Data private network state reasoning method based on power production knowledge graph

Abstract

The invention discloses a data private network state reasoning method based on an electric power production knowledge graph. The method comprises the following steps: 1) constructing the electric power production knowledge graph; 2) data acquisition; collecting operation state monitoring data of equipment from entity equipment of various power system components through a standardized interface, wherein the operation state monitoring data comprises the state, load data and environment data of the power system components; 3) associating the collected equipment operation state monitoring data with the knowledge graph through a data private network, and updating the power production knowledge graph based on the data collected in real time; and 4) analyzing the current state of the power system by adopting an inference engine, and predicting a potential fault. According to the method, the power production knowledge graph is constructed and data private network state reasoning is performed based on the power production knowledge graph, so that the operation and maintenance efficiency of the power system and the accuracy of fault diagnosis are remarkably improved.

Description

Technical Field

[0001] The present invention relates to natural language processing technology, and in particular to a method for inferring the state of a data private network based on a power production knowledge graph. Background Art

[0002] With the development of the information and digital age, the level of intelligence and digitization of the power system has been continuously improved. The application of intelligent power transmission and distribution technologies has significantly improved the operation efficiency and reliability of the power grid. At the same time, the wide application of data technology, artificial intelligence, and Internet of Things technology provides strong support for real-time monitoring, predictive analysis, and fault diagnosis, further promoting the development of the power system towards intelligence and automation.

[0003] In the field of power system monitoring and fault prediction, traditional methods face multiple challenges, which limit the accuracy and intelligence level of system state analysis and prediction. First, existing methods usually rely on simple thresholds and rules to detect system anomalies, but this method lacks a deep understanding of the complex structure and changing environment of the power system, resulting in the inability to comprehensively capture the dynamic changes of system operation. Second, the data sources in the power system are extensive and highly heterogeneous, and the inconsistency of data quality and format makes data integration and analysis complex and difficult, limiting the accuracy of real-time monitoring and fault prediction. Third, traditional fault prediction methods often rely on static models or simple statistical analysis and cannot effectively handle complex interactions and emergencies, resulting in insufficient response speed and accuracy.

[0004] To effectively address these problems, an intelligent method that can monitor the state of the power system in real time and predict potential faults in the data private network state is needed. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide a method for inferring the state of a data private network based on a power production knowledge graph in view of the deficiencies in the prior art.

[0006] The technical solution adopted by the present invention to solve its technical problems is: A method for inferring the state of a data private network based on a power production knowledge graph, comprising the following steps:

[0007] 1) Construct a power production knowledge graph; including entity definitions, relationships, and attribute features of power system components;

[0008] Among them, the entities of power system components include generators, transformers, and lines;

[0009] The relationship is to describe the connection and dependence between entities according to the topological structure of the power system;

[0010] Entities are represented by nodes and relationships by edges. Through the combination of nodes and edges, "entity-relationship-entity" triples and a power production knowledge graph are formed;

[0011] 2) Data collection; collecting operation status monitoring data of devices from the entity devices of various power system components through a standardized interface, including the status, load data, and environmental data of power system components;

[0012] The status of the power system components includes normal and faulty;

[0013] The load data includes voltage, current, and power;

[0014] The environmental data includes temperature and humidity;

[0015] 3) Through a data private network, associate the collected device operation status monitoring data with the knowledge graph, and update the power production knowledge graph based on the real-time collected data;

[0016] 4) Based on the power production knowledge graph, use an inference engine to analyze the current state of the power system and predict potential faults;

[0017] The inference engine consists of a rule engine, a machine learning model, and a graph neural network model (GNN);

[0018] The inference engine performs in-depth inference analysis on the data in the knowledge graph based on the power production knowledge graph and preset inference rules, combined with the graph neural network model.

[0019] According to the above solution, in step 3), establish a connection between the real-time collected data and the corresponding nodes in the knowledge graph to achieve the dynamic association of data and the knowledge graph, ensure that real-time data can be correctly introduced into the knowledge graph, and form a comprehensive system view; the knowledge graph receives real-time data from data collection and uses a parsing algorithm based on a graph database to map this data to the corresponding nodes and edges of the power production knowledge graph;

[0020] The update process of the knowledge graph includes incremental update and full update. The incremental update is used to process minor changes detected in real time, such as temperature fluctuations or voltage changes of a certain device, while the full update is performed when there is a structural adjustment or new device addition in the power production system to ensure that the information in the graph always accurately reflects the actual state of the current system.

