A method and system for generating a knowledge graph of power grid equipment standards

Abstract

The present application provides a method and system for generating a standard knowledge graph for power grid equipment, which can intelligently integrate and optimize power grid equipment knowledge. If the target power grid data node that needs to be integrated is an existing data node in the reference power grid equipment standard knowledge graph, the knowledge vector is inferred by using the characteristics of the forward data node or the dependent data node through the power grid equipment feature expansion mechanism, thereby optimizing the reference power grid equipment standard knowledge graph. This not only improves the accuracy of the reference power grid equipment standard knowledge graph, but also realizes the standardization and intelligent management of power grid equipment knowledge, so as to construct an optimized power grid equipment standard knowledge graph as a powerful service engine, namely, a middle-end capability semantic network, so as to realize subsequent intelligent recommendation and association of standards, and provide an efficient and intelligent solution for power grid equipment management.

Description

Technical Field

[0001] The present invention relates to the field of knowledge graph technology, and in particular to a method and system for generating a standard knowledge graph for power grid equipment. Background Art

[0002] With the rapid development of information technology such as OCR technology and NLP technology, as well as the in-depth promotion of smart grid construction, OCR technology and natural language technology can be conveniently used to collect data from power grid equipment. This makes the amount of data from power grid equipment increasingly large and complex. How to effectively manage and use this data has become an important challenge facing the power industry. As a new type of data structure, knowledge graphs can display the relationship between entities in a graphical way, providing a new perspective for data organization and query. In power grid equipment management, the introduction of knowledge graph technology can help enterprises better understand and analyze the relationship between equipment, optimize resource allocation, and improve operational efficiency.

[0003] However, there are still many difficulties in the construction of the knowledge graph of power grid equipment. On the one hand, there are many types of power grid equipment with different attributes. How to standardize these devices and their attributes and integrate them into the knowledge graph is an urgent problem to be solved. On the other hand, with the continuous updating and changes of power grid equipment, the knowledge graph needs to be continuously expanded and updated to maintain its timeliness and accuracy. In addition, for the power grid data nodes that already exist in the knowledge graph, how to effectively use their existing information for feature expansion and optimization is also an important research direction. Summary of the invention

[0004] In order to at least overcome the above-mentioned deficiencies in the prior art, an embodiment of the present application aims to provide a method and system for generating a standard knowledge graph of power grid equipment.

[0005] According to one aspect of the present application, a method for generating a knowledge graph of power grid equipment standards is provided, the method comprising:

[0006] When constructing a grid equipment standard knowledge graph, if the target grid data node that needs to be integrated currently is an existing data node in the reference grid equipment standard knowledge graph, then determine the grid equipment feature extension mechanism of the target grid data node, the grid equipment feature extension mechanism includes at least one of a grid equipment feature extension mechanism based on a forward data node and a grid equipment feature extension mechanism based on X dependent data nodes, the forward data node is an integrated data node that is located before the target grid data node in the node integration order of the reference grid equipment standard knowledge graph, and X is a positive integer;

[0007] Inferring the knowledge vector of the target power grid data node according to the power grid device feature extension mechanism of the target power grid data node, and generating inferred knowledge vector parameters of the target power grid data node;

[0008] Based on the inferred knowledge vector parameters of the target power grid data node, the reconstructed knowledge parameters of the target power grid data node are determined to optimize and generate the reference power grid equipment standard knowledge graph.

[0009] According to one aspect of an embodiment of the present application, a power grid online service system is provided, which includes a processor and a machine-readable storage medium, wherein the machine-readable storage medium stores machine-executable instructions, and the machine-executable instructions are loaded and executed by the processor to implement a method for generating a standard knowledge graph for power grid equipment in any one of the aforementioned possible implementations.

[0010] According to one aspect of the embodiments of the present application, a computer program product or a computer program is provided, the computer program product or the computer program includes computer instructions, and the computer instructions are stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, so that the computer device executes the methods provided in various optional implementations of the above three aspects.

[0011] In the technical solutions provided by some embodiments of the present application, the embodiments of the present application can flexibly select appropriate feature extension methods according to the characteristics of the data node by determining the grid equipment feature extension mechanism of the target grid data node, whether it is based on the forward data node or the dependent data node, it can more accurately capture the features of the grid equipment, thereby improving the accuracy of knowledge graph construction. According to the grid equipment feature extension mechanism of the target grid data node, its knowledge vector is inferred to generate inferred knowledge vector parameters, which not only enriches the node information in the knowledge graph, but also enables the graph to better reflect the actual characteristics and relationships of the grid equipment. By determining the reconstructed knowledge parameters of the target grid data node based on the inferred knowledge vector parameters, the reference grid equipment standard knowledge graph can be optimized and generated, which not only improves the integrity and accuracy of the graph, but also makes it closer to the complex relationship network of the actual grid equipment, providing a more solid foundation for subsequent grid analysis and management. By constructing and optimizing the grid equipment standard knowledge graph, it provides strong support for the intelligent management of the grid. Through the rich node information and accurate relationship expression in the grid equipment standard knowledge graph, managers can have a deeper understanding of the operating status and mutual relationship of grid equipment, so as to make more scientific and reasonable decisions. Therefore, in the process of generating the standard knowledge graph of power grid equipment, the construction quality of the knowledge graph is significantly improved through technical means such as flexible feature extension mechanism, effective knowledge vector inference and graph optimization generation.

[0012] That is, it is able to intelligently integrate and optimize power grid equipment knowledge. If the target power grid data node that needs to be integrated is an existing data node in the reference power grid equipment standard knowledge graph, the knowledge vector is inferred through the power grid equipment feature expansion mechanism using the features of the forward data node or the dependent data node, thereby optimizing the reference power grid equipment standard knowledge graph. This not only improves the accuracy of the reference power grid equipment standard knowledge graph, but also realizes the standardization and intelligent management of power grid equipment knowledge, so as to construct the optimized power grid equipment standard knowledge graph as a powerful service engine, namely the middle-office capability semantic network, so as to realize the subsequent intelligent recommendation and association of standards, and provide an efficient and intelligent solution for power grid equipment management. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required to be used in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other relevant drawings can be extracted in combination with these drawings without creative work.

[0014] Figure 1A schematic diagram of a flow chart of a method for generating a standard knowledge graph for power grid equipment provided in an embodiment of the present application;

[0015] Figure 2 A schematic diagram of the architecture of a power grid online service system for implementing the above-mentioned power grid equipment standard knowledge graph generation method provided in an embodiment of the present application. DETAILED DESCRIPTION

[0016] The following description is intended to enable one of ordinary skill in the art to implement and combine the present application, and the description is provided in the context of a specific application scenario and its requirements. It will be apparent to one of ordinary skill in the art that various changes may be made to the disclosed embodiments, and that the general principles defined in the present application may be applied to other embodiments and application scenarios without departing from the principles and scope of the present application. Therefore, the present application is not limited to the described embodiments, but should be given the broadest scope consistent with the claims.

[0017] Figure 1 This is a flow chart of a method for generating a standard knowledge graph for power grid equipment provided by an embodiment of the present application. The method for generating a standard knowledge graph for power grid equipment is introduced in detail below.

[0018] Step S110, when constructing a power grid equipment standard knowledge graph, if the target power grid data node that currently needs to be integrated is an existing data node in the reference power grid equipment standard knowledge graph, then determine the power grid equipment feature extension mechanism of the target power grid data node, the power grid equipment feature extension mechanism includes at least one of a power grid equipment feature extension mechanism based on a forward data node and a power grid equipment feature extension mechanism based on X dependent data nodes, the forward data node is an integrated data node that is located before the target power grid data node in the node integration order of the reference power grid equipment standard knowledge graph, and X is a positive integer.

[0019] In detail, the grid equipment standard knowledge graph is a structured, graphical knowledge system that displays the relationships, attributes, and related information of grid equipment, which helps to better understand and analyze the complex relationships of the grid system. For example, a transformer may be connected to multiple circuit breakers, disconnectors, and other devices. In the grid equipment standard knowledge graph, these relationships can be intuitively displayed.

[0020] The target grid data node refers to a specific grid equipment data point that is currently being processed or prepared for integration when building or expanding the grid equipment standard knowledge graph. Assuming that a data node about "transformer" is being integrated, then this "transformer" is the current target grid data node.

