Hydropower station abnormal event emergency disposal method and device based on knowledge graph

Abstract

The invention discloses a hydropower station abnormal event emergency disposal method and device based on a knowledge graph. The method comprises the following steps: constructing a hydropower station abnormal event knowledge graph mode layer; identifying entities through the text classification model, and classifying the entities; knowledge fusion is carried out through a cosine similarity-based method, and a text classification result is optimized; the graph database is a medium storage database and constructs an abnormal event emergency disposal knowledge graph; and carrying out similarity calculation to obtain the most similar abnormal event entity, so that a corresponding processing flow can be obtained in the abnormal event emergency processing knowledge graph. The method has important significance on daily safe operation of the hydropower station, rapid processing of abnormal events and the like, loss and influence caused by improper emergency response can be effectively reduced, and a powerful guarantee is provided for safe and stable operation of the hydropower station.

Description

Technical Field

[0001] The present invention relates to the application field of knowledge graph of hydropower centralized control center, and specifically to a method and device for emergency handling of abnormal events in hydropower stations based on knowledge graph. Background Art

[0002] As an important power supply and water conservancy facility, the safe operation of hydropower stations is of great significance to ensuring energy supply and water resources management. Due to the special operating environment and complex engineering structure of hydropower stations, as well as the numerous monitoring equipment and numerous monitoring parameters, hydropower stations will inevitably encounter various abnormal events during operation. These abnormal events may threaten the normal operation and safety of hydropower stations, and the diagnosis process of abnormal events is significantly complex and cumbersome, resulting in the operation of the centralized control center to a large extent relying on detailed abnormal handling procedures and the dispatcher's experience analysis. When an abnormal event occurs in the power grid, the traditional emergency response method that relies on manual review and analysis makes hydropower stations face certain challenges in the safety and timely dispatch of water resources and electricity.

[0003] At present, in terms of emergency response to abnormal events, computer monitoring systems only monitor and analyze operations, while decision-making and execution rely on the experience and adaptability of control personnel, which not only reduces execution efficiency, but also makes it difficult to share and inherit operating experience and capabilities, thus affecting the accuracy and standardization of fault handling. With the use of remote centralized control of equipment, the amount of operations performed by staff is also increasing, requiring staff to accurately handle various possible abnormal situations. The dispatching business in the power system contains a large amount of complex data and rules, and it is difficult for staff to quickly and accurately handle complex abnormal events. Therefore, how to quickly and accurately respond to abnormal events in hydropower stations and improve the efficiency of emergency response has become an important issue in the management and operation of hydropower stations.

[0004] A knowledge graph is a multi-sided relationship graph consisting of nodes (representing entities) and edges (representing relationships between entities). It essentially constitutes a semantic network that reveals the connections between different entities. In the application scenario of abnormal handling of hydropower stations, the knowledge graph constructs a multi-dimensional information framework to associate scattered data, such as operating procedures, daily maintenance, and abnormal cases, to form a structured knowledge network. This not only optimizes the access and retrieval process of information, but also provides a scientific basis for the rapid diagnosis and accurate handling of abnormal events. By constructing a knowledge graph model based on relevant data in the field of emergency handling of hydropower stations, the complex relationships between entities can be quickly and accurately analyzed, thereby improving the efficiency of the emergency response process for abnormal events. With the continuous operation of the hydropower station, the data in the knowledge graph continues to increase and improve, providing a continuously updated resource for knowledge management and decision support. This dynamic knowledge base enables the continuous optimization of operating strategies, thereby enhancing the safety and operating efficiency of the hydropower station. He Zengliang et al. disclosed in the Chinese invention patent CN117648446A a "method for identifying and diagnosing abnormal events in hydropower station monitoring based on knowledge graphs", which applied knowledge graph technology to the daily operation of hydropower stations, and achieved the identification and diagnosis of abnormal events by building a comprehensive event library and signal relationship knowledge graph. However, its main function is to monitor and diagnose the occurrence of abnormal events, and it has failed to further expand the scope of application and provide operational plan support for emergency response to abnormal events in hydropower stations. Summary of the invention

[0005] The purpose of the present invention is to overcome the shortcomings and deficiencies in the prior art, and to provide a structured method and device for emergency handling of abnormal events in hydropower stations, aiming to improve the control capability and safe operation level of hydropower stations in the face of abnormal events, and to have certain practical and promotional value for improving the dispatching and decision-making capabilities of hydropower stations. The present invention constructs a structured knowledge graph for emergency handling of abnormal events, so that the abnormal detection results can be organically combined with the specific processing flow, providing a comprehensive intelligent diagnosis and emergency handling solution for abnormal events in hydropower stations, and providing systematic and operational plan support for emergency handling of abnormal events in hydropower stations.

