is time-consuming and labor-intensive, resulting in low efficiency in identifying power and communication maintenance conflicts. Furthermore, due to negligence or varying technical skills, the accuracy of identifying power and communication maintenance conflicts can be low. Summary of the Invention

[0004] The present invention provides a method and device for identifying power communication maintenance conflicts based on knowledge graph, which mainly aims to improve the recognition efficiency and accuracy of power communication maintenance conflicts.

[0005] According to a first aspect of the present invention, a method for identifying power communication maintenance conflicts based on a knowledge graph is provided, comprising:

[0006] Obtain power communication maintenance text information;

[0007] Extracting entities from the power communication maintenance text information and mining the relationships between the entities, and constructing a power communication maintenance knowledge graph based on the entities and entity relationships, wherein the entities serve as nodes in the power communication maintenance knowledge graph, and the entity relationships serve as edges between the nodes in the power communication maintenance knowledge graph;

[0008] Based on the power communication maintenance knowledge graph, power communication maintenance conflicts are identified.

[0009] Optionally, entities are extracted from the power communication maintenance text information, including:

[0010] Obtaining a preset entity extraction model, wherein the preset entity extraction model includes a text encoder for extracting feature vectors, a deep self-attention network for learning context-dependent vector representations, and a conditional random field for labeling; the preset entity extraction model is pre-trained based on a sample text dataset;

[0011] The electric power communication maintenance text information is input into the preset entity extraction model, the feature vector of the electric power communication maintenance text information is extracted by the text encoder to obtain a feature vector, the text context feature is extracted from the feature vector by the deep self-attention network to obtain a context feature, and the context feature is labeled by the conditional random field to obtain a sequence annotation label, wherein the sequence annotation label includes each entity and its corresponding entity category.

[0012] Optionally, mining entity relationships between the entities includes:

[0013] Obtaining a preset relationship recognition model, wherein the preset relationship recognition model includes an input representation layer for information embedding, a low-level feature extraction layer for low-level feature extraction, a dynamic routing capsule network feature clustering layer with a word attention mechanism for global relationship feature extraction, and a relationship prediction layer for entity relationship prediction; the preset relationship recognition model is pre-trained based on an entity text dataset;

[0014] The electric power communication maintenance text information and each of the entities are input into the preset relationship recognition model, and the electric power communication maintenance text information is subjected to word embedding, character embedding, part-of-speech embedding, and word position embedding through the input representation layer, and the word embedding, character embedding, part-of-speech embedding, and word position embedding are fused to obtain a fused embedding, and low-level feature extraction is performed on the fused embedding through the low-level feature extraction layer to obtain low-level features, and feature extraction is performed on the low-level features through the dynamic routing capsule network feature clustering layer of the word attention mechanism to obtain global relationship features, and relationship prediction is performed on the global relationship features through the relationship prediction layer to obtain entity relationships between the entities.

[0015] Optionally, the low-level feature extraction layer includes a bidirectional long short-term memory neural network, and the bidirectional long short-term memory neural network includes a forward network layer and a backward network layer;

[0016] The step of extracting low-level features from the fused embedding through the low-level feature extraction layer to obtain low-level features includes:

[0017] The fusion is embedded and input into the bidirectional long short-term memory neural network to perform global sequence feature extraction to obtain the forward global sequence features corresponding to the forward network layer and the backward global sequence features corresponding to the backward network layer, and the forward global sequence features and the backward global sequence features are fused to obtain the low-level features.

[0018] Optionally, the dynamic routing capsule network feature clustering layer of the word attention mechanism includes: a low-layer capsule network, a word attention mechanism, and a high-layer capsule network;

[0019] The dynamic routing capsule network feature clustering layer of the word attention mechanism extracts features from the low-level features to obtain global relationship features, including:

[0020] The low-level features are input into the low-level capsule network for feature extraction to obtain local features, the local features are input into the word attention mechanism for weight distribution to obtain weighted local features, and the weighted local features are input into the high-level capsule network for feature extraction to obtain global relationship features.

[0021] Optionally, the identifying power communication maintenance conflicts based on the power communication maintenance knowledge graph includes:

[0022] Based on the power communication maintenance knowledge graph, performing at least one of time conflict identification, resource conflict identification, and logic conflict identification on the power communication maintenance;

[0023] After identifying power communication maintenance conflicts based on the power communication maintenance knowledge graph, the method further includes:

[0024] Provide early warning for identified power communication maintenance conflicts.

[0025] Optionally, before identifying power communication maintenance conflicts based on the power communication maintenance knowledge graph, the method further includes:

[0026] The nodes and edges in the power communication maintenance knowledge graph are converted into vector representations, and the multi-head attention mechanism is used to calculate the vector representation of the power communication maintenance knowledge graph. In the multi-head attention mechanism, multiple attention heads are set, and each head pays attention to and calculates the vectors of the power communication maintenance knowledge graph from different perspectives. The output results of multiple attention heads are spliced and linearly transformed to obtain the power communication maintenance knowledge graph with attention weights.

[0027] According to a second aspect of the present invention, a knowledge graph-based power communication maintenance conflict identification device is provided, comprising:

[0028] An acquisition unit, used for acquiring power communication maintenance text information;

[0029] A construction unit is configured to extract entities from the power communication maintenance text information and mine entity relationships between the entities, and construct a power communication maintenance knowledge graph based on the entities and entity relationships, wherein the entities serve as nodes in the power communication maintenance knowledge graph, and the entity relationships serve as edges between the nodes in the power communication maintenance knowledge graph;

[0030] An identification unit, configured to identify power communication maintenance conflicts based on the power communication maintenance knowledge graph

[0031] According to a third aspect of the present invention, there is provided a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-mentioned knowledge graph-based power communication maintenance conflict identification method.

[0032] According to a fourth aspect of the present invention, a computer device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above-mentioned knowledge graph-based power communication maintenance conflict identification method when executing the program.