[0021] The existing data node is a data point that already exists in the power grid equipment standard knowledge graph and has been integrated before. For example, if the power grid equipment standard knowledge graph already contains a node about "circuit breaker", and its related attributes and relationships have been defined and integrated, then this "circuit breaker" node is an existing data node.

[0022] The grid equipment feature extension mechanism is a method or rule for extending or improving the features and attributes of the target grid data node, which can be based on existing information or other related data nodes. For example, for a "transformer" node, its attributes can be extended according to certain features of the "circuit breaker" or "disconnector" connected to it, for example, the maximum output current of the transformer can be estimated according to the rated current of the connected circuit breaker.

[0023] The forward data node is a data node that has been integrated and is located before the current target power grid data node in the node integration order of the reference power grid equipment standard knowledge graph. For example, if "circuit breaker" is integrated first, and then "transformer", then in this context, "circuit breaker" is the forward data node of "transformer".

[0024] The dependent data nodes refer to other data nodes that have a direct or indirect relationship with the target power grid data node, and can provide important feature or attribute information for the target node. For example, in the power grid, a "transformer" may rely on a "circuit breaker" to ensure its safe operation, so the "circuit breaker" can be regarded as a dependent data node of the "transformer".

[0025] Therefore, in one scenario of this embodiment, in the process of constructing the grid equipment standard knowledge graph, the server will first check whether the target grid data node that needs to be integrated already exists in the reference grid equipment standard knowledge graph. For example, suppose the server is processing a data node about a transformer, and after checking, it is found that this data node already exists in the grid equipment standard knowledge graph. At this time, the server can determine the grid equipment feature extension mechanism of this existing transformer data node.

[0026] According to the previous integration order, the server can identify the integrated data nodes that are located before the transformer data node in the standard knowledge graph of power grid equipment, such as a power station data node. These data nodes are called forward data nodes. At the same time, the server will also identify other data nodes that have a dependency relationship with the transformer data node, such as the transmission line data node or distribution station data node connected to it, which are called dependent data nodes.

[0027] The server can determine the power grid equipment feature extension mechanism of the transformer data node based on the information of these forward data nodes and dependent data nodes. The power grid equipment feature extension mechanism can be based on the features of the forward data nodes only, or based on the features of a certain number of dependent data nodes, or a combination of the two.

[0028] Step S120 , inferring the knowledge vector of the target power grid data node according to the power grid device feature extension mechanism of the target power grid data node, and generating inferred knowledge vector parameters of the target power grid data node.

[0029] In detail, a knowledge vector is a mathematical representation used to capture and quantify the characteristics of an entity. For example, the knowledge vector can represent various attributes, characteristics, and relationships of power grid equipment, usually in the form of a multidimensional vector, and can be constructed based on the text description, attributes, relationships, and other related information of the entity. For example, the knowledge vector of a transformer may include its model, capacity, voltage level, manufacturer, and other attributes, which form a unique vector representation in a multidimensional space.

[0030] The inferred knowledge vector parameters refer to the specific parameter values ​​of the knowledge vector of the target power grid data node inferred based on the power grid equipment feature extension mechanism, which reflects the position, attributes, and relationship of the data node with other nodes in the knowledge graph. For example, for a transformer node, its inferred knowledge vector parameters may include the maximum load capacity, working efficiency, etc. inferred from the specific attributes of the connected circuit breaker or disconnector.

[0031] Therefore, in one scenario of this embodiment, after determining the grid equipment feature extension mechanism, the server can start to infer the knowledge vector of the existing transformer data node. If the extension mechanism is based on the forward data node, the server can use the knowledge vector of the power station data node to infer the knowledge vector of the transformer data node. For example, the output power of the power station may affect the selection and use of the transformer, so this information can be used to infer certain characteristics of the transformer.

[0032] If the extension mechanism is based on data-dependent nodes, the server can consider the data of the transmission line or substation connected to the transformer. For example, the capacity and length of the transmission line may affect the selection and configuration of the transformer, and this information can also be used to infer the knowledge vector of the transformer.

[0033] During the inference process, the server can use machine learning algorithms or deep learning models to ensure the accuracy and reliability of inference.

[0034] Step S130, determining the reconstructed knowledge parameters of the target power grid data node based on the inferred knowledge vector parameters of the target power grid data node, so as to optimize and generate the reference power grid equipment standard knowledge graph.

[0035] In detail, the reconstructed knowledge parameters refer to further determining or adjusting the representation of the target power grid data nodes in the knowledge graph based on the inferred knowledge vector parameters, which may involve the update or optimization of the node's location, connection relationship, attribute value, etc., aiming to improve the accuracy and completeness of the knowledge graph. For example, based on the inferred knowledge vector parameters of the transformer, it may be found that the connection relationship between the transformer and certain circuit breakers needs to be updated, or that some of its attribute values ​​(such as rated capacity) need to be adjusted to more accurately reflect the actual situation. These adjusted and updated parameters are the reconstructed knowledge parameters.

[0036] That is, in this embodiment, once the server completes the inference of the knowledge vector of the transformer data node and generates the inferred knowledge vector parameters, these inferred knowledge vector parameters can be used to determine the reconstructed knowledge parameters of the transformer data node. This process involves correcting, enhancing or supplementing the original knowledge vector to ensure the accuracy and completeness of the knowledge graph.

[0037] For example, based on the information of the forward data nodes and dependent data nodes, the server may find that the settings of certain parameters in the original knowledge vector are not accurate or comprehensive enough. At this time, the server will use the inferred knowledge vector parameters to correct or supplement these parameters. The corrected knowledge vector will more accurately reflect the actual characteristics and working environment of the transformer. Finally, the server can integrate these reconstructed knowledge parameters into the reference power grid equipment standard knowledge graph, thereby completing the optimization generation of the reference power grid equipment standard knowledge graph. This process not only improves the quality and accuracy of the reference power grid equipment standard knowledge graph, but also provides more powerful data support for subsequent power grid equipment management and decision-making.

[0038] Based on the above steps, the embodiment of the present application can flexibly select the appropriate feature extension method according to the characteristics of the data node by determining the grid equipment feature extension mechanism of the target grid data node. Whether it is based on the forward data node or the dependent data node, it can more accurately capture the characteristics of the grid equipment, thereby improving the accuracy of the knowledge graph. According to the grid equipment feature extension mechanism of the target grid data node, its knowledge vector is inferred to generate inferred knowledge vector parameters, which not only enriches the node information in the knowledge graph, but also enables the graph to better reflect the actual characteristics and relationships of the grid equipment. By determining the reconstructed knowledge parameters of the target grid data node based on the inferred knowledge vector parameters, the reference grid equipment standard knowledge graph can be optimized and generated, which not only improves the integrity and accuracy of the graph, but also makes it closer to the complex relationship network of the actual grid equipment, providing a more solid foundation for subsequent grid analysis and management. By constructing and optimizing the grid equipment standard knowledge graph, it provides strong support for the intelligent management of the grid. Through the rich node information and accurate relationship expression in the grid equipment standard knowledge graph, managers can have a deeper understanding of the operating status and mutual relationship of grid equipment, so as to make more scientific and reasonable decisions. Therefore, in the process of generating the standard knowledge graph of power grid equipment, the construction quality of the knowledge graph is significantly improved through technical means such as flexible feature extension mechanism, effective knowledge vector inference and graph optimization generation.

[0039] That is, it is able to intelligently integrate and optimize power grid equipment knowledge. If the target power grid data node that needs to be integrated is an existing data node in the reference power grid equipment standard knowledge graph, the knowledge vector is inferred through the power grid equipment feature expansion mechanism using the features of the forward data node or the dependent data node, thereby optimizing the reference power grid equipment standard knowledge graph. This not only improves the accuracy of the reference power grid equipment standard knowledge graph, but also realizes the standardization and intelligent management of power grid equipment knowledge, so as to construct the optimized power grid equipment standard knowledge graph as a powerful service engine, namely the middle-office capability semantic network, so as to realize the subsequent intelligent recommendation and association of standards, and provide an efficient and intelligent solution for power grid equipment management.

[0040] In a possible implementation manner, if the grid device feature extension mechanism of the target grid data node is the grid device feature extension mechanism based on the forward data node, step S120 includes:

[0041] Step S121 : determining a knowledge vector inference sub-mechanism of the target power grid data node, wherein the knowledge vector inference sub-mechanism includes at least one of a single feature inference mechanism and a composite feature inference mechanism.