[0006] To achieve the purpose of the present invention, the present invention provides a method for emergency handling of abnormal events in a hydropower station based on a knowledge graph, comprising the following steps: Construct a knowledge graph model layer of abnormal events of hydropower stations in the study area, and formally describe the entities, their relationships, and the attributes of entities and relationships; Identify entities through text classification models and classify entities; The knowledge of the identified entities is integrated through the cosine similarity method to optimize the text classification results, and the text data is stored in the graph database as a medium to build a knowledge graph for emergency response to abnormal events. Calculate the similarity of the abnormal events involved in the alarm information when the abnormal event occurs, obtain the most similar abnormal event entity, and use the most similar abnormal event entity to obtain the corresponding abnormal phenomenon, cause and processing flow of the abnormal event in the abnormal event emergency disposal knowledge graph; The knowledge graph of emergency response to abnormal events is visualized through the back-end framework and the front-end framework.

[0007] As an optimal technical solution, when constructing the knowledge graph model layer of abnormal events in hydropower stations in the study area, the emergency response plan texts, regulations and dispatching procedures in the study area are collected, and a top-down approach is used to analyze the entity types, attribute types and entity relationships of concern to construct a knowledge graph model layer framework for emergency response to abnormal events in hydropower stations.

[0008] As a preferred technical solution, the text classification model is obtained in the following way: The emergency response plan texts, rules and regulations texts, and dispatching procedures texts of the study area were organized into a data set; Traverse each line in the data set, complete text preprocessing by combining regular expressions with word segmentation tools (such as Jieba word segmentation) to obtain the preprocessed data set; The data in the preprocessed data set is vectorized and converted into feature vectors, and then the word frequency-inverse document frequency method is used to extract the features of the field; The vectorized data is passed into the text classification model for model training to obtain a trained text classification model.

[0009] As a preferred technical solution, the text classification model can map each data point to a high-dimensional space. The hyperplane found in the high-dimensional space can maximize the interval between data points of different categories. The mathematical model of the text classification model is expressed as: ; For all training samples in the dataset ( , ),satisfy: ; The importance of a word in a data set to one of the data in a data set is evaluated by word frequency-inverse document frequency. The calculation formula of word frequency-inverse document frequency is: ; In the formula, represents the normal vector of the hyperplane; represents the bias term; is the decision function in the text classification model; Indicates feature vectors, It is with The category label corresponding to the feature vector; To indicate a word; Represents a specific piece of data; Indicates the total amount of data in the dataset; Representing words In the data In and The product of Importance in data; Representing words In the data The frequency in Representing words In the dataset The inverse document frequency in .

[0010] As a preferred technical solution, knowledge fusion is performed on the identified entities by a method based on cosine similarity. Knowledge fusion refers to the unified representation of multiple entities with the same meaning as a single entity. The mathematical model of knowledge fusion is expressed as follows: ; In the formula, and are two vectors, and is an element in a vector, indicating that the vector and No. The elements at the positions, is the total number of elements.

[0011] As a preferred technical solution, the construction steps of the abnormal incident emergency handling knowledge graph include: constructing corresponding nodes in the form of RDF triples for all entities after knowledge fusion, writing the constructed nodes and their relationships into the graph database through the query language, and forming a complete abnormal incident emergency handling knowledge graph. Combined with the retrieval function of the graph database, efficient query and reasoning of abnormal incident emergency handling knowledge can be further realized.

