Power grid operation and maintenance deployment method and system based on knowledge graph

Abstract

The invention discloses a power grid operation and maintenance deployment method and system based on a knowledge graph, and relates to the technical field of power grid operation and maintenance deployment, and the method comprises the steps: collecting key data, and carrying out the preprocessing of the key data; extracting entities from the key data, identifying relationships between the entities, constructing a knowledge graph based on the relationships between the entities, and storing the knowledge graph in a graph database; dynamically identifying newly appearing entities based on the key data, synchronously adjusting the entity relationship, and updating the knowledge graph based on the new entities and the entity relationship; extracting training data from the knowledge graph, and constructing a graph neural network model based on the training data; and predicting potential faults in the power system based on the knowledge graph and the graph neural network model, and generating corresponding decision suggestions. According to the method, the entities are extracted from the key data and the relationship between the entities is identified, so that the structured expression of the complex relationship in the power system is realized, and the multi-source heterogeneous data is converted into operable knowledge graph nodes and edges.

Description

Technical Field

[0001] The present invention relates to the field of electric power grid operation and maintenance deployment technology, and in particular to an electric power grid operation and maintenance deployment method and system based on knowledge graph. Background Art

[0002] Traditional power grid operation and maintenance methods mostly rely on static monitoring data and preset fault diagnosis rules. Although this method can ensure the stable operation of the power grid to a certain extent, it seems powerless when dealing with increasingly complex and dynamically changing power systems. In recent years, with the rapid development of big data technology, artificial intelligence and the Internet of Things, the data collection and processing capabilities in power systems have been significantly improved, and data-driven methods have gradually become the mainstream technical direction of power operation and maintenance. Against this background, knowledge graphs, as a technology that can effectively represent complex relationships and multi-dimensional information, have gradually been introduced into power systems to enhance the intelligence and automation level of power operation and maintenance. Knowledge graphs can provide more intuitive and comprehensive power system fault prediction and decision support through dynamic modeling of equipment, nodes and their relationships in power systems.

[0003] Although the existing data-driven power operation and maintenance methods can use historical data for fault prediction to a certain extent, most of them lack deep modeling of complex relationships in the power system, especially it is difficult to dynamically adapt to the emergence of new equipment, new nodes and their relationship changes in the power system. This limitation leads to significant deficiencies in the accuracy and timeliness of fault prediction in existing technologies. Knowledge graph technology has good relationship modeling and dynamic expansion capabilities, and can effectively capture the complex relationships between different equipment and nodes in the power system, providing more comprehensive and dynamic support for fault prediction. At the same time, the graph neural network model based on knowledge graph can learn and predict potential faults in the power system based on large-scale data, significantly improving the accuracy of fault prediction and decision-making efficiency. Therefore, how to combine knowledge graph and graph neural network technology to build an efficient, accurate and dynamically adaptable operation and maintenance deployment method that can dynamically adapt to changes in the power system has become an urgent problem to be solved. Summary of the invention

[0004] In view of the above existing problems, the present invention is proposed.

[0005] Therefore, the present invention provides a knowledge graph-based power grid operation and maintenance deployment method and system to solve the problems of insufficient fault prediction accuracy and decision-making efficiency caused by data complexity and dynamics in the power system.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0007] In a first aspect, an embodiment of the present invention provides a method and system for power grid operation and maintenance deployment based on a knowledge graph, which includes collecting key data and preprocessing it;

[0008] Extract entities from key data, identify the relationships between entities, build a knowledge graph based on the relationships between entities, and store it in a graph database;

[0009] Dynamically identify newly emerged entities based on key data, adjust entity relationships synchronously, and update the knowledge graph based on new entities and entity relationships;

[0010] Extract training data from the knowledge graph and build a graph neural network model based on the training data;

[0011] Predict potential faults in the power system based on knowledge graph and graph neural network model, and generate corresponding decision recommendations.