[0021] According to the above solution, in step 4), specifically as follows:

[0022] The inference engine first filters and preliminarily processes the real-time data through the rule engine according to the preset rules defined by the historical operation data, threshold conditions, and fault modes of the devices of the power system components;

[0023] Then, use a machine learning model to learn from historical operation data, predict potential faults, and identify changes in fault trends in the system;

[0024] On this basis, the graph neural network model conducts reasoning: The graph neural network model captures the relationships between components and devices in the power system through information propagation between nodes and edges, generates feature vectors of the power system state, and performs fault diagnosis on possible abnormal situations based on the inferred power system state results. Use the results of the machine learning model to optimize the graph neural network model, and generate warning information or maintenance suggestions for the diagnosis results of the optimized graph neural network model to help respond to potential risks in a timely manner.

[0025] According to the above solution, in step 4), the reasoning process of the graph neural network model includes the following steps:

[0026] First is the analysis of the power system topology structure. Through the graph neural network, identify the interdependent relationships between components and devices in the power system, and analyze the possible impact of a fault in a certain device on the overall system. Then determine which devices or areas to disconnect first to reduce the risk of fault spread;

[0027] Secondly, combine real-time load data and environmental data (temperature, humidity, etc.) to evaluate the impact on the device and the device state, generate feature vectors of the power system state, and predict the risk of possible device failures as the diagnosis result;

[0028] Finally, conduct a time-series analysis of the fault diagnosis results. Combine historical data and the operation trajectory of the device to identify potential fault trends, and through the analysis of long-term monitoring data of the device, identify abnormal behaviors or performance degradation of the device in advance.

[0029] The beneficial effects of the present invention are:

[0030] The present invention significantly improves the operation and maintenance efficiency of the power system and the accuracy of fault diagnosis by constructing a power production knowledge graph and performing reasoning on the data private network state based on this. The system uses the graph neural network algorithm to be able to monitor and analyze various types of data in power production in real time, discover potential fault risks in a timely manner, and generate effective maintenance suggestions.

[0031] The present invention shows higher flexibility, accuracy, and timeliness in a complex power environment, provides strong technical support for the intelligent operation and maintenance of the power system. The reasoning engine combines a dynamically updated knowledge graph to ensure the system's rapid response ability to emergencies, thereby greatly improving the reliability of the power system. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] The present invention will be further described below in conjunction with the drawings and embodiments. In the drawings:

[0033] Figure 1 is the method flowchart of an embodiment of the present invention. Detailed implementation manners

[0034] In order to make the objectives, technical solutions and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below in conjunction with embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0035] As Figure 1 shown, a data private network status inference method based on a power production knowledge graph includes the following steps:

[0036] 1) Construct a power production knowledge graph; including entity definitions, relationships, and attribute characteristics of power system components;

[0037] Among them, the entities of power system components include generators, transformers, and lines;

[0038] The relationship is to describe the connection and dependence between entities according to the topological structure of the power system;

[0039] Using nodes to represent entities and edges to represent relationships, through the combination of nodes and edges, a "entity-relationship-entity" triple and a power production knowledge graph are formed;

[0040] 2) Data collection; collect the operation status monitoring data of the equipment from the entity devices of various power system components through a standardized interface, including the status, load data, and environmental data of the power system components;

[0041] The status of the power system components includes normal and faulty;

[0042] The load data includes voltage, current, and power;

[0043] The environmental data includes temperature and humidity;

[0044] Data collection first establishes a connection with various devices in the power system through the hardware interface of the communication protocol adaptation module, and confirms whether the connection is successful through the interface status detection function. After success, the operation status data of the device is obtained in real time through a standardized communication protocol (such as TCP / IP, Modbus, etc.). Data collection also supports obtaining data from remote sensors and control devices through Internet of Things (IoT) technology and SCADA systems, further enhancing the breadth and depth of data and ensuring comprehensive monitoring of the system status.