[0012] As a preferred technical solution, the Jaccard entity similarity algorithm is used to calculate the similarity. The mathematical model of the Jaccard entity similarity algorithm is expressed as: ; In the formula, Represents the similarity value, and Represents a collection of attributes of two different entities.

[0013] As a preferred technical solution, the backend framework adopts a Django-based backend framework, and the mathematical model of the Django framework is expressed as: ; In the formula, Indicates the result set that meets the query conditions after the user's query; Represents a query operation in the knowledge graph of emergency response to abnormal events Search and return all the files that meet the query conditions Nodes or relationships; Representing knowledge graph of emergency response to abnormal events A node or relationship in Representing knowledge graph of emergency response to abnormal events The global set of nodes and relationships; is a Boolean function representing a node or relationship Whether it meets the query conditions .

[0014] As a preferred technical solution, the front-end framework adopts a front-end framework based on Vue 3.0. The mathematical model of the front-end framework based on Vue 3.0 is expressed as: ; In the formula, Represents the rendered visual page view; For a given frontend state Interaction with users Function of Represents the front-end status, including the data entered by the user and the data returned by the back-end; Represents a set of user interaction behaviors; is a Boolean function representing a visualization element Whether it is necessary to determine the front-end status be displayed; Indicates that it contains the visual elements that should be displayed in the current state, that is, A collection of .

[0015] The knowledge graph-based emergency response device for abnormal events in a hydropower station provided by the present invention includes the following modules: The pattern layer construction module is used to construct the knowledge graph pattern layer of abnormal events of hydropower stations in the study area, and to formally describe the entities, their relationships, and the attributes of the entities and their relationships; The entity recognition and classification module is used to identify entities through text classification models and classify entities; The knowledge graph construction module is used to perform knowledge fusion on the identified entities based on the cosine similarity method to optimize the text classification results, store text data in the graph database as a medium, and construct a knowledge graph for emergency response to abnormal events; The abnormal event identification module is used to calculate the similarity of the abnormal events involved in the alarm information when the abnormal event occurs, obtain the most similar abnormal event entity, and use the most similar abnormal event entity to obtain the corresponding abnormal phenomenon, cause and processing flow of the abnormal event in the abnormal event emergency disposal knowledge graph; The visualization module is used to visualize the knowledge graph of emergency response to abnormal events through the back-end framework and the front-end framework.

[0016] The present invention has the following advantages and effects compared with the prior art: 1) So far, there are few studies on the application of knowledge graphs in the field of hydropower stations. This paper focuses on the application of knowledge graph technology in the field of abnormal event handling in hydropower stations. By constructing a knowledge graph in the field of emergency response in hydropower stations, it aims to improve the regulation and control capabilities and safe operation levels of hydropower stations when facing abnormal events, and has certain practical and promotional value for improving the dispatching decision-making capabilities of hydropower stations.

[0017] 2) The text composition structure of the emergency response plan of a hydropower station is complex and contains many professional terms, which makes the currently commonly used methods have little effect in the process of constructing the knowledge graph in the field of abnormal event handling in hydropower stations. The present invention constructs a knowledge graph by adopting a method based on a vector machine text classification model, which can quickly and accurately extract key entities, and can calculate the similarity of the alarm information through an entity similarity algorithm, so that the correct content can be returned even when the alarm information is inconsistent with the characters of the procedure text.

[0018] 3) When an abnormal alarm occurs during the daily operation of a hydropower station, the knowledge graph for emergency response to abnormal events executes the processing flow in sequence according to the entity relationships in the graph and the order of edge transmission. The dispatcher only needs to follow the given process in the graph step by step to complete the emergency response, thereby improving the dispatcher's efficiency in handling complex abnormal events and significantly reducing the emergency response time. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 A flow chart of a method and device for emergency handling of abnormal events in a hydropower station based on a knowledge graph provided in an embodiment of the present invention.

[0020] Figure 2 This is a framework diagram of the knowledge graph model layer of abnormal events in a hydropower station in an embodiment of the present invention.

[0021] Figure 3 The figure is a flow chart of a text classification model based on a support vector machine in an embodiment of the present invention.