[0012] As a preferred solution of the power grid operation and maintenance deployment method and system based on knowledge graph described in the present invention, wherein: the key data includes equipment operation data, environmental data, historical operation and maintenance data and text data, and appropriate sensors are selected according to the data type. The sensors are deployed around power equipment, substations and environmental monitoring points, and the key data collected by the sensors are transmitted to the central processing system through the network;

[0013] According to different data types, the sampling frequency is set, the data format is unified, and a timestamp and device identifier are added to each data.

[0014] As a preferred solution of the power grid operation and maintenance deployment method and system based on knowledge graph described in the present invention, wherein: the extraction of entities from key data and identification of the relationship between entities include:

[0015] Define entity categories in the knowledge graph, where the entity categories include equipment entities, fault entities, environment entities, and operation and maintenance entities;

[0016] Use NER technology to extract entities from preprocessed text data, use BERT model for entity recognition, input preprocessed text data into BERT model, generate word vectors for each word, and annotate entity categories in text through the output layer of BERT model;

[0017] Standardize the identified entities to eliminate different expressions of the same entity;

[0018] Based on the needs of power grid operation and maintenance, define the relationship types between entities, which include causal relationships, spatial relationships, and association relationships;

[0019] The dependency syntactic analysis plus rule matching method is used to extract the relationship between entities from text data.

[0020] As a preferred solution of the power grid operation and maintenance deployment method and system based on knowledge graph of the present invention, wherein: the knowledge graph is constructed based on entity and entity relationship and stored in the graph database, including:

[0021] Entities are regarded as nodes in the graph, and the relationships between entities are regarded as edges in the graph. The graph structure of the knowledge graph is constructed, and attributes are set for each entity node and each edge.

[0022] Define a graph G = (V, E), where V is the set of entities and E is the set of relationships. Merge homogeneous nodes, merge nodes with the same attributes into one node, and set weights for each edge based on the importance and frequency of the relationship.

[0023] The constructed knowledge graph is stored in a graph database. Each entity node is stored as a node in the graph database, and the attributes of the entity are stored as the attributes of the node. Each relationship is stored as an edge in the graph database, and the attributes of the relationship are stored as the attributes of the edge.

[0024] As a preferred solution of the power grid operation and maintenance deployment method and system based on knowledge graph described in the present invention, wherein: the method of dynamically identifying newly emerged entities based on key data, synchronously adjusting entity relationships, and updating the knowledge graph based on new entities and entity relationships includes:

[0025] Collect key data in real time and automatically identify new entities that have not appeared in the knowledge graph from new data based on NER technology;

[0026] Standardize newly identified entities;

[0027] Add new entities and set their properties from historical data;

[0028] Create a new node instance in the graph database, write the standardized entity information into the attribute field of the node, and then store all the attributes of the entity according to the structure of the graph database.

[0029] Based on the analysis of real-time data, detect the relationship between new entities and existing entities. When a new relationship is detected between new entities and existing entities, add the relationship to the knowledge graph;

[0030] For existing relationships, dynamically adjust the weight of the relationship based on the latest data, regularly check the relationships in the knowledge graph, and delete relationships that have not been triggered for a long time or have been proven to be invalid;

[0031] After updating the entities and relationships, check whether there are isolated nodes in the knowledge graph, whether there are repeated or redundant relationships, detect and process the circular relationships in the knowledge graph, and after completing the structural adjustment, verify the consistency of the knowledge graph.

[0032] As a preferred solution of the power grid operation and maintenance deployment method and system based on knowledge graph described in the present invention, wherein: the training data is extracted from the knowledge graph, and the graph neural network model is constructed based on the training data, including:

[0033] Extract features from nodes, edges, and adjacency matrices in the knowledge graph, define the node feature matrix as X, the edge feature matrix as e, and the adjacency matrix as A, and divide the data set into training set, validation set, and test set;

[0034] Initial weight matrix W;

[0035] In the forward propagation process, a multi-head attention mechanism is used to update the node features, and a node-edge interaction function fi nteract is introduced to combine the node and edge features for feature update. i Updates;

[0036] Define a multi-task loss function that combines node classification and edge regression tasks;

[0037] In the back propagation process, based on the loss function Calculate the gradient and use the Adam optimizer to update the model parameters. To prevent overfitting, introduce the Dropout mechanism in each layer and use L2 regularization to constrain the weight matrix. size;

[0038] After training is completed, the validation set is used to evaluate the graph neural network model, calculate the accuracy, F1 score and mean square error indicators, use learning rate decay and early stopping techniques, dynamically adjust the hyperparameters λ and ζ, and select the optimal model based on the validation results.