[0033] According to a method and device for identifying power communication maintenance conflicts based on a knowledge graph provided by the present invention, compared with the current method of determining maintenance conflicts by manually reviewing maintenance tickets, the present invention obtains power communication maintenance text information, extracts entities from the power communication maintenance text information, and mines the entity relationships between the entities, and constructs a power communication maintenance knowledge graph based on the entities and the entity relationships, wherein the entities serve as nodes in the power communication maintenance knowledge graph, and the entity relationships serve as edges between the nodes in the power communication maintenance knowledge graph; finally, based on the power communication maintenance knowledge graph, power communication maintenance conflicts are identified. Therefore, through the structured text information of power communication, a power communication maintenance knowledge graph is constructed, and the attention mechanism is combined to improve the efficiency of conflict identification. Finally, the power communication maintenance conflicts are identified based on the power communication maintenance knowledge graph without human participation, that is, there is no need to manually check the maintenance order item by item, thereby improving the recognition automation and efficiency of power communication maintenance conflicts. At the same time, based on the knowledge graph, complex entity relationships can be accurately identified, such as the dependency relationship between equipment, the overlapping relationship of maintenance time, etc., while there may be recognition errors and negligence when manually identifying these complex relationships. Therefore, the present invention can improve the recognition accuracy of power communication maintenance conflicts. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of this application. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0035] Figure 1 A flow chart of a method for identifying power communication maintenance conflicts based on a knowledge graph provided by an embodiment of the present invention is shown;

[0036] Figure 2 A schematic structural diagram of a preset relationship recognition model provided by an embodiment of the present invention is shown;

[0037] Figure 3 A flow chart of another method for identifying power communication maintenance conflicts based on a knowledge graph provided by an embodiment of the present invention is shown;

[0038] Figure 4 A schematic diagram of the structure of a power communication maintenance conflict identification device based on a knowledge graph provided by an embodiment of the present invention is shown;

[0039] Figure 5 A schematic structural diagram of another power communication maintenance conflict identification device based on a knowledge graph provided by an embodiment of the present invention is shown;

[0040] Figure 6 A schematic diagram of the physical structure of a computer device provided by an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0041] The present invention will be described in detail below with reference to the accompanying drawings and in combination with embodiments. It should be noted that, unless there is a conflict, the embodiments and features in the embodiments of the present application can be combined with each other.

[0042] Currently, manually reviewing maintenance tickets to identify maintenance conflicts is time-consuming and labor-intensive. In addition, due to negligence of staff or uneven technical levels, the accuracy of identifying power communication maintenance conflicts may be low.

[0043] In order to solve the above problems, the embodiment of the present invention provides a method for identifying power communication maintenance conflicts based on knowledge graph, such as Figure 1 As shown, the method includes:

[0044] 101. Obtain text information on power communication maintenance.

[0045] Specifically, images of bills such as power communication maintenance bills can be converted into power communication maintenance text information. The images of bills such as power communication maintenance bills can be obtained by scanning paper documents, forms, etc. for power maintenance.

[0046] 102. Extract entities from the text information of electric power communication maintenance and mine the entity relationships between entities. Based on the entities and entity relationships, construct a knowledge graph of electric power communication maintenance, where each entity serves as a node in the knowledge graph of electric power communication maintenance, and the entity relationship serves as an edge between each node in the knowledge graph of electric power communication maintenance.

[0047] For an embodiment of the present invention, after obtaining the electric power communication maintenance text information, it is necessary to construct an electric power communication maintenance knowledge graph based on the text information. Before constructing the electric power communication maintenance knowledge graph, it is necessary to extract entities in the electric power communication maintenance text information. Based on this, step 102 specifically includes: obtaining a preset entity extraction model, wherein the preset entity extraction model includes a text encoder for extracting feature vectors, a deep self-attention network for learning context-related vector representations, and a conditional random field for labeling; the preset entity extraction model is pre-trained based on a sample text data set; the electric power communication maintenance text information is input into the preset entity extraction model, the feature vector of the electric power communication maintenance text information is extracted by the text encoder to obtain a feature vector, the text context feature is extracted by the deep self-attention network to obtain a context feature, and the context feature is identified by the conditional random field to obtain a sequence labeling label, wherein the sequence labeling label contains each entity and its corresponding entity category.

[0048] The pre-set entity extraction model consists of a text encoder, a deep self-attention network, and a conditional random field. To improve its extraction accuracy, the model must be trained and constructed. First, an initial model is constructed. A sample power and telecommunications maintenance text dataset with entity annotations is obtained and divided into training and test data. The initial model is trained with the training data and then tested with the test data. The model that meets the test conditions is the pre-set entity extraction model, which can be used to extract key entities and attributes from power and telecommunications maintenance text. Specifically, the deep self-attention network understands the context of the text. Trained on a large corpus of text, it has a large capacity and strong expressiveness, and performs well in downstream tasks. Within the named entity recognition framework of a text encoder and conditional random field, the text encoder produces features, the deep self-attention network learns the context vector, and the conditional random field generates sequence annotation labels. These labels are used to extract entities from the power and telecommunications maintenance text. The sequence annotation labels identify each entity and its corresponding entity category.

[0049] Furthermore, after extracting the entities, in order to construct the knowledge graph, it is also necessary to mine the entity relationships between the entities. Based on this, step 102 includes: obtaining a preset relationship recognition model, wherein the preset relationship recognition model includes an input representation layer for information embedding, a low-level feature extraction layer for low-level feature extraction, a dynamic routing capsule network feature clustering layer of a word attention mechanism for global relationship feature extraction, and a relationship prediction layer for entity relationship prediction; the preset relationship recognition model is pre-trained based on an entity text data set; the power communication maintenance text information and each of the entities are input into the preset In the relationship recognition model, the power communication maintenance text information is word-embedded, character-embedded, part-of-speech-embedded, and word-position-embedded through the input representation layer, and the word embedding, character embedding, part-of-speech embedding, and word-position embedding are fused to obtain a fused embedding. The fused embedding is subjected to low-level feature extraction through the low-level feature extraction layer to obtain low-level features. The low-level features are subjected to feature extraction through the dynamic routing capsule network feature clustering layer of the word attention mechanism to obtain global relationship features. The global relationship features are subjected to relationship prediction through the relationship prediction layer to obtain entity relationships between the entities. The low-level feature extraction layer includes a bidirectional long short-term memory neural network, which includes a forward network layer and a backward network layer. The method for extracting low-level features from the fused embedding through the low-level feature extraction layer to obtain low-level features includes: inputting the fused embedding into the bidirectional long short-term memory neural network to extract global sequence features, obtaining forward global sequence features corresponding to the forward network layer and backward global sequence features corresponding to the backward network layer, fusing the forward global sequence features and the backward global sequence features to obtain the low-level features. In another embodiment of the present invention, the dynamic routing capsule network feature clustering layer of the word attention mechanism includes: a low-level capsule network, a word attention mechanism, and a high-level capsule network. The method for extracting global relationship features from the low-level features through the dynamic routing capsule network feature clustering layer of the word attention mechanism includes: inputting the low-level features into the low-level capsule network for feature extraction to obtain local features, inputting the local features into the word attention mechanism for weight assignment to obtain weighted local features, and inputting the weighted local features into the high-level capsule network for feature extraction to obtain global relationship features.