[0022] Figures 4 (a) to 4 (d) are storage diagrams of the knowledge graphs for emergency response to some abnormal events at four hydropower stations in an embodiment of the present invention; Figures 4 (a), 4 (b), 4 (c), and 4 (d) are Shannipo Hydropower Station, Dongqing Hydropower Station, Mamaya Hydropower Station, and Guangzhao Hydropower Station, respectively.

[0023] Figure 5 This is a flow chart of emergency response to the abnormal event of "unit overspeed" according to an embodiment of the present invention.

[0024] Figure 6 This is a visualized flow chart of emergency handling of abnormal events according to an embodiment of the present invention. DETAILED DESCRIPTION

[0025] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not limited to the present invention.

[0026] like Figure 1 As shown, the knowledge graph-based emergency handling method for abnormal events in a hydropower station provided by an embodiment of the present invention includes the following steps: Step 1: Based on the basic data of the study area, a knowledge graph model layer of abnormal events in hydropower stations is constructed to formally describe the entities, relationships, and attributes of entities and relationships.

[0027] In this step, the emergency response plan texts, regulations and dispatching procedures in the study area are collected, and a top-down approach is used to analyze the entity types, attribute types and entity relationships in the fields of equipment parameters, operating procedures and abnormal events, and to construct a knowledge graph model layer framework for emergency response to abnormal events in hydropower stations.

[0028] In some embodiments of the present invention, Figure 2 As shown in the figure, the model layer framework of the knowledge graph of abnormal events in hydropower stations mainly includes eight entity types, namely, hydropower station groups, hydropower station entities, operating procedures, hydropower station generators and transformers and other equipment and their parameters, daily maintenance, abnormal phenomena, abnormal causes and processing procedures, and six entity relationships between them, namely, "include", "abnormal", "phenomenon", "cause", "processing" and "maintenance". The attribute type is not directly reflected in the model layer framework. As additional information connected to specific entity nodes, the attribute type is usually expressed and stored at the entity instance level. Such a design structure helps to separate the relationship between entities from the attribute data in the knowledge graph to improve the scalability and flexibility of the graph.

[0029] Step 2: Automatically identify entities through text classification models and classify them.

[0030] The emergency response plan texts, rules and regulations texts, and dispatch procedure texts of the study area are organized into a data set, which is used to construct a text classification model based on a support vector machine (SVM model).

[0031] The construction process of the text classification model is as follows Figure 3 As shown in the figure, first load the sorted data set, the text classification model traverses each line in the data set, completes text preprocessing through regular expressions combined with word segmentation tools (such as jieba word segmentation), and obtains the preprocessed data set; then the data in the preprocessed data set is vectorized and converted into feature vectors, and then the term frequency-inverse document frequency (TF-IDF) method is used to extract the features of the field; the vectorized data is passed to the text classification model for model training, and then the model parameters are optimized through cross-validation. The performance of the trained text classification model is evaluated, and the accuracy, recall rate, F1 score and other indicators are obtained by classifying and predicting the test set data. The text classification model is continuously iterated according to the evaluation results, and finally the trained text classification model is saved.

[0032] The text classification model can map each data point to a high-dimensional space. The hyperplane in the high-dimensional space maximizes the interval between data points of different categories. The mathematical model of the text classification model can be expressed as: ; For all training samples in the dataset ( , ),satisfy: ; Term Frequency-Inverse Document Frequency (TF-IDF) is a statistical method used to evaluate the importance of a word in a data set to one of the data in a data set. In the text classification model, TF-IDF can effectively highlight the words with discriminativeness in certain documents by assigning a weight to each word, thereby improving the accuracy and effect of text classification. The calculation formula of TF-IDF is: ; In the formula, represents the normal vector of the hyperplane; represents the bias term; is the decision function in the text classification model; represents the feature vector, is the category label corresponding to the feature vector; To indicate a word; Represents a specific piece of data; Indicates the total amount of data in the dataset; Representing words In the data In and The product of Importance in data; Representing words In the data The frequency in Representing words In the dataset The inverse document frequency in .