[0039] As a preferred solution of the power grid operation and maintenance deployment method and system based on knowledge graph described in the present invention, wherein: the prediction of potential faults in the power system based on the knowledge graph and graph neural network model and the generation of corresponding decision suggestions include:

[0040] Input real-time key data and node and edge features in the knowledge graph into the trained graph neural network model;

[0041] The graph neural network model calculates the failure probability of each device and node through the forward propagation process. The graph neural network model outputs the failure prediction result based on the characteristics of the node itself and the information of neighboring nodes.

[0042] Based on the fault prediction results, a maintenance plan is selected and emergency plan suggestions are generated;

[0043] Apply the decision-making plan to the operation and maintenance of the power system. During the implementation of the plan, monitor the system's operating status and fault conditions in real time, and dynamically adjust the decision-making plan based on actual conditions.

[0044] In a second aspect, the present invention provides a knowledge graph-based power grid operation and maintenance deployment system, comprising:

[0045] Data acquisition module: responsible for collecting key data in the power system; pre-processing module: performs pre-processing operations such as cleaning and standardization on the data;

[0046] Entity extraction module: extract key entities from preprocessed data; relationship identification module: and identify the relationship between entities; knowledge graph construction module: build a knowledge graph based on the extracted entities and relationships; storage module: store the knowledge graph in the graph database for subsequent query and analysis; dynamic entity recognition module: dynamically identify newly emerging entities based on new data and adjust entity relationships; graph update module: update the knowledge graph to maintain its timeliness and accuracy; training data extraction module: extract data sets for training graph neural network models from the knowledge graph; graph neural network model construction module: use the training data extracted from the knowledge graph to build and train the graph neural network model to capture complex graph structure features; fault prediction module: predict potential faults in the power system based on the updated knowledge graph and the trained graph neural network model.

[0047] In a third aspect, an embodiment of the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program is executed by the processor, it implements any step of the power grid operation and maintenance deployment method and system based on the knowledge graph as described in the first aspect of the present invention.

[0048] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the knowledge graph-based power grid operation and maintenance deployment method and system as described in the first aspect of the present invention.

[0049] The beneficial effects of the present invention are: 1. By extracting entities from key data and identifying the relationships between entities, the structured expression of complex relationships in the power system is realized, and multi-source heterogeneous data is converted into operational knowledge graph nodes and edges. These relationships are then stored in a graph database, and various types of equipment, faults, environment and other entities in the power system and their relationships are mapped into a knowledge graph, achieving the beneficial effects of improving the power operation and maintenance system's ability to understand complex information and enhancing information retrieval and reasoning efficiency.

[0050] 2. By dynamically identifying newly emerging entities and adjusting entity relationships based on key data, the dynamic update of the knowledge graph is realized, so that the system can adapt to new equipment or new failure modes appearing in the power system in real time, maintain the timeliness and accuracy of the knowledge graph, and be able to adjust the expression and processing of new information in a timely manner, thereby enhancing the adaptability of the power operation and maintenance system and its ability to respond quickly to new problems.

[0051] 3. By extracting training data from the knowledge graph and building a graph neural network model, complex feature learning based on graph structured data is realized, which can accurately capture the potential failure modes in the power system and generate accurate failure prediction results and decision-making recommendations through the prediction ability of the model. It can not only warn of possible failures in advance, but also optimize maintenance plans, thereby improving the accuracy of power system fault prediction, reducing power outage risks and operation and maintenance costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0053] Figure 1 This is a flow chart of the knowledge graph-based power grid operation and maintenance deployment method and system in Example 1. DETAILED DESCRIPTION

[0055] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.