[0042] Step S122: performing knowledge vector inference on the target power grid data node according to the knowledge vector inference sub-mechanism of the target power grid data node to generate inferred knowledge vector parameters of the target power grid data node.

[0043] In this embodiment, in this scenario, it is assumed that the server has identified the target power grid data node (for example, a specific transformer node) and determines its power grid equipment feature extension mechanism as a power grid equipment feature extension mechanism based on the forward data node. This means that the server will use the features of the forward data node (such as the integrated circuit breaker node, disconnector node, etc.) that is located before the transformer node in the power grid equipment standard knowledge graph to extend the features of the target node.

[0044] The server then needs to determine how to infer the knowledge vector for the target power grid data node (transformer node). Here, the server can select a knowledge vector inference sub-mechanism, which can be at least one of a single feature inference mechanism or a composite feature inference mechanism. Specifically:

[0045] Single feature inference mechanism: The server may infer a feature of the target transformer node (such as the maximum allowable current) based on only one key forward data node feature (such as the rated current of the circuit breaker). For example, the server may infer the maximum allowable current of the transformer based on the rated current of the connected circuit breaker.

[0046] Composite feature inference mechanism: The server may consider multiple features of multiple forward data nodes to infer the features of the target node. For example, the server can combine the rated current of multiple circuit breakers, the status of disconnectors, and the overall load of the power grid to comprehensively infer the operating status or potential risks of the transformer.

[0047] In this scenario, the server chooses the composite feature inference mechanism because it can provide more comprehensive and accurate inference results.

[0048] As a result, the server will now use the selected composite feature inference mechanism to perform knowledge vector inference on the target transformer node. First, the feature data of all relevant forward data nodes (such as circuit breakers, disconnectors) are collected, such as current, voltage, status, etc. Then, the server can use complex algorithms (such as machine learning models, statistical analysis, etc.) to comprehensively analyze these data and infer the knowledge vector parameters of the target transformer node.

[0049] For example, the server may find that when the connected circuit breaker has a larger rated current, the disconnector is in a closed state, and the overall load of the power grid is low, the transformer may have higher operating efficiency and lower failure risk. Based on these inference results, the server can generate inference knowledge vector parameters of the target transformer node, which may include the operating status, efficiency evaluation, potential risk level, etc. of the transformer.

[0050] Ultimately, these inferred knowledge vector parameters will be integrated into the standard knowledge graph of power grid equipment to optimize the accuracy and completeness of the graph and provide valuable decision support for subsequent power grid management and maintenance.

[0051] In a possible implementation, step S121 may include:

[0052] Step S1211, determine the category of the knowledge vector of the reference power grid equipment standard knowledge graph, the category of the knowledge vector includes at least one of a single knowledge vector, a compound knowledge vector and a hybrid knowledge vector, the hybrid knowledge vector includes at least one of the following: Q single knowledge vectors, W compound knowledge vectors, E single knowledge vectors and R compound knowledge vectors, Q and W are both positive integers greater than 1, and E and R are positive integers.

[0053] In this embodiment, the server first analyzes the category of the knowledge vector in the reference power grid equipment standard knowledge graph. The category of the knowledge vector can be at least one of a single knowledge vector, a composite knowledge vector, or a hybrid knowledge vector. Specifically:

[0054] Single knowledge vector: represents a single feature or attribute of a grid device, such as the rated voltage of a transformer.

[0055] Composite knowledge vector: A comprehensive index that combines multiple features or attributes, such as the total capacity and efficiency of a transformer.

[0056] Hybrid knowledge vector: It is a combination of a single knowledge vector and a compound knowledge vector. It can include multiple single knowledge vectors, multiple compound knowledge vectors, or a mixture of single and compound knowledge vectors.

[0057] In this scenario, the server identifies that the knowledge vector categories in the reference power grid equipment standard knowledge graph are mainly hybrid knowledge vectors, which specifically include two single knowledge vectors (such as the rated voltage and rated current of the transformer) and one compound knowledge vector (such as the operating efficiency of the transformer).

[0058] Step S1212: If the category of the knowledge vector of the reference power grid equipment standard knowledge graph is the single knowledge vector, the single feature inference mechanism is output as the knowledge vector inference sub-mechanism of the target power grid data node.

[0059] Step S1213: If the category of the knowledge vector of the reference power grid equipment standard knowledge graph is the composite knowledge vector, the composite feature inference mechanism is output as the knowledge vector inference sub-mechanism of the target power grid data node.

[0060] Step S1214: If the category of the knowledge vector of the reference power grid equipment standard knowledge graph is the hybrid knowledge vector, then based on the hybrid knowledge vector included in the reference power grid equipment standard knowledge graph, determine the knowledge vector inference sub-mechanism of the target power grid data node from the single feature inference mechanism and the composite feature inference mechanism.

[0061] In this embodiment, if the server determines that the knowledge vector in the reference grid equipment standard knowledge graph is mainly a single knowledge vector, a single feature inference mechanism can be selected as the knowledge vector inference sub-mechanism of the target grid data node. For example, if the reference grid equipment standard knowledge graph mainly focuses on the rated voltage of the transformer, the server will make inferences based on this single feature.

[0062] If the server determines that the knowledge vectors in the graph are mainly composite knowledge vectors, the composite feature inference mechanism can be selected. For example, if the reference grid equipment standard knowledge graph focuses on the comprehensive indicators of the total capacity and efficiency of the transformer, the server can consider these two features comprehensively for inference.

[0063] In this scenario, the server identifies that the category of the knowledge vector is a hybrid knowledge vector, and needs to determine a suitable inference sub-mechanism from a single feature inference mechanism and a composite feature inference mechanism.

[0064] The server can first analyze the single knowledge vectors and composite knowledge vectors contained in the hybrid knowledge vector, and then decide whether to use a single feature inference mechanism, a composite feature inference mechanism, or a combination of the two. In this example, the server may decide to consider the rated voltage, rated current (single knowledge vector), and operating efficiency (composite knowledge vector) of the transformer at the same time, and thus select an inference sub-mechanism that can integrate these features.

[0065] Based on the above analysis, the server finally determined an inference sub-mechanism that integrates single features and composite features. This mechanism can simultaneously consider the rated voltage, rated current and operating efficiency of the transformer to more comprehensively infer the knowledge vector of the target power grid data node. The inference sub-mechanism is then applied to the target power grid data node (such as a specific transformer node) to generate its inferred knowledge vector parameters.

[0066] In a possible implementation, step S1214 may include:

[0067] Step S1214-1, if the reference power grid equipment standard knowledge graph includes Q single knowledge vectors, the single feature inference mechanism is output as a knowledge vector inference sub-mechanism of at least one single knowledge vector among the Q single knowledge vectors of the target power grid data node.

[0068] In this embodiment, when analyzing the reference grid equipment standard knowledge graph, the server finds that the reference grid equipment standard knowledge graph mainly contains Q single knowledge vectors, such as the rated voltage, rated current, and rated power factor of the transformer. In this case, the server can determine to use a single feature inference mechanism to infer the knowledge vector of the target grid data node.

[0069] For example, if the focus of the reference grid equipment standard knowledge graph is on the rated voltage of the transformer, the server can apply a single feature inference mechanism to the target transformer node to specifically infer the knowledge vector parameters of its rated voltage. The server may collect relevant data of the transformer node, such as design specifications, historical operation data, etc., and infer the reasonable range or specific value of its rated voltage through an algorithm model.

[0070] Step S1214-2: If the reference power grid equipment standard knowledge graph includes W kinds of composite knowledge vectors, the composite feature inference mechanism is output as a knowledge vector inference sub-mechanism of at least one of the W kinds of composite knowledge vectors of the target power grid data node.

[0071] In another scenario, if the server finds that the reference grid equipment standard knowledge graph mainly contains W kinds of composite knowledge vectors, such as the comprehensive efficiency index of the transformer (combining the efficiency evaluation of multiple operating parameters), the server can select the composite feature inference mechanism.

[0072] For example, the reference grid equipment standard knowledge graph may include an efficiency index that integrates multiple factors such as transformer load rate, temperature, and loss. The server can use the composite feature inference mechanism to comprehensively consider these factors to infer the comprehensive efficiency index of the target transformer node. This may require the server to perform complex data processing and model calculations to obtain accurate inference results.