[0050] In an embodiment of the present invention, in order to improve the recognition accuracy of the preset relationship recognition model, it is first necessary to train and construct the preset relationship recognition model. Based on this, the method includes: constructing an initial relationship recognition model, and obtaining an entity text dataset, wherein the entity text dataset includes sample power communication maintenance text information with relationship annotation labels and each sample entity in the sample power communication maintenance text information; dividing the entity text dataset into training data and test data, using the training data to train the initial relationship recognition model, and using the test data to test the trained initial relationship recognition model, and finally determining the initial relationship recognition model that meets the test conditions as the preset relationship recognition model. Furthermore, entity relationship mining refers to extracting the relationship between two entities in text information, and finally forming a graph. In the embodiment of the present invention, the association between maintenance equipment and maintenance type, the association between optical path and interruption of business, etc. are mainly identified. The specific implementation method includes: in order to effectively identify multiple relationships from power communication maintenance text information and solve the problem of keyword sparsity, a capsule network based on word attention mechanism is proposed to realize knowledge relationship recognition. The structure of the preset relationship recognition model is as follows: Figure 2 The proposed model is a variation of the original capsule network. The architecture of the model consists of four layers: input representation layer, low-level feature extraction layer, dynamic routing capsule network feature clustering based on word attention mechanism, and relationship prediction layer. Figure 2 WE and CE represent word embedding and character embedding respectively, POS represents the part of speech of a word, P represents the position of a word in a sentence, U1, U2, ..., U LRepresent the outputs of the low-level capsule network (features of each entity), S1, S2, S3...represent the features of each entity after noise reduction, V1, V2, V3...represent the outputs of the feature clustering layer of the dynamic routing capsule network based on the word attention mechanism, and squashing is a specific function that performs nonlinear transformations on the outputs of the capsule. Y1, Y2, Y3...represent the outputs of the relationship prediction layer. Input representation layer: The model's input representation is a combination of four embeddings: character embedding, word embedding, part-of-speech embedding, and position embedding. This method captures the semantic information of each character and word, as well as the context of the sentence and the word order of the words within the sentence. Low-level feature extraction layer: The word representation of the input layer is passed through the forward and backward networks of the bidirectional long short-term memory neural network to obtain global sequence information, thereby capturing the low-level features of the sentence. For example, if the input sentence describes information about maintenance equipment for a power communication system, this step can better extract the various attributes of this equipment. The feature clustering layer uses a capsule network with dynamic routing based on a word-attention mechanism to reduce noise generated by irrelevant words and extract high-level features. This allows for the capture of multiple relationships between entities. For example, if the input contains multiple entities and their characteristics, such as power communication system maintenance equipment, its attributes, and maintenance types, this step effectively captures the relationships between other maintenance equipment and maintenance types, thus capturing multiple relationships between entities. The relationship prediction layer uses a marginal loss function to predict relationships between entities. Specifically, the entity relationship mining method using a pre-set relationship recognition model is as follows: Based on low-level feature extraction: The low-level feature extraction layer can be a bidirectional long short-term memory neural network. Bidirectional long short-term memory neural networks are effective for modeling sequential data in relationship extraction tasks. Therefore, a bidirectional long short-term memory neural network algorithm, consisting of a forward network and a backward network, is used to obtain global sequence information, thereby acquiring low-level features of the sentence. In the relationship extraction task, the embedding combination of the input representation layer is first input into the bidirectional long short-term memory neural network. Then, the forward and backward networks of the bidirectional long short-term memory neural network are used to simultaneously obtain global sequence features. Finally, the global sequence feature t is used as the output of the bidirectional long short-term memory neural network and as the input of the first-layer capsule network. Enriched feature extraction based on capsule networks: To further obtain the relevance of candidate relations and entities and address the problem of multiple relations in sentences, a capsule network framework for multiple relation extraction is proposed. Capsule networks use neuron vectors trained by a dynamic routing algorithm to implement model training. The capsule outputs a vector that richly expresses the features extracted from the node. Different dimensions in the vector are used to record different attributes of the same feature, improving the expressiveness of the text. Each directional capsule vector can effectively capture the implicit grammatical, semantic, and word order features of the text. Capsule networks are used to extract relevance features between relations and entities.Capsule networks are divided into two parts for feature extraction: low-level capsules and high-level capsules. Low-level capsules contain local and trivial features, while high-level capsules are aggregated from low-level capsules. Active capsules use a transition matrix to predict the instantiation parameters of high-level capsules. When multiple predictions match, higher-level capsules are activated. Low-level semantic information captured by a bidirectional long short-term memory neural network is input into the first-layer capsules. The current layer's capsules then use a nonlinear squeezing function to compress the length of each low-level capsule. Dynamic Routing Capsule Network Feature Clustering Layer Based on a Word Attention Mechanism: Using capsule networks, high-level capsules representing relational features can be obtained. High-level capsule networks can more richly express relational features. However, due to the long source sequences in power and telecommunications maintenance text, the noise generated by irrelevant words leads to significant gradient update attenuation, making it impossible to effectively update sequence parameters. This paper proposes introducing a word attention mechanism into the dynamic routing of capsule networks to extract informative words related to inter-entity relationships in sentences and aggregate the representations of these words into a sentence vector, thereby addressing the sparsity of informative words in long sentences. The word attention mechanism aims to ignore irrelevant words and focus on informative words, with informative words being given a higher weight coefficient than irrelevant words. The word attention mechanism assigns different weights to different words, that is, informative words have higher weights, while irrelevant words have lower weights, which can increase the strength of information words. The word attention mechanism is used to combine local information at a higher level through local perception, which helps to reduce useless information in relationship extraction. Dynamic routing algorithm based on word attention mechanism: In order to further capture the relevance and important features of the context and reduce the attenuation of useful information in long sentences, a dynamic routing algorithm based on the word attention mechanism is designed to iteratively correct the connection strength. The word attention mechanism enables the decoder to obtain important input information to solve the problem of keyword sparsity. That is, the embodiment of the present invention ultimately inputs the power communication maintenance text information and each of the entities into the preset relationship recognition model, and can realize the mining of entity relationships through operations such as the input representation layer, the low-level feature extraction layer, the dynamic routing capsule network feature clustering layer of the word attention mechanism, and the feature extraction of the relationship prediction layer in the preset relationship recognition model. For example, entity relationships include carrying relationships, which describe the relationship between optical paths and optical cables. For example, optical path A was originally carried by optical cable B, and during maintenance, it was detoured to a transitional detour optical cable from A to B to C to D; time association relationships, which describe the relationship between maintenance tasks and time. For example, the communication maintenance task of optical cable A was carried out between February 14 and February 18; and conflict relationships, which are used to identify the relationship between entities that have conflicts. This relationship can be subsequently determined and established based on the judgment rules.