[0056] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0057] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.

[0058] Example 1

[0059] Reference Figure 1, which is the first embodiment of the present invention, and which provides a method and system for power grid operation and maintenance deployment based on knowledge graph, comprising the following steps: collecting key data and preprocessing the data;

[0060] The key data includes equipment operation data, environmental data, historical operation and maintenance data, and text data. According to the data type, appropriate sensors are selected. Sensors are deployed around power equipment, substations, and environmental monitoring points. The key data collected by the sensors are transmitted to the central processing system through the network;

[0061] According to different data types, the sampling frequency is set, the data format is unified, and a timestamp and device identifier are added to each data.

[0062] Extract entities from key data, identify the relationships between entities, build a knowledge graph based on the relationships between entities, and store it in a graph database;

[0063] The extracting entities from the key data and identifying the relationships between entities include:

[0064] Define entity categories in the knowledge graph, where the entity categories include equipment entities, fault entities, environment entities, and operation and maintenance entities;

[0065] Use NER technology to extract entities from preprocessed text data, use BERT model for entity recognition, input preprocessed text data into BERT model, generate word vectors for each word, and annotate entity categories in text through the output layer of BERT model;

[0066] Standardize the identified entities to eliminate different expressions of the same entity;

[0067] Based on the needs of power grid operation and maintenance, define the relationship types between entities, which include causal relationships, spatial relationships, and association relationships;

[0068] The dependency syntactic analysis plus rule matching method is used to extract the relationship between entities from text data.

[0069] The knowledge graph is constructed based on entities and entity relationships and stored in a graph database, including:

[0070] Entities are regarded as nodes in the graph, and the relationships between entities are regarded as edges in the graph. The graph structure of the knowledge graph is constructed, and attributes are set for each entity node and each edge.

[0071] Define a graph G = (V, E), where V is the set of entities and E is the set of relationships. Merge homogeneous nodes, merge nodes with the same attributes into one node, and set weights for each edge based on the importance and frequency of the relationship.

[0072] The constructed knowledge graph is stored in a graph database. Each entity node is stored as a node in the graph database, and the attributes of the entity are stored as the attributes of the node. Each relationship is stored as an edge in the graph database, and the attributes of the relationship are stored as the attributes of the edge.

[0073] Dynamically identify newly emerged entities based on key data, adjust entity relationships synchronously, and update the knowledge graph based on new entities and entity relationships;

[0074] Collect key data in real time and automatically identify new entities that have not appeared in the knowledge graph from new data based on NER technology;

[0075] Standardize newly identified entities;

[0076] Add new entities and set their properties from historical data;

[0077] Create a new node instance in the graph database, write the standardized entity information into the attribute field of the node, and then store all the attributes of the entity according to the structure of the graph database.

[0078] Based on the analysis of real-time data, detect the relationship between new entities and existing entities. When a new relationship is detected between new entities and existing entities, add the relationship to the knowledge graph;

[0079] For existing relationships, dynamically adjust the weight of the relationship based on the latest data, regularly check the relationships in the knowledge graph, and delete relationships that have not been triggered for a long time or have been proven to be invalid;

[0080] After updating the entities and relationships, check whether there are isolated nodes in the knowledge graph, whether there are repeated or redundant relationships, detect and process the circular relationships in the knowledge graph, and after completing the structural adjustment, verify the consistency of the knowledge graph.

[0081] Extract training data from the knowledge graph and build a graph neural network model based on the training data;

[0082] Extract features from nodes, edges, and adjacency matrices in the knowledge graph, define the node feature matrix as X, the edge feature matrix as e, and the adjacency matrix as A, and divide the data set into training set, validation set, and test set;

[0083] The initial weight matrix W has the following initialization formula:

[0084]

[0085] in: is the initial weight matrix; is a random matrix sampled from a standard normal distribution, N is a node in the knowledge graph, deg(v i ) is the node v i The degree.