[0033] In some embodiments of the present invention, the text of the procedures and plans of the four hydropower stations of Shannipo, Guangzhao, Mamaya and Dongqing in the Beipanjiang River Basin are used as the data source, and the text classification task is performed through the text classification model to identify 8 entity types such as the generator unit operating procedures, transformer operating procedures, and distribution device operating procedures of each hydropower station. The recognition results of each entity type are shown in Table 1.

[0034] Table 1 SVM model training results

[0035] Step 3: Use a cosine similarity-based method to perform knowledge fusion on the entities identified in step 2 to optimize the text classification results, and use a graph database as a medium to store text data and build a knowledge graph for emergency response to abnormal events.

[0036] Knowledge fusion refers to unifying multiple entities with the same meaning into a single entity. In the process of constructing the knowledge graph, the code texts of different hydropower stations were written by different people and at different times, resulting in inconsistent descriptions of some professional terms and abnormal phenomena. Therefore, the entities are unified through knowledge fusion. The mathematical model of knowledge fusion can be expressed as: ; In the formula, and are two vectors, and is an element in a vector, indicating that the vector and No. The elements at the positions, is the total number of elements.

[0037] In some embodiments of the present invention, the graph database uses the Neo4j graph database, which is a NoSQL graph database developed based on Java. It has a significant speed advantage when processing complex queries, is suitable for large-scale projects, and can reflect complex relationships between multiple entities. In other embodiments, other types of databases are also used, such as relational databases (MySQL, PostgreSQL), document databases (MongoDB), and other graph databases (ArangoDB, Amazon Neptune), etc. The database type is selected according to the data structure and query requirements.

[0038] In some embodiments of the present invention, by performing text classification on the text of the procedures and plans of the four hydropower stations, namely Shannipo, Guangzhao, Mamaya and Dongqing, various types of text entities are obtained, and then these text entities are subjected to knowledge fusion. Finally, all the text entities after knowledge fusion construct corresponding nodes in the form of RDF (entity, attribute and relationship) triples, and the constructed nodes and the relationships between the nodes are written into the Neo4j graph database through the Cypher query language to form a complete knowledge graph for emergency response to abnormal events. Combined with the retrieval function of the graph database, efficient query and reasoning of emergency response knowledge to abnormal events can be achieved.

[0039] The storage results of the knowledge graphs for emergency response to some abnormal events of the four hydropower stations are shown in Figures 4 (a) to 4 (d), where Figure 4 (a) is the knowledge graph for emergency response to some abnormal events of the butterfly valve of Shannipo Hydropower Station, which shows abnormal events such as "automatic closure of butterfly valve". When the abnormal phenomenon is determined to be "the indication of the unit active power meter gradually decreases" or "the red light of the butterfly valve full open position indication is off, the butterfly valve is closing, and there is a strong vibration sound", the abnormal event can be determined to be "automatic closure of butterfly valve", and the corresponding processing flow is given, which is shown in the figure as the node "operation starting point". Figure 4 (b) is the knowledge graph for emergency response to some abnormal events of the cooling system of Dongqing Hydropower Station, which shows abnormal events such as "cooling water interruption". When the abnormal phenomenon is determined to be "the water pressure indication of each part decreases or there is no indication" or "the light sign of "central control room host computer briefing, unit cooling water interruption" is on, and voice alarm", the abnormal event can be determined to be "cooling water interruption", and the corresponding processing flow is given, which is shown in the figure as the node "operation starting point". Figure 4 (c) is a knowledge graph of emergency response to some abnormal events of the Mamaya Hydropower Station unit, which shows abnormal events such as "unit overspeed". When the abnormal phenomenon is determined to be "the unit has an overspeed sound" or "the guide vane may be closed, the unit active power and stator current indication may be zero, and the butterfly valve may be closed", the abnormal event can be determined to be "unit overspeed", and the corresponding processing flow is given, which is shown in the figure as "node operation starting point". The node is also connected to the corresponding processing flow, such as Figure 5As shown. Figure 4 (d) is a knowledge graph of emergency response to abnormal events of bearings in the Guangzhao Hydropower Station, which shows abnormal events such as "bearing temperature is too high". When the abnormal phenomenon is determined to be "the oil level of a certain bearing reaches or exceeds the fault oil level" or "the host computer briefing in the central control room, the "bearing oil level exceeds the limit" light-emitting sign is on, and the voice alarm is on", the abnormal event can be determined to be "bearing temperature is too high", and the corresponding processing flow is given, which is shown in the figure as the node "operation starting point".