[0051] Furthermore, after extracting the entities in the power communication maintenance text information and mining the entity relationships between the entities, the entities are used as nodes of the power communication maintenance knowledge graph to be constructed, and the entity relationships are used as edges between the corresponding nodes. Finally, the power communication maintenance knowledge graph is constructed based on the edges between the nodes.

[0052] Furthermore, when building a knowledge graph in the power communication maintenance scenario, considering specific maintenance time periods, optical cables, optical paths, and services, the graph becomes voluminous, increasing the computational complexity and traversal effort. Therefore, we introduce an attention mechanism to focus on key information to improve efficiency and accuracy. Power communication systems are dynamic, with new equipment, services, or maintenance tasks constantly emerging, necessitating updates to the knowledge graph. The attention mechanism can better adapt to this dynamic dynamic, flexibly adjusting the focus on different information based on current circumstances. When a new maintenance task is added, the attention mechanism can quickly identify the relevant parts of the existing knowledge graph, avoiding the need to recalculate and analyze the entire graph. For example, when inspecting optical cable A, the status information of the connected optical terminal equipment and other service optical paths transmitting signals through the cable can have a significant impact on the planning and risk assessment of the current maintenance task.

[0053] In another embodiment of the present invention, it is also possible to jointly train models related to the knowledge graph (such as a preset entity extraction model, a preset relationship recognition model, and a knowledge graph embedding model) with a neural network model. Through joint optimization, different models can work together better and complement each other's information. For example, during the training process, the knowledge graph model provides a more accurate structured knowledge representation for the neural network, and the feedback from the neural network can help the knowledge graph model further optimize the extraction and representation of entities and relationships, thereby improving the performance of the entire conflict analysis method.

[0054] 103. Based on the power communication maintenance knowledge graph, power communication maintenance conflicts are identified.

[0055] According to the embodiment of the present invention, after the electric power communication maintenance knowledge graph is constructed, the electric power communication maintenance conflicts are finally identified based on the electric power communication maintenance knowledge graph.

[0056] For example, time conflict judgment: the time information of the maintenance task needs to be extracted from the knowledge graph. For example, for the maintenance task of disconnecting the communication of optical cable A, its maintenance time is obtained as February 14th to February 18th; traverse other maintenance tasks, that is, traverse all maintenance task entities in the knowledge graph, and for each maintenance task, obtain its corresponding time information; judge time conflict, that is, for each traversed maintenance task, check whether its time overlaps with the time of the maintenance task of disconnecting the communication of optical cable A. If there is an overlap, establish a conflict relationship between the two maintenance tasks in the knowledge graph and mark it as a time conflict; for example, if it is found that the time of another maintenance task that may cause the planned interruption of the detour optical cable is February 16th to February 20th, then it has a time conflict with the maintenance task of disconnecting the communication of optical cable A, and establish a conflict relationship between them in the knowledge graph.

[0057] Resource conflict determination: (1) Determine the transition optical cable and related optical paths, determine the transition detour optical cable entity from the knowledge graph, and the optical path entity carried on the transition optical cable during the maintenance period, traverse other optical paths and service adjustment information, and traverse all optical path entities, maintenance tickets, method orders, transition plans and other related information in the knowledge graph. For each optical path entity, check whether there is an adjustment operation (such as adjusting the other port protection service of the AB port and its optical path to the optical cable route, or adjusting another single-port security control service in the same direction to the optical cable route). (2) If it is found that the adjustment operation of a certain optical path entity will cause it to be carried on the transition detour optical cable, and the optical cable has already carried other optical paths that are not allowed to be adjusted during the maintenance period, then establish a conflict relationship between the optical path entity and the transition detour optical cable in the knowledge graph and mark it as a resource conflict. For example, if a maintenance ticket is found to plan to adjust the optical path of another protection service of port AB protection to a transitional detour optical cable, this will cause the optical cable to carry the AB port channels of the same set of protection services simultaneously from February 14 to February 18, which will exceed 8 hours in maintenance mode and pose an operational risk. In this case, a resource conflict is determined to exist, and a corresponding conflict relationship is established in the knowledge graph.

[0058] Logical conflict determination: Logical conflicts are usually determined by rules, time resources and relationship consistency. Specifically, (1) Rule-based conflict determination: Some logical rules are pre-set in the knowledge graph. For example, "If the maintenance plan clearly states that a certain optical path carries a specific business, then the affected business module of the maintenance ticket should include these businesses", or "In the same maintenance time period, the same optical cable cannot be in both normal operation and disconnection at the same time". When new information is entered into the knowledge graph, the system will check whether the relationship and attributes between entities are logical according to these rules. If not, it is determined that there is a logical conflict. For example, in the above example, the maintenance plan indicates that the optical path carries protection and security control business, but the affected business module of the maintenance ticket does not list security control business, which violates the rules and is determined to be a conflict. (2) Time and resource conflict detection: For conflicts involving time and resources, the relationship between time entities and resource entities such as equipment and optical paths in the knowledge graph is analyzed to determine whether different maintenance tasks overlap in time, or whether the same resource (such as a section of optical cable) is occupied by multiple contradictory operations at the same time. If two maintenance tasks are found in the knowledge graph to perform different operations on the same section of optical cable within the same time period (e.g., one task is to disconnect the optical cable, while the other task is to perform normal communication on the optical cable), then a time and resource conflict is determined to exist. (3) Relationship consistency check: Verify the consistency of the relationships between entities in the knowledge graph. For example, check whether the carrying relationship between the optical path and the business, the relationship between the maintenance task and the affected business, etc. are reasonable. If a business is found to be carried by a certain optical path, but there is no corresponding carrying relationship between the optical path and the business in the knowledge graph, or the relationship is incomplete, then a logical conflict is determined to exist.