[0086] In the forward propagation process, a multi-head attention mechanism is used to update the node features, and a node-edge interaction function fi nteract is introduced to combine the node and edge features for feature update. i The update formula is:

[0087]

[0088] in: For node v i In the features of the l+1th layer, σ is the activation function, is the attention weight, f interact is the node-edge interaction function, H is the number of attention heads in the multi-head attention mechanism, For node v i The set of neighbor nodes, e ij is the edge feature between node i and node j.

[0089] Define the multi-task loss function combining node classification and edge regression tasks, expressed as:

[0090]

[0091] in: is the total loss function, N is the total number of nodes, M is the total number of edges, C is the total number of fault types, y ic For node v i The true label on category c, For node v i The predicted probability of class c, w ij For edge e ij The weight, r ij For edge e ij The true value of For edge e ij The predicted value of , λ is the weight balance coefficient between node classification and edge regression loss.

[0092] In the back propagation process, based on the loss function Calculate the gradient and use the Adam optimizer to update the model parameters. To prevent overfitting, introduce the Dropout mechanism in each layer and use L2 regularization to constrain the weight matrix. The size of is expressed as:

[0093]

[0094] in: is the weight matrix of the hth head in the lth layer, η is the learning rate, is the loss function The weight matrix The gradient of ζ is the L2 regularization coefficient, which is used to prevent overfitting.

[0095] After training is completed, the validation set is used to evaluate the graph neural network model, calculate the accuracy, F1 score and mean square error indicators, use learning rate decay and early stopping techniques, and dynamically adjust the hyperparameters λ and ζ.

[0096] And select the optimal model based on the verification results.

[0097] Predict potential faults in the power system based on knowledge graph and graph neural network model, and generate corresponding decision recommendations.

[0098] Input real-time key data and node and edge features in the knowledge graph into the trained graph neural network model;

[0099] The graph neural network model calculates the failure probability of each device and node through the forward propagation process. The graph neural network model outputs the failure prediction result based on the characteristics of the node itself and the information of neighboring nodes.

[0100] In each layer of the graph neural network, the node v i The features of will interact with the features of its neighboring nodes to generate new nodes. This process is called message passing, and its formula is:

[0101]

[0102] in: For node v i Features at level l; For node v i Updated features at layer l+1; For node v i The set of neighbor nodes of For slave node v j To node v i The attention weight indicates the importance of message transmission; W (l) is the weight matrix of the lth layer; b (l) is the bias vector of the lth layer; σ is the activation function.

[0103] After multiple layers of message passing, the feature representation of the node in the last layer is obtained. The failure probability of each node is predicted based on this feature, expressed as:

[0104]

[0105] in: For node v i The predicted probability distribution corresponding to T types of faults, where T is the number of fault types; W fault is the weight matrix of the fault classifier; b fault is the bias vector of the fault classifier.

[0106] For each node v i The fault type with the highest probability is selected as the predicted fault type of the node, expressed as:

[0107]

[0108] in: For node v i The most likely fault type.

[0109] Based on the fault prediction results, a maintenance plan is selected and emergency plan suggestions are generated;

[0110] Apply the decision-making plan to the operation and maintenance of the power system. During the implementation of the plan, monitor the system's operating status and fault conditions in real time, and dynamically adjust the decision-making plan according to actual conditions to ensure its effectiveness.

[0111] This embodiment also provides a system for power grid operation and maintenance deployment based on knowledge graph, including:

[0112] Data acquisition module: responsible for collecting key data in the power system; pre-processing module: performs pre-processing operations such as cleaning and standardization on the data;

[0113] Entity extraction module: extract key entities from preprocessed data; relationship identification module: and identify the relationship between entities; knowledge graph construction module: build a knowledge graph based on the extracted entities and relationships; storage module: store the knowledge graph in the graph database for subsequent query and analysis; dynamic entity recognition module: dynamically identify newly emerging entities based on new data and adjust entity relationships; graph update module: update the knowledge graph to maintain its timeliness and accuracy; training data extraction module: extract data sets for training graph neural network models from the knowledge graph; graph neural network model construction module: use the training data extracted from the knowledge graph to build and train the graph neural network model to capture complex graph structure features; fault prediction module: predict potential faults in the power system based on the updated knowledge graph and the trained graph neural network model.