[0073] Step S1214-3, if the reference power grid equipment standard knowledge graph includes E single knowledge vectors and R compound knowledge vectors, the single feature inference mechanism is output as a knowledge vector inference sub-mechanism of at least one single knowledge vector among the E single knowledge vectors, and / or the compound feature inference mechanism is output as a knowledge vector inference sub-mechanism of at least one compound knowledge vector among the R compound knowledge vectors.

[0074] In more complex scenarios, the reference grid equipment standard knowledge graph may contain E single knowledge vectors and R compound knowledge vectors. For example, the reference grid equipment standard knowledge graph may contain both single features such as the rated voltage and rated current of the transformer and compound features such as comprehensive efficiency indicators.

[0075] In this case, the server can choose to use a single feature inference mechanism or a composite feature inference mechanism, or a combination of the two, based on actual needs. If the rated voltage of the transformer needs to be inferred, the server can use a single feature inference mechanism; if the comprehensive performance of the transformer needs to be evaluated, the server will use a composite feature inference mechanism.

[0076] Therefore, the server can flexibly select and apply the corresponding inference mechanism (single feature inference mechanism or composite feature inference mechanism) according to the type of knowledge vectors (single, composite or hybrid) contained in the reference grid equipment standard knowledge graph to accurately infer the knowledge vector parameters of the target grid data node.

[0077] In a possible implementation, step S130 may include:

[0078] Step S131, parsing the power grid device data stream to determine the absolute value of the characteristic difference amount of the target power grid data node.

[0079] In this embodiment, the server first receives and parses the power grid equipment data stream, which contains the real-time operation data of each data node in the power grid. For the target power grid data node (for example, a specific transformer), the server can analyze the difference between the real-time data of the target power grid data node and its historical data or standard data. This difference may be reflected in voltage fluctuations, current changes, temperature rise and fall, etc.

[0080] Through comparison and analysis, the server can determine the absolute value of the characteristic difference of the target power grid data node within a specific time period or under specific conditions. The absolute value reflects the degree of deviation between the actual state of the node and the ideal state or historical state.

[0081] Step S132: determining a difference description matrix of the characteristic difference quantity of the target power grid data node according to the knowledge vector inference sub-mechanism of the target power grid data node.

[0082] Next, the server can use the previously determined knowledge vector inference sub-mechanism (which can be a single feature inference mechanism or a composite feature inference mechanism) of the target power grid data node to further analyze these feature differences. The server can construct a difference description matrix that can describe the nature, size, and direction of each feature difference in detail.

[0083] For example, if the voltage at a transformer experiences abnormal fluctuations, the variance description matrix may contain information about the magnitude of the voltage fluctuation, its duration, and the type of fluctuation (e.g., swell, sag, or periodic fluctuation).

[0084] Step S133: generating the characteristic difference amount of the target power grid data node according to the absolute value of the characteristic difference amount and the difference description matrix.

[0085] The server will combine the absolute value of the feature difference and the difference description matrix to generate a comprehensive feature difference report, which includes not only the numerical size of the difference, but also the specific nature of the difference and the possible impact.

[0086] Using a transformer as an example, the report might indicate how many percentage points the voltage fluctuated, whether the fluctuation was sustained or momentary, and what effect the fluctuation might have on the transformer's operating efficiency and lifespan.

[0087] Step S134: generating a reconstructed knowledge parameter of the target power grid data node according to the characteristic difference amount and the inferred knowledge vector parameter of the target power grid data node.

[0088] Finally, the server can use the inferred knowledge vector parameters (these parameters can be inferred based on historical data, standard data or data from other related nodes) combined with the feature difference report to generate the reconstructed knowledge parameters of the target power grid data node.

[0089] These reconstructed knowledge parameters are designed to provide an updated and more accurate description of the node status, which can be used to optimize the operation of the power grid, improve the maintenance efficiency of equipment, or provide valuable diagnostic information when a fault occurs. For example, if the voltage of a transformer continues to fluctuate, the reconstructed knowledge parameters may suggest adjusting the transformer's tap switch position or checking the relevant control system to ensure the stable operation of the power grid.

[0090] In a possible implementation, the single feature inference mechanism includes at least one of the following:

[0091] The reconstructed knowledge parameters of the first knowledge vector of the forward data node are output as the inferred knowledge vector parameters of the first knowledge vector of the target power grid data node, and the predefined description matrix is ​​output as the difference description matrix of the first knowledge vector of the target power grid data node, or the difference description matrix of the first knowledge vector is parsed by the knowledge matrix parsing mechanism, and the first knowledge vector is a single knowledge vector. Alternatively, the predefined parameter sequence is output as the inferred knowledge vector parameters of the first knowledge vector of the target power grid data node, and the predefined description matrix is ​​output as the difference description matrix of the first knowledge vector of the target power grid data node, or the difference description matrix of the first knowledge vector is parsed by the knowledge matrix parsing mechanism.

[0092] In this embodiment, when processing power grid data, the server first identifies the target power grid data node (for example, a specific distribution transformer) and the related forward data node (which may be an upstream node on the power supply line). In this scenario, the server adopts the first method of the single feature inference mechanism.

[0093] The server obtains the reconstructed knowledge parameter of the first knowledge vector from the forward data node. Assuming that the first knowledge vector is a voltage value, the voltage reconstructed knowledge parameter of the forward data node is 240V. The server directly uses this 240V as the voltage inference knowledge vector parameter of the target power grid data node (distribution transformer).

[0094] The server also needs to generate a difference description matrix for the first knowledge vector (voltage) of the target grid data node. In this scenario, the server can directly output a predefined difference description matrix, which can contain information such as voltage fluctuation range and stability. For example, the difference description matrix may indicate that under normal circumstances, the voltage is allowed to fluctuate between 230V and 250V.

[0095] Another possibility is that the server parses the difference description matrix through a knowledge matrix parsing mechanism. For example, the server may dynamically generate a difference description matrix based on historical data, real-time data, and grid operation rules. This difference description matrix will more accurately reflect the actual voltage fluctuation and allowable range of the target grid data node.

[0096] In another scenario, the server may not have the reconstruction knowledge parameters of the forward data nodes directly available, or choose not to use these parameters for some reason. In this case, the server can adopt the second method of a single feature inference mechanism.

[0097] Outputting the predefined parameter sequence as the inferred knowledge vector parameter: The server selects a suitable predefined parameter sequence from the preset parameter library. Taking voltage as an example, the server may select a standard voltage value sequence, such as 220V, 230V, 240V, etc., as the voltage inferred knowledge vector parameter of the target power grid data node. This selection may be based on the design specifications of the power grid, the rated parameters of the equipment, or other relevant factors.

[0098] Similar to scenario 1, the server can directly output a predefined difference description matrix, or dynamically generate a difference description matrix through the knowledge matrix parsing mechanism. The matrix will be used to describe the actual voltage fluctuation and allowable range to help grid operators better understand and monitor the operating status of the target grid data nodes.

[0099] Therefore, when processing power grid data, the server will select different single feature inference mechanisms according to the actual situation to generate the inferred knowledge vector parameters and difference description matrix of the target power grid data node. These parameters and matrices will provide important reference information for the operation and maintenance of the power grid.

[0100] In one possible implementation, the composite feature inference mechanism is to output the reconstructed knowledge parameters of at least one knowledge component element among the H knowledge component elements of the second knowledge vector of the forward data node as inferred knowledge vector parameters of at least one knowledge component element of the second knowledge vector of the target power grid data node, where the second knowledge vector is a composite knowledge vector including H knowledge component elements, and H is a positive integer greater than 1.

[0101] In a possible implementation, the composite feature inference mechanism includes at least one of the following:

[0102] 1. Output the reconstructed knowledge parameters of the H knowledge components of the second knowledge vector of the forward data node as the inferred knowledge vector parameters of the H knowledge components of the second knowledge vector of the target power grid data node, and output the predefined description matrix as the difference description matrix of the H knowledge components of the second knowledge vector of the target power grid data node, or output the predefined description matrix as the difference description matrix of the H1 knowledge components in the second knowledge vector of the target power grid data node, and parse the difference description matrix of the H2 knowledge components of the H knowledge components of the second knowledge vector of the target power grid data node except the H1 knowledge components through the knowledge matrix parsing mechanism, or parse the difference description matrix of the H knowledge components of the second knowledge vector of the target power grid data node through the knowledge matrix parsing mechanism, wherein H1 and H2 are both positive integers, and the sum of H1 and H2 is equal to H.