[0040] The emergency response process of the above knowledge graph is as follows: Figure 5 As shown in the figure, the processing flow of the abnormal event of "unit overspeed" is taken as an example. When an abnormal event occurs during the daily operation of a hydropower station, the knowledge graph of emergency handling of abnormal events of each hydropower station executes the processing flow in sequence according to the entity relationship and edge transmission order in the graph. The dispatcher only needs to follow the preset processing flow according to the current different situations to complete the emergency handling of abnormal events. Figure 5 When an abnormal event is confirmed, the dispatcher starts from the node "operation starting point". The first step of the operation is "the centralized control dispatcher immediately monitors whether the overspeed protection device of the unit is operating normally". If the "overspeed protection is operating normally", "monitor the operation of the overspeed protection". If there are no other abnormalities, the emergency disposal is terminated; if the "overspeed protection refuses to operate or operates abnormally", "immediately start the emergency shutdown". After the shutdown, if it is found that "the unit is overspeeding due to the shear pin being cut off or the main pressure regulating valve being stuck", "monitor whether the pure mechanical overspeed protection is operating". If there are no other abnormalities, the emergency disposal is terminated; if it is found that "the unit cannot be stopped", "emergency closure of the water inlet gate with remote operation function is carried out, and the controlled emergency attendant is notified." For the power station side, "after the shutdown, the controlled power station shall take relevant safety measures, conduct a comprehensive inspection of the unit, comprehensively organize, record and analyze the cause of the accident for various signals, information and instructions, and report the inspection results to the centralized control dispatcher"; for the dispatcher, "report to the superior dispatching agency, monitor the situation of the overspeed unit, and order the on-site personnel to deal with it", and finally, if "the overspeed unit does not have the conditions for recovery", then "report to the superior dispatching agency and apply for isolation of the faulty equipment according to the on-site work requirements"; if "the overspeed unit has the conditions for recovery", then "report to the superior dispatching agency and determine whether the unit is connected to the grid according to the dispatching order"; at the same time, the "dispatcher" needs to "report the handling situation to the department head and make detailed records of the emergency handling situation."

[0041] Step 4: When an abnormal event occurs, the hydropower station control center will send a corresponding alarm message, receive the alarm message and calculate the similarity of the abnormal events involved in the alarm message through the entity similarity algorithm, return the entity with the highest similarity, and obtain the most similar abnormal event entity. In this way, the corresponding abnormal phenomenon, cause and processing flow of the abnormal event can be further obtained in the abnormal event emergency response knowledge graph.

[0042] In some embodiments of the present invention, the entity similarity algorithm adopts the Jaccard entity similarity algorithm, which is an algorithm for measuring the similarity of two sets. The similarity value is calculated by comparing the intersection and union of two entity sets. Its mathematical model can be expressed as: ; In the formula, Similarity value, and Represents a collection of attributes of two different entities.

[0043] Step 5: Build a visualization interface of the abnormal event emergency response knowledge graph through the back-end framework and the front-end framework.

[0044] In some embodiments of the present invention, the back-end framework and the front-end framework respectively adopt a back-end framework based on Django and a front-end framework based on Vue 3.0. The back-end framework based on Django can provide a complete back-end service architecture, which can better realize the knowledge graph query function and clearly return the results for the decision-making of the dispatcher. In other embodiments, other back-end frameworks (such as Flask, FastAPI and other back-end frameworks) can also be applied. The implementation of different frameworks is slightly different in specific query operations and architectural methods, but the core structure of the mathematical model remains consistent, all of which are based on the query conditions to filter qualified nodes or relationships from the knowledge graph. Among them, the mathematical model of the back-end framework based on Django can be expressed as: ; In the formula, Indicates the result set that meets the query conditions after the user's query; Represents a query operation in the knowledge graph of emergency response to abnormal events Search and return all the files that meet the query conditions Nodes or relationships; Representing knowledge graph of emergency response to abnormal events A node or relationship in Representing knowledge graph of emergency response to abnormal events The global set of nodes and relationships; is a Boolean function representing a node or relationship Whether it meets the query conditions .