[0114] This embodiment also provides a computer device, which is suitable for the electric power grid operation and maintenance deployment method and system based on the knowledge graph, including: a memory and a processor; the memory is used to store computer executable instructions, and the processor is used to execute computer executable instructions to implement the electric power grid operation and maintenance deployment method and system based on the knowledge graph as proposed in the above embodiment.

[0115] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.

[0116] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, it implements the power grid operation and maintenance deployment method and system based on the knowledge graph as proposed in the above embodiment; the storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, referred to as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, referred to as EEPROM), erasable programmable read-only memory (Erasable Programmable Read On ly Memory, referred to as EPROM), programmable read-only memory (Programmable Red-Only Memory, referred to as PROM), read-only memory (Read-Only Memory, referred to as ROM), magnetic storage, flash memory, disk or optical disk.

[0117] In summary, the present invention realizes the monitoring and understanding of user behavior through the analysis of user file operation behavior, can accurately capture the user's operation habits and behavior patterns, provide basic data for the subsequent generation of encryption strategies, and also help the system to detect and prevent abnormal behaviors; predict the user's future operations according to the user's behavior habits, so as to formulate encryption strategies more intelligently; more accurately protect user privacy, reduce the risk of data leakage, and improve the data security and confidentiality of the system; the present invention better adapts to the user's operation habits and needs, improves the intelligence level of the encryption algorithm and user experience, and enhances the flexibility and adaptability of the system.

[0118] Example 2

[0119] Referring to Table 1, which is the second embodiment of the present invention, in order to further verify the advancement of the present invention, experimental simulation data of a power grid operation and maintenance deployment method and system based on a knowledge graph are provided.

[0120] In order to verify the effectiveness of the power grid operation and maintenance deployment method based on knowledge graph, a simulated power system consisting of 10 substations was constructed. Each substation was deployed with a variety of sensors to collect equipment operation data, environmental data, historical operation and maintenance data, and text data. The data was transmitted to the central processing system through the network for unified preprocessing operations, including data cleaning, standardization, format unification, and the addition of timestamps and device identifiers.

[0121] After data preprocessing is completed, NER technology is used to extract entities from text data, and the BERT model is used for entity recognition. The extracted entities include equipment entities, fault entities, environmental entities, and operation and maintenance entities. The differences in different expressions are eliminated through standardization. Then, based on the power operation and maintenance needs, the causal relationship, spatial relationship, and association relationship between entities are defined, and the entity relationship is extracted from the data through dependency syntactic analysis combined with rule matching.

[0122] The graph structure of the knowledge graph is constructed by taking the extracted entities as nodes and the entity relationships as edges, and the corresponding attributes and weights are set for each entity node and edge. The constructed knowledge graph is stored in the graph database, which ensures structured storage of data while providing a basis for subsequent data analysis and model training.

[0123] Afterwards, based on the key data collected in real time, newly emerging entities are dynamically identified, and the dynamic update and timeliness of the knowledge graph are ensured by updating the node and edge structure of the knowledge graph. Finally, training data is extracted from the knowledge graph and a graph neural network model is constructed. The model uses a multi-head attention mechanism and a node-edge interaction function to update node features and is optimized through a multi-task loss function. After training, the validation set is used to evaluate the model, and the model is applied to fault prediction to generate the fault probability of the power system and corresponding decision recommendations.

[0124] Table 1: Comparison of the conventional method and the method of the present invention

[0125]

[0126]

[0127] By comparing the traditional method with the knowledge graph-based power grid operation and maintenance deployment method, it is found that the new method has significant advantages in multiple key indicators. First, in terms of fault detection rate, the knowledge graph-based method has significantly higher fault detection rates in each substation than the traditional method, with an average increase of about 20 percentage points. This shows that through the construction of knowledge graphs and the application of graph neural network models, the system's ability to identify potential faults is effectively enhanced, and more potential fault signals can be captured, avoiding serious consequences caused by missed detections.