[0045] 3) Through the data private network, associate the collected device operation status monitoring data with the knowledge graph, and update the power production knowledge graph based on the real-time collected data;

[0046] Establish connections between the real-time collected data and the corresponding nodes in the knowledge graph to achieve the dynamic association between the data and the knowledge graph, ensure that the real-time data can be correctly introduced into the knowledge graph, and form a comprehensive system view; the knowledge graph receives the real-time data from data collection and uses the parsing algorithm based on the graph database to map this data to the corresponding nodes and edges of the power production knowledge graph;

[0047] The update process of the knowledge graph includes incremental update and full update. The incremental update is used to process the tiny changes detected in real time, such as the temperature fluctuation or voltage change of a certain device, while the full update is performed when there is a structural adjustment or new device addition in the power production system, ensuring that the information in the graph always accurately reflects the actual state of the current system.

[0048] 4) Use an inference engine to analyze the current state of the power system and predict potential faults;

[0049] The inference engine consists of a rule engine, a machine learning model, and a graph neural network model (GNN);

[0050] The inference engine is based on the power production knowledge graph and the preset inference rules, and combines with the graph neural network model to conduct in-depth inference and analysis on the data in the knowledge graph.

[0051] Step 4) is specifically as follows:

[0052] The inference engine first passes through the rule engine to screen and preliminarily process the real-time data according to the preset rules defined by the historical operation data, threshold conditions, and fault modes of the devices of the power system components;

[0053] Then, use the machine learning model to learn the historical operation data, predict potential faults, and identify the change trend of faults in the system;

[0054] On this basis, the graph neural network model conducts inference: the graph neural network model captures the relationships between the devices of each component of the power system through the information propagation between nodes and edges, generates the feature vectors of the power system state, and conducts fault diagnosis on the possible abnormal situations according to the inference results of the power system state. Use the results of the machine learning model to optimize the graph neural network model, and generate early warning information or maintenance suggestions for the diagnosis results of the optimized graph neural network model to help respond to potential risks in a timely manner.

[0055] The inference process of the graph neural network model includes the following steps:

[0056] First, it is the analysis of the power system topology. Through the graph neural network, the interdependencies between the components and devices of the power system are identified, and the possible impact of a fault in a certain device on the overall system is analyzed. Furthermore, it is determined which devices or areas should be disconnected first to reduce the risk of fault spread.

[0057] Secondly, combining real-time load data and environmental data (such as temperature, humidity, etc.), the impact on the devices and the device status are evaluated, a feature vector of the power system state is generated, and the risk of possible device failures is predicted as the diagnostic result.

[0058] Finally, a time-series analysis is performed on the fault diagnosis results. Combining historical data and the operation trajectories of the devices, potential fault trends are identified. Through the analysis of the long-term monitoring data of the devices, abnormal behaviors or performance degradations of the devices are identified in advance.

[0059] In this process, the inference engine operates based on a series of preset inference rules. The set inference rules include device fault models, load change impact rules, temperature and humidity change response rules, etc. These rules are preset by the expert system according to device characteristics, historical fault patterns, and the operation experience of the power system. For example, when the load of a certain device exceeds its rated power, the inference engine will, according to the load change impact rule, judge whether there is a potential risk of failure for this device. In addition, the rule engine will also evaluate the response rules according to the changes in environmental temperature or humidity, combined with the physical characteristics of the device, to predict the fault risk caused by environmental factors. Through the setting of these rules, the inference engine can conduct comprehensive analysis based on real-time data and generate the most accurate fault prediction and early warning information.

[0060] Once the inference engine identifies potential faults or abnormal situations, the system will generate relevant instruction information and convert the instructions into the communication protocol format that the target device can accept (such as IEC 61850, DNP3, etc.) through the communication protocol adaptation module. These instructions transmit the information to the power devices through standardized communication interfaces (such as Ethernet interfaces, analog interfaces, serial ports, etc.) to perform corresponding maintenance operations.