[0045] The front-end framework based on Vue 3.0 is a flexible progressive framework for building human-computer interaction of system pages. In other embodiments, other front-end frameworks, such as React, Angular and other front-end frameworks, can also be applied. The differences between different front-end frameworks are mainly reflected in the specific implementation method, view update mechanism and state management. From the perspective of mathematical models, the core structure and expression of the model are consistent. The mathematical model of the front-end framework based on Vue 3.0 can be expressed as: ; In the formula, Represents the rendered visual page view; For a given frontend state Interaction with users Function of Represents the front-end status, including the data entered by the user and the data returned by the back-end; Represents a set of user interaction behaviors; is a Boolean function representing a visualization element Is it necessary to adjust the current status be displayed; Indicates that it contains the visual elements that should be displayed in the current state, that is, A collection of .

[0046] In some embodiments of the present invention, the process of visualizing the knowledge graph of emergency response to abnormal events is as follows: Figure 6 As shown in the figure. First, after the front-end framework based on Vue 3.0 receives the user's query request (the query request here can be a query manually entered by the dispatcher, or the monitoring system triggers an alarm and automatically performs an abnormal query after discovering an abnormal event), it is transmitted to the back-end framework through the HTTP protocol for data processing. Then the corresponding keywords are retrieved in the Neo4j graph database, and the back-end framework based on Django executes the corresponding Cypher query statement in the graph database to retrieve the entity node calculated by the entity similarity and the node associated with the node. These data are converted into JSON format and returned to the front-end through the application program interface (API). The front-end parses these JSON data and visualizes the user's query content. The front-end framework based on Vue 3.0 allows dispatchers to query abnormal events and return the corresponding disposal process to achieve visual auxiliary decision-making. The back-end framework based on Django processes the query content received by the front-end and returns the results to the front-end for display. The two use HTTP API to transmit information, so as to provide a visual interface for emergency disposal of abnormal events for hydropower station dispatchers and assist dispatchers in making decisions.

[0047] In some embodiments of the present invention, a knowledge graph-based emergency response device for abnormal events in a hydropower station is provided, which is used to implement the method provided in the above embodiments. The device includes the following modules: The pattern layer construction module is used to construct the knowledge graph pattern layer of abnormal events of hydropower stations in the study area, and to formally describe the entities, their relationships, and the attributes of the entities and their relationships; The entity recognition and classification module is used to identify entities through text classification models and classify entities; The knowledge graph construction module is used to perform knowledge fusion on the identified entities based on the cosine similarity method to optimize the text classification results, store text data in the graph database as a medium, and construct a knowledge graph for emergency response to abnormal events; The abnormal event identification module is used to calculate the similarity of the abnormal events involved in the alarm information when the abnormal event occurs, obtain the most similar abnormal event entity, and use the most similar abnormal event entity to obtain the corresponding abnormal phenomenon, cause and processing flow of the abnormal event in the abnormal event emergency disposal knowledge graph; The visualization module is used to visualize the knowledge graph of emergency response to abnormal events through the back-end framework and the front-end framework.

[0048] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined in the present invention may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown in the present invention, but will conform to the widest scope consistent with the principles and novel features disclosed in the present invention.

Claims

1. The emergency response method for abnormal events in hydropower stations based on knowledge graph is characterized by: The steps include: Construct a knowledge graph model layer of abnormal events of hydropower stations in the study area, and formally describe the entities, their relationships, and the attributes of entities and relationships; Identify entities through text classification models and classify entities; The knowledge of the identified entities is integrated through the cosine similarity method to optimize the text classification results, and the text data is stored in the graph database as a medium to build a knowledge graph for emergency response to abnormal events. Calculate the similarity of the abnormal events involved in the alarm information when the abnormal event occurs, obtain the most similar abnormal event entity, and use the most similar abnormal event entity to obtain the corresponding abnormal phenomenon, cause and processing flow of the abnormal event in the abnormal event emergency disposal knowledge graph; The knowledge graph of emergency response to abnormal events is visualized through the back-end framework and the front-end framework.