[0059] In another embodiment of the present invention, in order to improve the accuracy of power communication maintenance conflict identification, it is also necessary to optimize the power communication maintenance knowledge graph. Based on this, the method includes: converting the nodes and edges in the power communication maintenance knowledge graph into vector representations, and using a multi-head attention mechanism to calculate the vector representation of the power communication maintenance knowledge graph; in the multi-head attention mechanism, multiple attention heads are set, and each head pays attention to and calculates the vectors of the power communication maintenance knowledge graph from different perspectives; the output results of multiple attention heads are spliced and linearly transformed to obtain a power communication maintenance knowledge graph with attention weights.

[0060] Specifically, the multimodal knowledge graph vector initialization utilizes an extended Heterogeneous Graph Attention Network (HGAT) model to fuse graph structure and semantic features, providing targeted processing for various entities and relationships within the power communication knowledge graph. For example, in the case of optical cable disconnection and maintenance, the multimodal knowledge graph vector initialization requires more targeted conflict detection. For the entity of the transitional optical cable, the vector generation must not only consider its topological connectivity with surrounding equipment (such as optical transmission equipment at both ends and surrounding optical cables) and the functional semantics of the cable itself (transmission capacity, type of communication services carried), but also consider factors that may cause conflicts. For example, the correlation between the services carried by the optical cable in the transitional stage is analyzed. A high correlation may indicate that the services carried by the optical cable in the transitional stage lack N-1 robustness during maintenance. For maintenance tasks, when generating vectors by integrating task process logic (such as the optical path detour stage, the optical cable disconnection and maintenance stage, and the optical path restoration stage), potential conflicts between each stage and other work order tasks must be considered. For example, during the optical path detour and restoration phase, it is determined whether it will affect the normal communication of surrounding equipment and conflict with other maintenance work on the detour route. During the optical cable disconnection and maintenance phase, it is assessed whether it will conflict with other maintenance tasks or methods, and whether related work will interfere with the ongoing temporary detour service. Furthermore, position coding is introduced to record node structure location information, such as the location of a specific line optical cable in the network topology. This, combined with the location of potentially affected surrounding equipment and optical cables, further enhances the integrity of the vector representation in conflict judgment.

[0061] In terms of adaptive attention head allocation, under the framework of the multi-head attention mechanism, based on task types such as optical cable disconnection and maintenance, when using reinforcement learning algorithms to train the attention head allocator, it is necessary to conduct a more in-depth analysis of conflict-related factors. A detailed study of historical conflict cases similar to optical cable disconnection and maintenance should be conducted, focusing not only on the temporal correlation between resources and maintenance tasks and the resource allocation relationship, but also on the conflict situations at each ticket submission stage. For example, for the conflicts that may arise at the different stages of optical path detour, optical cable disconnection, and restoration of the original optical path, it is necessary to intelligently assign weights to different attention heads based on conflict judgments, so that they can accurately focus on the key conflict points.

[0062] Hierarchical attention computation must be closely centered around conflict determination. During coarse-grained global attention computation, the attention mechanism is used to quickly locate key subgraph regions relevant to the optical cable maintenance task. This not only identifies the subgraphs containing potentially conflicting equipment groups, but also considers the potential conflicting regions involved in each ticketing phase. For example, during the optical path detour and restoration phases, attention is focused on areas that may interfere with communications with surrounding devices; during the optical cable maintenance phase, attention is focused on areas that may impact the normal operation of the detour services. Fine-grained local attention computation is then performed within the subgraph regions. By introducing residual connections to avoid information loss, the algorithm accurately captures detailed information and complex relationships within the subgraph, focusing specifically on conflict-related information such as the connection status between devices within the subgraph, the traffic load on different optical paths, and the impact of operations at each ticketing phase on these states and loads. Finally, the global and local computation results are combined and weighted to produce a final knowledge graph representation with attention weights, ensuring that conflict determination balances the overall network structure with key local details.

[0063] The dynamic attention weight update process requires real-time monitoring of changes in the attributes of key nodes in the knowledge graph related to conflict judgment and new relationship data. Online learning algorithms and backpropagation algorithms are used to instantly update attention weights and adjust model parameters so that the weights promptly reflect the possibility of conflict under the latest power communication system status. At the same time, due to the modeling of the specific time period, optical cable, optical path, and business entity for each maintenance, the amount of graph information increases. In each of the above steps, the algorithm needs to be optimized to reduce the computational complexity and complexity of the traversal operation. For example, a pruning strategy is used in global attention calculations to reduce the unnecessary subgraph search range and focus only on areas where conflicts may arise; in local attention calculations, the calculation method of residual connections is optimized to improve computational efficiency, allowing for more efficient conflict judgment.

[0064] According to a method for identifying power communication maintenance conflicts based on a knowledge graph provided by the present invention, compared with the current method of determining maintenance conflicts by manually reviewing maintenance tickets, the present invention obtains power communication maintenance text information; then extracts entities from the power communication maintenance text information and mines the entity relationships between the entities, and constructs a power communication maintenance knowledge graph based on the entities and the entity relationships, wherein the entities serve as nodes in the power communication maintenance knowledge graph, and the entity relationships serve as edges between the nodes in the power communication maintenance knowledge graph; finally, based on the power communication maintenance knowledge graph, power communication maintenance conflicts are identified. Therefore, a power communication maintenance knowledge graph is constructed through the power communication maintenance order image, and finally the power communication maintenance conflict is identified based on the power communication maintenance knowledge graph without manual participation, that is, there is no need to manually check the maintenance order item by item, thereby improving the recognition automation and efficiency of power communication maintenance conflicts. At the same time, based on the knowledge graph, complex entity relationships can be accurately identified, such as the dependency relationship between equipment, the overlapping relationship of maintenance time, etc., while manual identification of these complex relationships may have recognition errors and negligence problems, so the present invention can improve the recognition accuracy of power communication maintenance conflicts.