[0103] In this embodiment, when processing power grid data, the server identifies the target power grid data node (for example, a substation) and the forward data node (which may be the previous substation or power station). The server uses a composite feature inference mechanism to infer the second knowledge vector containing multiple knowledge components (such as voltage, current, power factor, etc.).

[0104] The server directly uses the reconstructed knowledge parameters of all H knowledge components (assuming H=3, including voltage, current, and power factor) of the second knowledge vector of the forward data node as the inferred knowledge vector parameters of the second knowledge vector of the target power grid data node. For example, the voltage of the forward node is 220kV, the current is 1000A, and the power factor is 0.95. These parameters are directly used as the inferred parameters of the target node.

[0105] The server then outputs difference description matrices for these inferred knowledge vector parameters. For voltage, current, and power factor, the server can directly output predefined difference description matrices that may represent the normal fluctuation range of each parameter. Alternatively, the server can use the knowledge matrix parsing mechanism to dynamically generate difference description matrices that are based on historical data, real-time data, and grid operation rules to more accurately reflect the actual fluctuation and allowable range of the parameters.

[0106] 2. Output the reconstructed knowledge parameters of Y knowledge components out of the H knowledge components of the second knowledge vector of the forward data node as inferred knowledge vector parameters of the Y knowledge components of the second knowledge vector of the target power grid data node, output the predefined parameter sequence as inferred knowledge vector parameters of HY knowledge components of the second knowledge vector of the target power grid data node, and output the predefined description matrix as a difference description matrix of the H knowledge components of the second knowledge vector of the target power grid data node, or output the predefined description matrix as a difference description matrix of the Y knowledge components of the second knowledge vector of the target power grid data node, and parse the difference description matrix of the HY knowledge components of the second knowledge vector of the target power grid data node through a knowledge matrix parsing mechanism, or parse the difference description matrix of the H knowledge components of the second knowledge vector of the target power grid data node through a knowledge matrix parsing mechanism, wherein Y is a positive integer less than H.

[0107] In this embodiment, the server only uses the reconstructed knowledge parameters of some knowledge components (such as voltage and current, i.e., Y=2) in the second knowledge vector of the forward data node as the inferred knowledge vector parameters of the target power grid data node. For the remaining knowledge components (such as power factor, i.e., HY=1), the server uses a predefined parameter sequence to fill them. For example, the voltage and current parameters of the forward node are directly used as the inferred parameters of the target node, while the power factor adopts the preset standard value of 0.9.

[0108] The output of the difference description matrix is ​​similar to the above, which can be predefined or dynamically generated through the knowledge matrix parsing mechanism.

[0109] 3. Output the reconstructed knowledge parameters of L knowledge components among the H knowledge components of the second knowledge vector of the forward data node as inferred knowledge vector parameters of the L knowledge components of the second knowledge vector of the target power grid data node, and output them as inferred knowledge vector parameters of HL knowledge components of the second knowledge vector of the target power grid data node using a knowledge vector inference strategy based on X dependent data nodes, and output the predefined description matrix as a difference description matrix of the H knowledge components of the second knowledge vector of the target power grid data node, or output the predefined description matrix as a difference description matrix of the L knowledge components of the second knowledge vector of the target power grid data node, and parse the difference description matrix of the HL knowledge components of the second knowledge vector of the target power grid data node through a knowledge matrix parsing mechanism, or parse the difference description matrix of the H knowledge components of the second knowledge vector of the target power grid data node through a knowledge matrix parsing mechanism, where L is a positive integer less than H.

[0110] In this embodiment, the server uses the reconstructed knowledge parameters of some knowledge components (such as voltage, i.e., L=1) in the second knowledge vector of the forward data node as the inferred knowledge vector parameters of the target power grid data node. For other knowledge components (such as current and power factor, i.e., HL=2), the server derives the inferred knowledge vector parameters based on the knowledge vector inference strategy of other dependent data nodes (such as adjacent substations or load points). For example, the voltage parameter is directly obtained from the forward node, while the current and power factor are inferred through the algorithm model based on the real-time data and historical data of the adjacent nodes.

[0111] The output of the difference description matrix can also be predefined or dynamically generated through the knowledge matrix parsing mechanism to reflect the fluctuation and allowable range of each knowledge component element.

[0112] Therefore, when processing power grid data, the server will select different composite feature inference mechanism implementation methods according to actual conditions to generate inferred knowledge vector parameters and difference description matrices of target power grid data nodes, providing decision support for the operation and maintenance of the power grid.

[0113] In a possible implementation manner, determining whether the target power grid data node is an existing data node includes:

[0114] A. Determine the degree of association between the target grid data node and candidate grid data nodes in a candidate grid data node sequence of the target grid data node, and determine whether the target grid data node is the existing data node based on the degree of association.

[0115] In this embodiment, the server first collects relevant information of the target power grid data node, such as device ID, name, model, location, etc. The server compares the relevant information with the candidate power grid data nodes in the candidate power grid data node sequence. During the comparison process, the server can calculate the correlation between the target power grid data node and the candidate power grid data node. The calculation of the correlation may be based on multiple factors, such as the similarity of device attributes, the connection relationship between devices, etc.

[0116] The server sets a correlation threshold. When the calculated correlation exceeds this threshold, the target power grid data node is considered to be highly correlated with a candidate power grid data node. If the correlation exceeds the set threshold, the server can determine that the target power grid data node is actually an existing data node, but may not be directly matched due to minor differences in data update or entry.

[0117] B. Determine whether the target power grid data node is an existing data node based on the topological position of the target power grid data node and the topological position of the data nodes integrated in the reference power grid equipment standard knowledge graph.

[0118] In this embodiment, the server first determines the location of the target power grid data node in the power grid topology. This is usually based on the connection relationship of the power grid and the physical location information of the equipment. The server compares the topological location of the target power grid data node with the topological location of the integrated data node in the reference power grid equipment standard knowledge graph. If the topological location of the target power grid data node is the same as or similar to the topological location of an integrated data node, the server can determine that the target power grid data node is an existing data node.

[0119] C. Determine whether the target power grid data node is an existing data node based on the classification identifier of the target power grid data node and the classification identifier of the data node integrated in the reference power grid equipment standard knowledge graph, wherein the classification identifier is a coded identifier obtained by converting the topological position of the data node in the knowledge graph during knowledge graph classification.

[0120] In this embodiment, the server generates a classification identifier for the target power grid data node based on the topological position and other relevant information of the target power grid data node. The classification identifier is a coded identifier after converting the topological position of the data node in the knowledge graph classification. The server compares the classification identifier of the target power grid data node with the classification identifier of the integrated data node in the reference power grid equipment standard knowledge graph.

[0121] If the classification identifier of the target power grid data node is the same as the classification identifier of an integrated data node, the server may determine that the target power grid data node is an existing data node.

[0122] In a possible implementation manner, if the reference grid equipment standard knowledge graph includes a hybrid knowledge vector, determining the degree of association between the target grid data node and a candidate grid data node in a candidate grid data node sequence of the target grid data node includes:

[0123] A1. Determine the characteristic correlation between the target power grid data node and the candidate power grid data node based on the topological data of the target power grid data node and the topological data of the candidate power grid data node.

[0124] In this embodiment, the server first collects topological data of the target power grid data node and the candidate power grid data node. The topological data may include information such as the location of the node in the power grid, the connection relationship with other nodes, and the power grid level to which it belongs.

[0125] The server analyzes the collected topological data and compares the relative positions and connection relationships between the target grid data node and the candidate grid data node in the grid topology. For example, the server can check whether the two data nodes are directly connected or whether they are at the same grid level. Based on the analysis results of the topological data, the server calculates the feature correlation between the target grid data node and the candidate grid data node, which reflects the similarity or closeness of the two data nodes in the grid topology. If the two data nodes are very close in topology or have similar connection patterns, their feature correlation will be high.

[0126] A2. Determine a feature correlation degree between the target grid data node and the candidate grid data node based on the knowledge vector of the target grid data node and the knowledge vector of the candidate grid data node.