2. The method for emergency handling of abnormal events in a hydropower station based on knowledge graph according to claim 1 is characterized in that: When constructing the knowledge graph model layer of abnormal events in hydropower stations in the study area, the emergency response plan texts, regulations and dispatching procedures in the study area are collected, the entity types, attribute types and entity relationships of concern are analyzed, and the knowledge graph model layer framework for emergency response to abnormal events in hydropower stations is constructed.

3. The method for emergency handling of abnormal events in a hydropower station based on knowledge graph according to claim 1 is characterized in that: The text classification model is obtained as follows: The emergency response plan texts, rules and regulations texts, and dispatching procedures texts of the study area were organized into a data set; Traverse each row in the data set, complete text preprocessing by combining regular expressions with word segmentation tools, and obtain the preprocessed data set; The data in the preprocessed data set is vectorized and converted into feature vectors, and then the word frequency-inverse document frequency method is used to extract the features of the field; The vectorized data is passed into the text classification model for model training to obtain a trained text classification model.

4. The method for emergency handling of abnormal events in a hydropower station based on knowledge graph according to claim 1 is characterized in that: The text classification model can map each data point to a high-dimensional space. The hyperplane found in the high-dimensional space can maximize the interval between data points of different categories. The mathematical model of the text classification model is expressed as: ； For all training samples in the dataset ( , ),satisfy: ； In the formula, represents the normal vector of the hyperplane; represents the bias term; is the decision function in the text classification model; Indicates feature vectors, It is with The category label corresponding to the feature vector; The importance of a word in a data set to one of the data in a data set is evaluated by term frequency-inverse document frequency.

5. The method for emergency handling of abnormal events in a hydropower station based on knowledge graph according to claim 1 is characterized in that: In the knowledge fusion of the identified entities based on the cosine similarity method, knowledge fusion means that multiple entities with the same meaning are uniformly represented as a single entity. The mathematical model of knowledge fusion is expressed as: ； In the formula, and are two vectors, and is an element in a vector, indicating that the vector and No. The elements at the positions, is the total number of elements.

6. The method for emergency handling of abnormal events in a hydropower station based on knowledge graph according to claim 1 is characterized in that: The steps for constructing the knowledge graph for emergency response to abnormal events include: constructing corresponding nodes in the form of RDF triples for all entities after knowledge fusion, writing the constructed nodes and their relationships into the graph database through the query language, and forming a complete knowledge graph for emergency response to abnormal events.

7. The method for emergency handling of abnormal events in a hydropower station based on knowledge graph according to claim 1 is characterized in that: The Jaccard entity similarity algorithm is used to calculate the similarity.

8. The method for emergency handling of abnormal events in a hydropower station based on a knowledge graph according to any one of claims 1 to 7, characterized in that: The backend framework uses a backend framework based on Django.

9. The method for emergency handling of abnormal events in a hydropower station based on a knowledge graph according to any one of claims 1 to 7, characterized in that: The front-end framework uses a front-end framework based on Vue 3.

0.

10. The knowledge graph-based emergency response device for abnormal events in a hydropower station is characterized by: For implementing the method according to any one of claims 1 to 9, the device comprises the following modules: The pattern layer construction module is used to construct the knowledge graph pattern layer of abnormal events of hydropower stations in the study area, and to formally describe the entities, their relationships, and the attributes of the entities and their relationships; The entity recognition and classification module is used to identify entities through text classification models and classify entities; The knowledge graph construction module is used to perform knowledge fusion on the identified entities based on the cosine similarity method to optimize the text classification results, store text data in the graph database as a medium, and construct a knowledge graph for emergency response to abnormal events; The abnormal event identification module is used to calculate the similarity of the abnormal events involved in the alarm information when the abnormal event occurs, obtain the most similar abnormal event entity, and use the most similar abnormal event entity to obtain the corresponding abnormal phenomenon, cause and processing flow of the abnormal event in the abnormal event emergency disposal knowledge graph; The visualization module is used to visualize the knowledge graph of emergency response to abnormal events through the back-end framework and the front-end framework.

Images