[0065] Furthermore, in order to better illustrate the above process of identifying power communication maintenance conflicts, as a refinement and extension of the above embodiment, the embodiment of the present invention provides another power communication maintenance conflict identification method based on knowledge graph, such as Figure 3 As shown, the method includes:

[0066] 201. Obtain the text information of the power communication maintenance order.

[0067] Specifically, maintenance-related documents in various forms such as PDF, Word text, Excel spreadsheets, etc., such as maintenance tickets, maintenance plans, and three measures and one plan, are structured and extracted to obtain the text information of power communication maintenance orders.

[0068] 202. Extract entities from the text information of electric power communication maintenance and mine the entity relationships between entities. Based on the entities and entity relationships, construct a knowledge graph of electric power communication maintenance, wherein each entity serves as a node in the knowledge graph of electric power communication maintenance, and the entity relationship serves as an edge between each node in the knowledge graph of electric power communication maintenance.

[0069] Specifically, after structuring the power communication maintenance related data into power communication maintenance text information, each entity is extracted from the power communication maintenance text information, and the entity relationship between each entity is mined based on the contextual relationship of each entity in the power communication maintenance text information. Then, each entity is used as each node in the power communication maintenance knowledge graph to be constructed, and the entity relationship between each entity is used as the edge between the corresponding nodes. Finally, the power communication maintenance knowledge graph is constructed based on each node and the edges between each node.

[0070] 203. Based on the power communication maintenance knowledge graph, power communication maintenance conflicts are identified.

[0071] For the embodiment of the present invention, after constructing the power communication maintenance knowledge graph, it is necessary to identify the power communication maintenance conflicts based on the power communication maintenance knowledge graph. Based on this, step 203 specifically includes: based on the power communication maintenance knowledge graph, performing at least one of time conflict identification, resource conflict identification, and logical conflict identification on the power communication maintenance.

[0072] Specifically, in one embodiment of the present invention, a power communication maintenance knowledge graph can be input into a preset conflict identification model for conflict identification, thereby obtaining power maintenance conflicts. The preset conflict identification model is constructed by: constructing an initial conflict identification model and obtaining a sample dataset, wherein the sample dataset includes a sample power communication maintenance knowledge graph with conflict annotation information; dividing the sample dataset into training data and test data, and using the training data to train the initial conflict identification model; testing the trained initial conflict identification model using the test data; and ultimately determining the initial conflict identification model that meets the test conditions as the preset conflict identification model. In another embodiment of the present invention, conflict identification rules can be formulated based on professional knowledge and experience in the field of power communication maintenance; and using the entities and relationships in the power maintenance knowledge graph to match the rules to identify potential conflicts. For example, if the graph contains multiple conflicting fault handling measures, it can be considered a conflict. This embodiment of the present invention uses knowledge graph technology to model and analyze text data, thereby accurately identifying and warning of potential maintenance conflicts in advance.

[0073] 204. Optimization of conflict identification efficiency in power communication maintenance based on attention mechanism.

[0074] For the embodiment of the present invention, in the process of maintenance conflict identification, in order to improve the efficiency and accuracy of maintenance conflict identification, it is necessary to introduce an attention mechanism for optimization. Based on this, step 204 specifically includes: an input layer, which converts data into vector form input; an attention layer, which uses the attention mechanism to calculate the weights between different data vectors to highlight key information; a conflict identification layer, which judges conflicts based on specific algorithms or rules. In the conflict detection stage, the focus is on related optical path services and conflict-prone areas based on business priorities and historical data; in the conflict location stage, the attention mechanism is used to quickly lock conflict nodes and edges in the knowledge graph, and the equipment location and connection relationship are clarified in combination with position coding; in the conflict resolution stage, weights are assigned to different resolution strategies according to the real-time nature, importance and maintenance resource status of the business, and a comprehensive evaluation is conducted.

[0075] 205. Issue early warning for identified power communication maintenance conflicts.

[0076] Specifically, when a conflict in power communication maintenance is detected, an early warning of the power communication maintenance conflict is performed based on information such as the conflict type.

[0077] According to another method for identifying power communication maintenance conflicts based on knowledge graph provided by the present invention, compared with the current method of determining maintenance conflicts by manually reviewing maintenance tickets, the present invention obtains power communication maintenance text information, extracts entities in the power communication maintenance text information and mines the entity relationships between the entities, and constructs a power communication maintenance knowledge graph based on the entities and the entity relationships, wherein the entities serve as nodes in the power communication maintenance knowledge graph, and the entity relationships serve as edges between the nodes in the power communication maintenance knowledge graph; finally, based on the power communication maintenance knowledge graph, power communication maintenance conflicts are identified. In this way, a power communication maintenance knowledge graph is constructed, and finally, power communication maintenance conflicts are identified based on the power communication maintenance knowledge graph without manual participation, that is, there is no need to manually check the maintenance order item by item, thereby improving the automation and efficiency of power communication maintenance conflict identification. At the same time, based on the knowledge graph, complex entity relationships can be accurately identified, such as carrying relationships, time associations, conflict relationships, etc., while manual identification of these complex relationships may result in recognition errors and negligence, so the present invention can improve the recognition accuracy of power communication maintenance conflicts.

[0078] Further, as Figure 1 The specific implementation of the present invention provides a power communication maintenance conflict identification device based on knowledge graph, such as Figure 4 As shown, the device includes: an acquisition unit 31, a construction unit 32, and an identification unit 33.

[0079] The acquisition unit 31 can be used to acquire power communication maintenance text information.

[0080] The construction unit 32 can be used to extract entities and mine the entity relationships between the entities in the power communication maintenance text information, and construct a power communication maintenance knowledge graph based on the entities and entity relationships, wherein each entity serves as a node in the power communication maintenance knowledge graph, and the entity relationships serve as edges between the nodes in the power communication maintenance knowledge graph.