[0127] In this embodiment, the server extracts respective knowledge vectors from the target power grid data node and the candidate power grid data node. These knowledge vectors may contain information on multiple dimensions such as device type, performance parameters, and operating status.

[0128] The server compares the knowledge vectors of the two data nodes and analyzes their similarities and differences in various dimensions. For example, the server may compare whether the device types of the two data nodes are the same, whether the performance parameters are similar, etc. Based on the comparison results of the knowledge vectors, the server calculates another feature correlation between the target power grid data node and the candidate power grid data node, which reflects the similarity or matching degree of the two data nodes in the attributes represented by the knowledge vectors. If the knowledge vectors of the two data nodes are very similar, their feature correlation will be high.

[0129] A3. Determine the global correlation between the target power grid data node and the candidate power grid data node based on the feature correlation and the knowledge vector.

[0130] What is this? The server integrates the two feature correlations based on topological data and knowledge vectors. This can be achieved through weighted average, product or other suitable combination. The result of the integration is the global correlation, which fully reflects the overall similarity or correlation between the target power grid data node and the candidate power grid data node in terms of topological structure and knowledge attributes. The higher the global correlation, the more similar or related the two data nodes are.

[0131] A4. Determine the degree of association between the target power grid data node and the candidate power grid data node based on the global degree of association.

[0132] The server evaluates the global correlation to determine whether it reaches a preset threshold or meets specific conditions. This can help the server determine whether there is a significant correlation or similarity between two data nodes. If the global correlation meets the requirements, the server uses it as the final correlation between the target power grid data node and the candidate power grid data node, which can be used in subsequent data integration, knowledge reasoning or other power grid data analysis tasks. For example, if the correlation is high enough, the server can treat the two data nodes as the same entity or highly related entities for processing.

[0133] In a possible implementation manner, if the power grid device feature extension mechanism of the target power grid data node is the power grid device feature extension mechanism based on X dependent data nodes, step S120 may include:

[0134] Step S123: determining X dependent data nodes from the candidate grid data node sequence of the target grid data node.

[0135] Step S124 : determining the inferred knowledge vector parameters of the target power grid data node according to the reconstructed knowledge parameters of at least one dependent data node among the X dependent data nodes.

[0136] In this embodiment, the server selects X dependent data nodes from the candidate power grid data node sequence according to a preset rule or algorithm. These dependent data nodes can be nodes that are directly connected to or have a certain specific relationship with the target power grid data node in the power grid topology, or they can be important nodes selected according to other criteria (such as data quality, update frequency, etc.).

[0137] The server further verifies the actual dependency relationship between the X dependent data nodes and the target power grid data node to ensure that the association between them is valid and reliable. This may involve checking historical data, analyzing communication records or data flows between nodes, etc.

[0138] The server collects and reconstructs knowledge parameters from the determined X dependent data nodes. These parameters may include the node's device type, operating status, historical data records, interaction information with other nodes, etc. This information may have been processed and reconstructed into a specific knowledge parameter form in the previous steps.

[0139] The server performs an in-depth analysis of the collected reconstructed knowledge parameters, which may include, for example, comparing parameter differences between different dependent data nodes, identifying correlations or trends between parameters, etc. For example, the server may find that the operating status of certain dependent data nodes is highly correlated with the operating status of the target power grid data node.

[0140] Based on the above analysis, the server begins to infer the knowledge vector parameters of the target power grid data node. The inference process may involve complex computing models or algorithms, such as machine learning models, statistical analysis, etc. The server uses the reconstructed knowledge parameters collected from the dependent data nodes as input and infers the knowledge vector parameters of the target power grid data node through the computing model.

[0141] After inferring the knowledge vector parameters, the server can verify and make necessary corrections. This can include comparing with actual operating data, checking the rationality and consistency of parameters, etc. If it is found that there is a large deviation between the inferred results and the actual data, the server can adjust the calculation model or reselect the dependent data node for inference. After verification and correction, the server generates the final inferred knowledge vector parameters of the target power grid data node. These final inferred knowledge vector parameters will serve as an important basis for subsequent tasks such as power grid data analysis, fault prediction, and optimization decision-making.

[0142] Wherein, step S124 may include:

[0143] Output the trend characteristic parameters of the reconstructed knowledge parameters of the X dependent data nodes as the inferred knowledge vector parameters of the target power grid data node. Alternatively, determine X1 dependent data nodes whose association with the target power grid data node is not less than a set association from the X dependent data nodes, and determine the inferred knowledge vector parameters of the target power grid data node based on the reconstructed knowledge parameters of the X1 dependent data nodes, where X1 is a positive integer not greater than X.

[0144] In this embodiment, the server first performs a trend analysis on the reconstructed knowledge parameters of each dependent data node. This includes identifying patterns of parameter changes over time, such as an upward trend, a downward trend, or a periodic change. Through these analyses, the server extracts trend characteristic parameters of each dependent data node. The server uses these trend characteristic parameters as inferred knowledge vector parameters of the target power grid data node. These inferred knowledge vector parameters reflect the changing trend of the dependent data node knowledge parameters, and therefore can be used as a basis for predicting the future state of the target power grid data node.

[0145] For example, if the server detects that the reconstructed knowledge parameter of a dependent data node shows a stable upward trend, this may mean that the device represented by the dependent data node is experiencing some performance degradation. The server can use this trend characteristic parameter as an inferred knowledge vector parameter of the target power grid data node to warn of possible performance problems.

[0146] The server first calculates the correlation between each dependent data node and the target power grid data node. This can be achieved by comparing the historical data, operating status, topological relationship, etc. of the two. The correlation can be calculated using a variety of algorithms, such as Pearson correlation coefficient, cosine similarity, etc.

[0147] The server sets a correlation threshold and selects X1 nodes from the X dependent data nodes whose correlation with the target power grid data node is not less than the threshold. These nodes are highly correlated with the target power grid data node, so their reconstructed knowledge parameters are of great value for inferring the state of the target node. The server determines the inferred knowledge vector parameters of the target power grid data node based on the reconstructed knowledge parameters of these X1 highly correlated dependent data nodes through weighted averaging, voting or other statistical methods. These parameters integrate the information of other nodes that are highly correlated with the target node, so they can more accurately reflect the actual state or future trend of the target node.

[0148] For example, suppose the server finds through calculation that the correlation between three dependent data nodes and the target grid data node exceeds the set threshold. The reconstructed knowledge parameters of these three nodes indicate the increase in device temperature, power reduction, and increased failure rate. The server takes these parameters into consideration, infers the possible performance problems of the target grid data node, and generates the corresponding inferred knowledge vector parameters for subsequent analysis and processing.

[0149] In a possible implementation manner, determining the inferred knowledge vector parameters of the target power grid data node based on the reconstructed knowledge parameters of the X1 dependent data nodes includes:

[0150] The trend characteristic parameters of the reconstructed knowledge parameters of the X1 dependent data nodes are output as the inferred knowledge vector parameters of the target power grid data node. Alternatively, the importance coefficients of X1 dependent data nodes among the X dependent data nodes are determined, and the fused trend characteristic parameters of the reconstructed knowledge parameters of the X1 dependent data nodes are output as the inferred knowledge vector parameters of the target power grid data node. Alternatively, the importance coefficients of X1 dependent data nodes and the importance coefficients of X-X1 dependent data nodes among the X dependent data nodes are determined, and based on the importance coefficients of each dependent data node among the X dependent data nodes, the fused trend characteristic parameters of the reconstructed knowledge parameters of the X dependent data nodes are output as the inferred knowledge vector parameters of the target power grid data node.

[0151] In this embodiment, the server conducts an in-depth analysis of the reconstructed knowledge parameters of the selected X1 highly correlated dependent data nodes. By using data analysis techniques, such as time series analysis, the server extracts trend characteristics of these parameters, such as rising, falling or stable trends. The server directly uses these trend characteristic parameters as inferred knowledge vector parameters of the target power grid data node. These trend characteristic parameters reflect the overall change trend of the X1 dependent data nodes, and can therefore be used as a reference for predicting the future state of the target node.