[0128] Secondly, in terms of false alarm rate, the knowledge graph-based method has a lower false alarm rate, which is reduced by 7 percentage points on average. This shows that this method not only improves the detection rate in fault prediction, but also reduces the occurrence of false alarms, further improving the accuracy and reliability of operation and maintenance. Based on the accurate modeling and dynamic updating of the relationship between entities in the knowledge graph, the system can more effectively distinguish the difference between faults and normal operating states.

[0129] In addition, in terms of fault handling time, the knowledge graph-based method also shows significant advantages. The average fault handling time is reduced by about 25 minutes compared with the traditional method. This shows that the knowledge graph-based method can not only identify faults faster, but also provide more accurate decision support, helping operation and maintenance personnel to take effective measures in a short time and quickly restore the normal operation of the system.

[0130] Comprehensive analysis shows that the power grid operation and maintenance deployment method based on knowledge graph significantly improves the fault detection rate, reduces the false alarm rate, and shortens the fault handling time, reflecting its innovation and practicality in power system operation and maintenance. Compared with traditional methods, this method forms an efficient and intelligent operation and maintenance system through dynamic data collection and processing, construction and updating of knowledge graphs, and application of graph neural network models, which can better adapt to the complex and changeable power system environment.

[0131] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A method for power grid operation and maintenance deployment based on knowledge graph, characterized in that: include: Collect key data and pre-process it; Extract entities from key data, identify the relationships between entities, build a knowledge graph based on the relationships between entities, and store it in a graph database; Dynamically identify newly emerged entities based on key data, adjust entity relationships synchronously, and update the knowledge graph based on new entities and entity relationships; Extract training data from the knowledge graph and build a graph neural network model based on the training data; Predict potential faults in the power system based on knowledge graph and graph neural network model, and generate corresponding decision recommendations.

2. The power grid operation and maintenance deployment method based on knowledge graph according to claim 1, characterized in that: The key data includes equipment operation data, environmental data, historical operation and maintenance data, and text data. According to the data type, appropriate sensors are selected. Sensors are deployed around power equipment, substations, and environmental monitoring points. The key data collected by the sensors are transmitted to the central processing system through the network; According to different data types, the sampling frequency is set, the data format is unified, and a timestamp and device identifier are added to each data.

3. The power grid operation and maintenance deployment method based on knowledge graph according to claim 2, characterized in that: The extracting entities from the key data and identifying the relationships between entities include: Define entity categories in the knowledge graph, where the entity categories include equipment entities, fault entities, environment entities, and operation and maintenance entities; Use NER technology to extract entities from preprocessed text data, use BERT model for entity recognition, input preprocessed text data into BERT model, generate word vectors for each word, and annotate entity categories in text through the output layer of BERT model; Standardize the identified entities to eliminate different expressions of the same entity; Based on the needs of power grid operation and maintenance, define the relationship types between entities, which include causal relationships, spatial relationships, and association relationships; The dependency syntactic analysis plus rule matching method is used to extract the relationship between entities from text data.

4. The power grid operation and maintenance deployment method based on knowledge graph according to claim 3, characterized in that: The knowledge graph is constructed based on entities and entity relationships and stored in a graph database, including: Entities are regarded as nodes in the graph, and the relationships between entities are regarded as edges in the graph. The graph structure of the knowledge graph is constructed, and attributes are set for each entity node and each edge. Define a graph G = (V, E), where V is the set of entities and E is the set of relationships. Merge homogeneous nodes, merge nodes with the same attributes into one node, and set weights for each edge based on the importance and frequency of the relationship. The constructed knowledge graph is stored in a graph database. Each entity node is stored as a node in the graph database, and the attributes of the entity are stored as the attributes of the node. Each relationship is stored as an edge in the graph database, and the attributes of the relationship are stored as the attributes of the edge.