[0081] In a specific application scenario, in order to extract entities from power communication maintenance text information, such as Figure 5 As shown, the construction unit 32 includes an acquisition module 321 and an entity extraction module 322.

[0082] The acquisition module 321 can be used to obtain a preset entity extraction model, wherein the preset entity extraction model includes a text encoder for extracting feature vectors, a deep self-attention network for learning context-related vector representations, and a conditional random field for label annotation; the preset entity extraction model is pre-trained based on a sample text dataset.

[0083] The entity extraction module 322 can be used to input the power communication maintenance text information into the preset entity extraction model, extract the feature vector of the power communication maintenance text information through the text encoder to obtain a feature vector, extract text context features of the feature vector through the deep self-attention network to obtain context features, and perform label recognition on the context features through the conditional random field to obtain a sequence annotation label, wherein the sequence annotation label contains each entity and its corresponding entity category.

[0084] In a specific application scenario, in order to mine entity relationships, the construction unit 32 further includes a relationship mining module 323 .

[0085] The acquisition module 321 can also be used to obtain a preset relationship recognition model, wherein the preset relationship recognition model includes an input representation layer for information embedding, a low-level feature extraction layer for low-level feature extraction, a dynamic routing capsule network feature clustering layer of a word attention mechanism for global relationship feature extraction, and a relationship prediction layer for entity relationship prediction; the preset relationship recognition model is pre-trained based on an entity text dataset.

[0086] The relationship mining module 323 can be used to input the power communication maintenance text information and each of the entities into the preset relationship recognition model, perform word embedding, character embedding, part-of-speech embedding, and word position embedding on the power communication maintenance text information through the input representation layer, and fuse the word embedding, character embedding, part-of-speech embedding, and word position embedding to obtain a fused embedding, perform low-level feature extraction on the fused embedding through the low-level feature extraction layer to obtain low-level features, perform feature extraction on the low-level features through the dynamic routing capsule network feature clustering layer of the word attention mechanism to obtain global relationship features, perform relationship prediction on the global relationship features through the relationship prediction layer to obtain entity relationships between the entities.

[0087] In a specific application scenario, the low-level feature extraction layer includes a bidirectional long short-term memory neural network, and the bidirectional long short-term memory neural network includes a forward network layer and a backward network layer; in order to perform low-level feature extraction, the relationship mining module 323 can be specifically used to input the fusion embedding into the bidirectional long short-term memory neural network for global sequence feature extraction, obtain the forward global sequence features corresponding to the forward network layer and the backward global sequence features corresponding to the backward network layer, and fuse the forward global sequence features and the backward global sequence features to obtain the low-level features.

[0088] In a specific application scenario, the dynamic routing capsule network feature clustering layer of the word attention mechanism includes: a low-level capsule network, a word attention mechanism, and a high-level capsule network; in order to extract global relationship features, the relationship mining module 323 can be specifically used to input the low-level features into the low-level capsule network for feature extraction to obtain local features, input the local features into the word attention mechanism for weight distribution to obtain weighted local features, and input the weighted local features into the high-level capsule network for feature extraction to obtain global relationship features.

[0089] In a specific application scenario, in order to perform conflict identification, the identification unit 33 can be specifically used to perform at least one of time conflict identification, resource conflict identification, and logical conflict identification on power communication maintenance based on the power communication maintenance knowledge graph.

[0090] In a specific application scenario, in order to provide an early warning for power communication maintenance conflicts, the device further includes an early warning unit 34 .

[0091] The early warning unit 34 can be used to issue an early warning for the identified power communication maintenance conflict.

[0092] In a specific application scenario, in order to optimize the power communication maintenance knowledge graph, the device also includes an optimization unit 35.

[0093] The nodes and edges in the power communication maintenance knowledge graph are converted into vector representations, and the multi-head attention mechanism is used to calculate the vector representation of the power communication maintenance knowledge graph. In the multi-head attention mechanism, multiple attention heads are set, and each head pays attention to and calculates the vectors of the power communication maintenance knowledge graph from different perspectives. The output results of multiple attention heads are spliced and linearly transformed to obtain the power communication maintenance knowledge graph with attention weights.

[0094] It should be noted that for other corresponding descriptions of the functional modules involved in the power communication maintenance conflict identification device based on knowledge graph provided by the embodiment of the present invention, please refer to Figure 1 The corresponding description of the method shown will not be repeated here.

[0095] Based on the above Figure 1 The method shown, accordingly, an embodiment of the present invention also provides a computer-readable storage medium, on which a computer program is stored, which implements the following steps when executed by a processor: obtaining power communication maintenance text information; extracting entities and mining the relationships between the entities in the power communication maintenance text information, and constructing a power communication maintenance knowledge graph based on the entities and entity relationships, wherein the entities serve as nodes in the power communication maintenance knowledge graph, and the entity relationships serve as edges between the nodes in the power communication maintenance knowledge graph; based on the power communication maintenance knowledge graph, identifying power communication maintenance conflicts.

[0096] Based on the above Figure 1 The method shown and Figure 4 The embodiment of the device shown in the figure, the embodiment of the present invention also provides a physical structure diagram of a computer device, such as Figure 6 As shown, the computer device includes: a processor 41, a memory 42, and a computer program stored in the memory 42 and executable on the processor, wherein the memory 42 and the processor 41 are both arranged on a bus 43; when the processor 41 executes the program, the following steps are implemented: obtaining power communication maintenance text information; extracting entities and mining the relationships between the entities in the power communication maintenance text information, and constructing a power communication maintenance knowledge graph based on the entities and entity relationships, wherein the entities serve as nodes in the power communication maintenance knowledge graph, and the entity relationships serve as edges between the nodes in the power communication maintenance knowledge graph; and identifying power communication maintenance conflicts based on the power communication maintenance knowledge graph.