[0061] Considering the complexity of the power system operation environment (such as temperature and humidity changes may lead to performance degradation of devices, etc.), the present invention conducts state analysis and fault prediction by comprehensively considering various factors through an inference algorithm based on a knowledge graph. When the system detects abnormalities or potential faults, it automatically triggers an alarm mechanism and takes predetermined measures, thereby ensuring the stable operation and fault prevention of the power system.

[0062] It should be understood that for those of ordinary skill in the art, improvements or transformations can be made according to the above description, and all such improvements and transformations should fall within the protection scope of the appended claims of the present invention.

Claims

1. A data private network state reasoning method based on power production knowledge graph, characterized in that: The following steps are involved: 1) Construct a knowledge graph of power production, including entity definitions, relationships, and attribute characteristics of power system components; Among them, the entities of power system components include generators, transformers, and lines; The relationship describes the connection and dependency between entities according to the topological structure of the power system; Nodes represent entities, edges represent relationships, and through the combination of nodes and edges, a triple of "entity-relationship-entity" and a knowledge graph of power production are formed; 2) Data collection: Collecting equipment operation status monitoring data from the physical equipment of various power system components, including the status of power system components, load data and environmental data; The states of the power system components include normal and faulty; The load data includes voltage, current, and power; The environmental data includes temperature and humidity; 3) Through the data dedicated network, the collected equipment operation status monitoring data is associated with the knowledge graph, and the power production knowledge graph is updated based on the real-time collected data; 4) Based on the knowledge graph of power production, an inference engine is used to analyze the current state of the power system and predict potential failures; The inference engine consists of a rule engine, a machine learning model, and a graph neural network model; The inference engine is based on the power production knowledge graph and preset inference rules, combined with the graph neural network model, to perform deep reasoning analysis on the data in the knowledge graph to predict potential failures.

2. The data private network state reasoning method based on the power production knowledge graph according to claim 1 is characterized in that: In the step 3), the real-time collected data is linked to the corresponding nodes in the knowledge graph to achieve dynamic association between the data and the knowledge graph, ensuring that the real-time data can be correctly introduced into the knowledge graph to form a comprehensive system view; The knowledge graph receives real-time data from data collection and uses a parsing algorithm based on a graph database to map these data to the corresponding nodes and edges of the power production knowledge graph; The update process of the knowledge graph includes incremental updates and full updates. The incremental updates are used to process small changes monitored in real time, such as temperature fluctuations or voltage changes of a device, while the full updates are performed when the power production system undergoes structural adjustments or new equipment is added, ensuring that the information in the graph always accurately reflects the actual status of the current system.

3. The data private network state reasoning method based on the power production knowledge graph according to claim 1 is characterized in that: In the step 4), the details are as follows: The reasoning engine first screens and preliminarily processes the real-time data through the rule engine based on the historical operating data of the equipment, threshold conditions, and preset rules defined by the failure modes of the power system components; Then, the machine learning model is used to learn from historical operation data to predict potential failures and identify changes in failure trends in the system; On this basis, the graph neural network model performs reasoning: the graph neural network model captures the relationship between the components of the power system through information propagation between nodes and edges, generates a characteristic vector of the power system state, and predicts faults for possible abnormal situations based on the inference results of the power system state. The graph neural network model is optimized using the results of the machine learning model, and the diagnostic results of the optimized graph neural network model are used to generate early warning information or maintenance recommendations.

4. The data private network state reasoning method based on the power production knowledge graph according to claim 3 is characterized in that: In step 4), the reasoning process of the graph neural network model includes the following steps: The first is to analyze the topology of the power system. Through the graph neural network, the interdependence between the components of the power system is identified, and the impact of the failure of a certain device on the entire system is analyzed, so as to determine which devices or areas should be disconnected first to reduce the risk of fault spread. Secondly, by combining real-time load data and environmental data, the impact on the equipment and the equipment status are evaluated, the characteristic vector of the power system status is generated, and the risk of equipment failure is predicted as a diagnostic result; Finally, the fault diagnosis results are analyzed in time series, combined with historical data and the equipment's operating trajectory to identify potential fault trends and identify abnormal behavior or performance degradation of the equipment in advance.

5. An electronic device, characterized in that: include: one or more processors; as well as a storage device for storing one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors are caused to execute the method according to any one of claims 1 to 4.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 4 is implemented.

Images