[0152] Alternatively, the server assigns an importance coefficient to each dependent data node based on factors such as the correlation between the X1 dependent data nodes and the target power grid data node, data quality, and update frequency. The coefficient reflects the weight of the node when inferring the state of the target node. The server uses the trend feature parameters of the extracted X1 dependent data nodes and combines them with their importance coefficients to perform weighted averaging or other fusion algorithms to obtain the fused trend feature parameters. The server uses the fused trend feature parameters as the inferred knowledge vector parameters of the target power grid data node. This method comprehensively considers the importance and trend characteristics of different dependent data nodes and improves the accuracy of inference.

[0153] Alternatively, the server assigns importance coefficients not only to the X1 highly correlated dependent data nodes, but also to the remaining X-X1 dependent data nodes. These coefficients may be low, but still reflect their certain contribution in the inference process. The server extracts the trend feature parameters of all X dependent data nodes and performs weighted averaging or fusion according to their importance coefficients to obtain more comprehensive fused trend feature parameters. Finally, the server uses the fused trend feature parameters that comprehensively consider the importance and trend characteristics of all dependent data nodes as the inferred knowledge vector parameters of the target power grid data node. This approach utilizes more information and may further improve the accuracy and robustness of the inference.

[0154] Figure 2 The hardware structure of the power grid online service system 100 provided in the embodiment of the present application for implementing the above-mentioned power grid equipment standard knowledge graph generation method is shown. Figure 2 As shown, the grid online service system 100 may include a processor 110 , a machine-readable storage medium 120 , a bus 130 , and a communication unit 140 .

[0155] In an alternative embodiment, the power grid online service system 100 may be a single server or a server group. The server group may be centralized or distributed (for example, the power grid online service system 100 may be a distributed system). In an alternative embodiment, the power grid online service system 100 may be local or remote. For example, the power grid online service system 100 may access information and / or data stored in a machine-readable storage medium 120 via a network. For another example, the power grid online service system 100 may be directly connected to the machine-readable storage medium 120 to access the stored information and / or data. In an alternative embodiment, the power grid online service system 100 may be implemented on a cloud platform. By way of example only, the cloud platform may include a private cloud, a public cloud, an aggregated cloud, a community cloud, a distributed cloud, an internal cloud, a multi-layer cloud, etc. or any integration thereof.

[0156] The machine-readable storage medium 120 may store data and / or instructions. In an alternative embodiment, the machine-readable storage medium 120 may store data obtained from an external terminal. In an alternative embodiment, the machine-readable storage medium 120 may store data and / or instructions used by the power grid online service system 100 to execute or use to complete the exemplary method described in this application. In an alternative embodiment, the machine-readable storage medium 120 may include a large-capacity memory, a removable memory, a volatile read-write memory, a read-only memory, etc. or any integration thereof. Exemplary large-capacity memory may include a magnetic disk, an optical disk, a solid-state disk, etc. Exemplary removable memory may include a flash drive, a floppy disk, an optical disk, a memory card, a compressed disk, a magnetic tape, etc.

[0157] During the specific implementation process, multiple processors 110 execute computer executable instructions stored in the machine-readable storage medium 120, so that the processor 110 can execute the method for generating a standard knowledge graph of power grid equipment in the above method embodiment. The processor 110, the machine-readable storage medium 120 and the communication unit 140 are connected through the bus 130, and the processor 110 can be used to describe the sending and receiving actions of the communication unit 140.

[0158] The specific implementation process of the processor 110 can refer to the various method embodiments executed by the above-mentioned power grid online service system 100. The implementation principles and technical effects are similar, and this embodiment will not be repeated here.

[0159] In addition, an embodiment of the present application also provides a readable storage medium, in which computer executable instructions are pre-installed. When the processor executes the computer executable instructions, the above-mentioned method for generating a standard knowledge graph of power grid equipment is implemented.

[0160] Similarly, it should be noted that in order to simplify the description disclosed in this application, thereby helping to understand one or more embodiments of the invention, in the foregoing description of the embodiments of the present application, multiple features are sometimes combined into one embodiment, drawings, or descriptions thereof. Similarly, it should be noted that in order to simplify the description disclosed in this application, thereby helping to understand one or more embodiments of the invention, in the foregoing description of the embodiments of the present application, multiple features are sometimes combined into one embodiment, drawings, or descriptions thereof.

Claims

1. A method for generating a knowledge graph of power grid equipment standards, characterized in that: The method comprises: When constructing a grid equipment standard knowledge graph, if the target grid data node that needs to be integrated currently is an existing data node in the reference grid equipment standard knowledge graph, then determine the grid equipment feature extension mechanism of the target grid data node, the grid equipment feature extension mechanism includes at least one of a grid equipment feature extension mechanism based on a forward data node and a grid equipment feature extension mechanism based on X dependent data nodes, the forward data node is an integrated data node that is located before the target grid data node in the node integration order of the reference grid equipment standard knowledge graph, and X is a positive integer; Inferring the knowledge vector of the target power grid data node according to the power grid device feature extension mechanism of the target power grid data node, and generating inferred knowledge vector parameters of the target power grid data node; Determining the reconstructed knowledge parameters of the target power grid data node according to the inferred knowledge vector parameters of the target power grid data node, so as to optimize and generate the reference power grid equipment standard knowledge graph; If the grid equipment feature extension mechanism of the target grid data node is the grid equipment feature extension mechanism based on the forward data node, then the knowledge vector of the target grid data node is inferred based on the grid equipment feature extension mechanism of the target grid data node to generate the inferred knowledge vector parameters of the target grid data node, including: Determining a knowledge vector inference sub-mechanism of the target power grid data node, the knowledge vector inference sub-mechanism comprising at least one of a single feature inference mechanism and a composite feature inference mechanism; According to the knowledge vector inference sub-mechanism of the target power grid data node, the knowledge vector of the target power grid data node is inferred to generate an inferred knowledge vector parameter of the target power grid data node; The knowledge vector inference sub-mechanism for determining the target power grid data node includes: Determine the category of the knowledge vector of the reference power grid equipment standard knowledge graph, the category of the knowledge vector includes at least one of a single knowledge vector, a composite knowledge vector and a hybrid knowledge vector, the hybrid knowledge vector includes at least one of the following: Q single knowledge vectors, W composite knowledge vectors, E single knowledge vectors and R composite knowledge vectors, Q and W are both positive integers greater than 1, and E and R are positive integers; If the category of the knowledge vector of the reference power grid equipment standard knowledge graph is the single knowledge vector, outputting the single feature inference mechanism as the knowledge vector inference sub-mechanism of the target power grid data node; If the category of the knowledge vector of the reference power grid equipment standard knowledge graph is the composite knowledge vector, outputting the composite feature inference mechanism as the knowledge vector inference sub-mechanism of the target power grid data node; If the category of the knowledge vector of the reference power grid equipment standard knowledge graph is the hybrid knowledge vector, then according to the hybrid knowledge vector included in the reference power grid equipment standard knowledge graph, the knowledge vector inference sub-mechanism of the target power grid data node is determined from the single feature inference mechanism and the composite feature inference mechanism; Wherein, the determining of the knowledge vector inference sub-mechanism of the target power grid data node from the single feature inference mechanism and the composite feature inference mechanism based on the hybrid knowledge vector included in the reference power grid equipment standard knowledge graph includes: If the reference power grid equipment standard knowledge graph includes Q single knowledge vectors, outputting the single feature inference mechanism as a knowledge vector inference sub-mechanism of at least one single knowledge vector among the Q single knowledge vectors of the target power grid data node; If the reference power grid equipment standard knowledge graph includes W kinds of composite knowledge vectors, the composite feature inference mechanism is output as a knowledge vector inference sub-mechanism of at least one composite knowledge vector among the W kinds of composite knowledge vectors of the target power grid data node; If the reference power grid equipment standard knowledge graph includes E single knowledge vectors and R compound knowledge vectors, the single feature inference mechanism is output as a knowledge vector inference sub-mechanism of at least one single knowledge vector among the E single knowledge vectors, and / or the compound feature inference mechanism is output as a knowledge vector inference sub-mechanism of at least one compound knowledge vector among the R compound knowledge vectors.