[0097] The technical solution of the present invention obtains power communication maintenance text information; then extracts entities from the power communication maintenance text information and mines the entity relationships between the entities; based on the entities and the entity relationships, constructs a power communication maintenance knowledge graph, wherein each entity serves as a node in the power communication maintenance knowledge graph, and the entity relationships serve as edges between the nodes in the power communication maintenance knowledge graph; and finally, based on the power communication maintenance knowledge graph, identifies power communication maintenance conflicts. Thus, a power communication maintenance knowledge graph is constructed using the power communication maintenance order image, and finally, based on the power communication maintenance knowledge graph, power communication maintenance conflicts are identified without manual intervention, that is, without manually checking the maintenance orders item by item, thereby improving the automation and efficiency of power communication maintenance conflict identification. At the same time, the knowledge graph can accurately identify complex entity relationships, such as dependencies between devices and overlapping maintenance time relationships, while manual identification of these complex relationships may result in recognition errors and oversights, thereby improving the accuracy of power communication maintenance conflict identification.

[0098] Obviously, those skilled in the art will appreciate that the various modules or steps of the present invention described above can be implemented using a general-purpose computing device, centralized on a single computing device, or distributed across a network of multiple computing devices. Alternatively, they can be implemented using program code executable by a computing device, which can then be stored in a storage device and executed by the computing device. In some cases, the steps shown or described can be performed in a different order than that shown, or can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. Thus, the present invention is not limited to any particular combination of hardware and software.

[0099] The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Those skilled in the art will readily appreciate that various modifications and variations of the present invention are possible. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention shall be included within the scope of protection of the present invention.

Claims

1. A method for identifying conflicts in power communication maintenance based on knowledge graph, characterized in that: include: Obtain power communication maintenance text information; Extracting entities from the power communication maintenance text information and mining the relationships between the entities, and constructing a power communication maintenance knowledge graph based on the entities and entity relationships, wherein the entities serve as nodes in the power communication maintenance knowledge graph, and the entity relationships serve as edges between the nodes in the power communication maintenance knowledge graph; Based on the power communication maintenance knowledge graph, power communication maintenance conflicts are identified.

2. The method according to claim 1, characterized in that Extract entities from power communication maintenance text information, including: Obtaining a preset entity extraction model, wherein the preset entity extraction model includes a text encoder for extracting feature vectors, a deep self-attention network for learning context-dependent vector representations, and a conditional random field for labeling; the preset entity extraction model is pre-trained based on a sample text dataset; The electric power communication maintenance text information is input into the preset entity extraction model, the feature vector of the electric power communication maintenance text information is extracted by the text encoder to obtain a feature vector, the text context feature is extracted from the feature vector by the deep self-attention network to obtain a context feature, and the context feature is labeled by the conditional random field to obtain a sequence annotation label, wherein the sequence annotation label includes each entity and its corresponding entity category.

3. The method according to claim 1, characterized in that Mining entity relationships between the entities, including: Obtaining a preset relationship recognition model, wherein the preset relationship recognition model includes an input representation layer for information embedding, a low-level feature extraction layer for low-level feature extraction, a dynamic routing capsule network feature clustering layer with a word attention mechanism for global relationship feature extraction, and a relationship prediction layer for entity relationship prediction; the preset relationship recognition model is pre-trained based on an entity text dataset; The electric power communication maintenance text information and each of the entities are input into the preset relationship recognition model, and the electric power communication maintenance text information is subjected to word embedding, character embedding, part-of-speech embedding, and word position embedding through the input representation layer, and the word embedding, character embedding, part-of-speech embedding, and word position embedding are fused to obtain a fused embedding, and low-level feature extraction is performed on the fused embedding through the low-level feature extraction layer to obtain low-level features, and feature extraction is performed on the low-level features through the dynamic routing capsule network feature clustering layer of the word attention mechanism to obtain global relationship features, and relationship prediction is performed on the global relationship features through the relationship prediction layer to obtain entity relationships between the entities.

4. The method according to claim 3, characterized in that The low-level feature extraction layer includes a bidirectional long short-term memory neural network, and the bidirectional long short-term memory neural network includes a forward network layer and a backward network layer; The step of extracting low-level features from the fused embedding through the low-level feature extraction layer to obtain low-level features includes: The fusion is embedded and input into the bidirectional long short-term memory neural network to perform global sequence feature extraction to obtain the forward global sequence features corresponding to the forward network layer and the backward global sequence features corresponding to the backward network layer, and the forward global sequence features and the backward global sequence features are fused to obtain the low-level features.

5. The method according to claim 4, characterized in that The dynamic routing capsule network feature clustering layer of the word attention mechanism includes: a low-level capsule network, a word attention mechanism, and a high-level capsule network; The dynamic routing capsule network feature clustering layer of the word attention mechanism extracts features from the low-level features to obtain global relationship features, including: The low-level features are input into the low-level capsule network for feature extraction to obtain local features, the local features are input into the word attention mechanism for weight distribution to obtain weighted local features, and the weighted local features are input into the high-level capsule network for feature extraction to obtain global relationship features.

6. The method according to claim 1, characterized in that The identifying of power communication maintenance conflicts based on the power communication maintenance knowledge graph includes: Based on the power communication maintenance knowledge graph, performing at least one of time conflict identification, resource conflict identification, and logic conflict identification on the power communication maintenance; After identifying power communication maintenance conflicts based on the power communication maintenance knowledge graph, the method further includes: Provide early warning for identified power communication maintenance conflicts.

7. The method according to claim 1, characterized in that Before identifying power communication maintenance conflicts based on the power communication maintenance knowledge graph, the method further includes: The nodes and edges in the power communication maintenance knowledge graph are converted into vector representations, and the multi-head attention mechanism is used to calculate the vector representation of the power communication maintenance knowledge graph. In the multi-head attention mechanism, multiple attention heads are set, and each head pays attention to and calculates the vectors of the power communication maintenance knowledge graph from different perspectives. The output results of multiple attention heads are spliced and linearly transformed to obtain the power communication maintenance knowledge graph with attention weights.

8. A power communication maintenance conflict identification device based on knowledge graph, characterized in that: include: An acquisition unit, used for acquiring power communication maintenance text information; A construction unit is configured to extract entities from the power communication maintenance text information and mine entity relationships between the entities, and construct a power communication maintenance knowledge graph based on the entities and entity relationships, wherein the entities serve as nodes in the power communication maintenance knowledge graph, and the entity relationships serve as edges between the nodes in the power communication maintenance knowledge graph; An identification unit is used to identify power communication maintenance conflicts based on the power communication maintenance knowledge graph.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images