5. The power grid operation and maintenance deployment method based on knowledge graph according to claim 4, characterized in that: The method dynamically identifies newly emerged entities based on key data, synchronously adjusts entity relationships, and updates the knowledge graph based on new entities and entity relationships. include, Collect key data in real time and automatically identify new entities that have not appeared in the knowledge graph from new data based on NER technology; Standardize newly identified entities; Add new entities and set their properties from historical data; Create a new node instance in the graph database, write the standardized entity information into the attribute field of the node, and then store all the attributes of the entity according to the structure of the graph database; Based on the analysis of real-time data, detect the relationship between new entities and existing entities. When a new relationship is detected between new entities and existing entities, add the relationship to the knowledge graph; For existing relationships, dynamically adjust the weight of the relationship based on the latest data, regularly check the relationships in the knowledge graph, and delete relationships that have not been triggered for a long time or have been proven to be invalid; After updating the entities and relationships, check whether there are isolated nodes in the knowledge graph, whether there are repeated or redundant relationships, detect and process the circular relationships in the knowledge graph, and after completing the structural adjustment, verify the consistency of the knowledge graph.

6. The power grid operation and maintenance deployment method based on knowledge graph according to claim 5, characterized in that: The extracting of training data from the knowledge graph and constructing a graph neural network model based on the training data include: Extract features from nodes, edges, and adjacency matrices in the knowledge graph, define the node feature matrix as X, the edge feature matrix as e, and the adjacency matrix as A, and divide the data set into training set, validation set, and test set; Initial weight matrix W; In the forward propagation process, a multi-head attention mechanism is used to update the node features. The node-edge interaction function finteract is introduced to combine the node and edge features for feature update. i Updates; Define a multi-task loss function that combines node classification and edge regression tasks; In the back propagation process, based on the loss function Calculate the gradient and use the Adam optimizer to update the model parameters. To prevent overfitting, introduce the Dropout mechanism in each layer and use L2 regularization to constrain the weight matrix. size; After training is completed, the validation set is used to evaluate the graph neural network model, calculate the accuracy, F1 score and mean square error indicators, use learning rate decay and early stopping techniques, dynamically adjust the hyperparameters λ and ζ, and select the optimal model based on the validation results.

7. The power grid operation and maintenance deployment method based on knowledge graph according to claim 6, characterized in that: The knowledge graph and graph neural network model are used to predict potential faults in the power system and generate corresponding decision suggestions. include, Input real-time key data and node and edge features in the knowledge graph into the trained graph neural network model; The graph neural network model calculates the failure probability of each device and node through the forward propagation process. The graph neural network model outputs the failure prediction result based on the characteristics of the node itself and the information of neighboring nodes. Based on the fault prediction results, a maintenance plan is selected and emergency plan suggestions are generated; Apply the decision-making plan to the operation and maintenance of the power system. During the implementation of the plan, monitor the system's operating status and fault conditions in real time, and dynamically adjust the decision-making plan based on actual conditions.

8. A system for power grid operation and maintenance deployment based on knowledge graph, based on the power grid operation and maintenance deployment method based on knowledge graph according to any one of claims 1 to 7, characterized in that: include: Data acquisition module: responsible for collecting key data in the power system; Preprocessing module: performs preprocessing operations such as cleaning and standardization on data; Entity extraction module: extract key entities from preprocessed data; Relationship identification module: and identify the relationship between entities; Knowledge graph construction module: builds a knowledge graph based on the extracted entities and relationships; Storage module: stores the knowledge graph in the graph database for subsequent query and analysis; Dynamic entity recognition module: dynamically identifies newly emerged entities based on new data and adjusts entity relationships; Graph update module: updates the knowledge graph to maintain its timeliness and accuracy; Training data extraction module: extracts data sets from knowledge graphs for training graph neural network models; Graph neural network model building module: Use the training data extracted from the knowledge graph to build and train the graph neural network model to capture complex graph structure features; Fault prediction module: Predicts potential faults in the power system based on the updated knowledge graph and the trained graph neural network model.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the power grid operation and maintenance deployment method and system based on the knowledge graph are implemented as described in any one of claims 1 to 7.

10. 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 power grid operation and maintenance deployment method and system based on knowledge graph are implemented as described in any one of claims 1 to 7.

Images