2. The method for generating a knowledge graph of power grid equipment standards according to claim 1, characterized in that: The step of determining the reconstructed knowledge parameters of the target power grid data node according to the inferred knowledge vector parameters of the target power grid data node comprises: Parsing the data stream of the power grid equipment to determine the absolute value of the characteristic difference amount of the target power grid data node; Determine a difference description matrix of characteristic difference quantities of the target power grid data node according to a knowledge vector inference sub-mechanism of the target power grid data node; Generate the characteristic difference amount of the target power grid data node according to the absolute value of the characteristic difference amount and the difference description matrix; The reconstructed knowledge parameters of the target power grid data node are generated according to the characteristic difference amount and the inferred knowledge vector parameters of the target power grid data node.

3. The method for generating a knowledge graph of power grid equipment standards according to claim 2, characterized in that: The single feature inference mechanism includes at least one of the following: The reconstructed knowledge parameters of the first knowledge vector of the forward data node are output as inferred knowledge vector parameters of the first knowledge vector of the target power grid data node, and the predefined description matrix is ​​output as the difference description matrix of the first knowledge vector of the target power grid data node, or the difference description matrix of the first knowledge vector is parsed by a knowledge matrix parsing mechanism, and the first knowledge vector is a single knowledge vector; or, a predefined parameter sequence is output as inferred knowledge vector parameters of the first knowledge vector of the target power grid data node, and the predefined description matrix is ​​output as the difference description matrix of the first knowledge vector of the target power grid data node, or the difference description matrix of the first knowledge vector is parsed by a knowledge matrix parsing mechanism.

4. The method for generating a standard knowledge graph of power grid equipment according to any one of claims 2 to 3, characterized in that: The composite feature inference mechanism is to output the reconstructed knowledge parameter of at least one knowledge component element among the H knowledge component elements of the second knowledge vector of the forward data node as the inferred knowledge vector parameter of at least one knowledge component element of the second knowledge vector of the target power grid data node, wherein the second knowledge vector is a composite knowledge vector including H knowledge component elements, and H is a positive integer greater than 1; The composite feature inference mechanism includes at least one of the following: Output the reconstructed knowledge parameters of the H knowledge components of the second knowledge vector of the forward data node as the inferred knowledge vector parameters of the H knowledge components of the second knowledge vector of the target power grid data node, and output the predefined description matrix as the difference description matrix of the H knowledge components of the second knowledge vector of the target power grid data node, or output the predefined description matrix as the difference description matrix of the H1 knowledge components in the second knowledge vector of the target power grid data node, and parse the difference description matrix of the H2 knowledge components among the H knowledge components of the second knowledge vector of the target power grid data node except the H1 knowledge components through the knowledge matrix parsing mechanism, or parse the difference description matrix of the H knowledge components of the second knowledge vector of the target power grid data node through the knowledge matrix parsing mechanism, wherein H1 and H2 are both positive integers, and the sum of H1 and H2 is equal to H; Outputting the reconstructed knowledge parameters of Y knowledge components among the H knowledge components of the second knowledge vector of the forward data node as inferred knowledge vector parameters of the Y knowledge components of the second knowledge vector of the target power grid data node, outputting the predefined parameter sequence as inferred knowledge vector parameters of HY knowledge components of the second knowledge vector of the target power grid data node, and outputting the predefined description matrix as a difference description matrix of the H knowledge components of the second knowledge vector of the target power grid data node, or outputting the predefined description matrix as a difference description matrix of the Y knowledge components of the second knowledge vector of the target power grid data node, and parsing the difference description matrix of the HY knowledge components of the second knowledge vector of the target power grid data node through a knowledge matrix parsing mechanism, or parsing the difference description matrix of the H knowledge components of the second knowledge vector of the target power grid data node through a knowledge matrix parsing mechanism, wherein Y is a positive integer less than H; The reconstructed knowledge parameters of L knowledge components among the H knowledge components of the second knowledge vector of the forward data node are output as inferred knowledge vector parameters of the L knowledge components of the second knowledge vector of the target power grid data node, and the inferred knowledge vector parameters of HL knowledge components of the second knowledge vector of the target power grid data node are output using a knowledge vector inference strategy based on X dependent data nodes, and the predefined description matrix is ​​output as a difference description matrix of the H knowledge components of the second knowledge vector of the target power grid data node, or the predefined description matrix is ​​output as a difference description matrix of the L knowledge components of the second knowledge vector of the target power grid data node, and the difference description matrix of the HL knowledge components of the second knowledge vector of the target power grid data node is parsed by a knowledge matrix parsing mechanism, or the difference description matrix of the H knowledge components of the second knowledge vector of the target power grid data node is parsed by a knowledge matrix parsing mechanism, where L is a positive integer less than H.

5. The method for generating a knowledge graph of power grid equipment standards according to claim 1, characterized in that: Determining whether the target power grid data node is an existing data node includes: Determine the degree of association between the target grid data node and candidate grid data nodes in the candidate grid data node sequence of the target grid data node, and determine whether the target grid data node is the existing data node based on the degree of association; or, Determining whether the target power grid data node is an existing data node based on the topological position of the target power grid data node and the topological position of the data nodes integrated in the reference power grid equipment standard knowledge graph; or, Based on the classification identifier of the target power grid data node and the classification identifier of the integrated data node in the reference power grid equipment standard knowledge graph, determine whether the target power grid data node is an existing data node, wherein the classification identifier is a coded identifier obtained by converting the topological position of the data node in the knowledge graph in the knowledge graph classification.

6. The method for generating a knowledge graph of power grid equipment standards according to claim 5, characterized in that: If the reference grid equipment standard knowledge graph includes a hybrid knowledge vector, determining the degree of association between the target grid data node and a candidate grid data node in a candidate grid data node sequence of the target grid data node includes: Determining a characteristic correlation degree between the target power grid data node and the candidate power grid data node according to the topological data of the target power grid data node and the topological data of the candidate power grid data node; Determining a characteristic correlation degree between the target power grid data node and the candidate power grid data node according to the knowledge vector of the target power grid data node and the knowledge vector of the candidate power grid data node; Determining a global correlation between the target power grid data node and the candidate power grid data node according to the feature correlation and the knowledge vector; According to the global correlation, the correlation between the target power grid data node and the candidate power grid data node is determined.

7. The method for generating a knowledge graph of power grid equipment standards according to claim 1, characterized in that: If the power grid device feature extension mechanism of the target power grid data node is the power grid device feature extension mechanism based on X dependent data nodes, then the power grid device feature extension mechanism based on the target power grid data node is used to infer the knowledge vector of the target power grid data node to generate the inferred knowledge vector parameters of the target power grid data node, including: Determine X dependent data nodes from a sequence of candidate power grid data nodes of the target power grid data node; Determining an inferred knowledge vector parameter of the target power grid data node according to a reconstructed knowledge parameter of at least one dependent data node among the X dependent data nodes; The step of determining the inferred knowledge vector parameter of the target power grid data node according to the reconstructed knowledge parameter of at least one dependent data node among the X dependent data nodes includes: Outputting the trend characteristic parameters of the reconstructed knowledge parameters of the X dependent data nodes as inferred knowledge vector parameters of the target power grid data node; or, Determine X1 dependent data nodes whose association degree with the target power grid data node is not less than a set association degree from the X dependent data nodes, and determine the inferred knowledge vector parameters of the target power grid data node according to the reconstructed knowledge parameters of the X1 dependent data nodes, where X1 is a positive integer not greater than X; The step of determining the inferred knowledge vector parameters of the target power grid data node based on the reconstructed knowledge parameters of the X1 dependent data nodes includes: Output the trend characteristic parameters of the reconstructed knowledge parameters of the X1 dependent data nodes as inferred knowledge vector parameters of the target power grid data node; or, Determine the importance coefficients of X1 dependent data nodes among the X dependent data nodes, and output the fusion trend characteristic parameters of the reconstructed knowledge parameters of the X1 dependent data nodes as the inferred knowledge vector parameters of the target power grid data node; or, Determine the importance coefficients of X1 dependent data nodes and X-X1 dependent data nodes among the X dependent data nodes, and output the fusion trend characteristic parameters of the reconstructed knowledge parameters of the X dependent data nodes as the inferred knowledge vector parameters of the target power grid data node according to the importance coefficient of each dependent data node among the X dependent data nodes.

8. A power grid online service system, characterized in that: The power grid online service system includes a processor and a machine-readable storage medium, wherein the machine-readable storage medium stores machine-executable instructions, and the machine-executable instructions are loaded and executed by the processor to implement the standard knowledge graph of power grid equipment of any one of claims 1-7.

Images