Power system monitoring information table auditing method, equipment and medium

Abstract

The power system monitoring information table auditing method provided by the invention comprises the following steps: receiving input data of a power system monitoring information table, extracting power equipment naming information, and obtaining prior knowledge and associated information of related power equipment based on a pre-constructed power field knowledge graph; encoding the text of the power system monitoring information table based on the natural language processing model to generate a text feature vector; processing the heterogeneous graph of the power system by adopting a graph neural network to generate a graph feature vector; performing multi-modal fusion on the text feature vector and the map feature vector to obtain comprehensive feature representation; constructing a naming auditing core network, and performing bidirectional feature extraction and long sequence dependence modeling on the comprehensive feature representation; and performing normative evaluation on the naming of the power equipment through the multi-dimensional scoring index system to obtain an audit result. Through deep fusion of the knowledge graph and deep learning, organic combination of explicit knowledge and implicit features is realized, and the accuracy and efficiency of auditing are improved.

Description

Technical Field

[0001] The present application relates to the technical field of power system monitoring, and more specifically, to a method, device, and medium for reviewing a power system monitoring information table. Background Art

[0002] With the continuous advancement of smart grid construction, the standardized management of power system monitoring information tables has become increasingly important. Power system monitoring information tables are the basic data for power system operation and maintenance. The standardization of power equipment naming directly affects the reliability of power system operation and the convenience of maintenance.

[0003] However, traditional methods for reviewing power system monitoring information tables have numerous shortcomings: first, they lack a deep understanding of power sector expertise; second, they struggle to handle the complex semantics and structural relationships within device naming; and third, they fail to effectively leverage power system topology information. Traditional power system monitoring information table review relies primarily on manual processes, which are not only inefficient but also prone to omissions and inconsistencies, making them inadequate for the management needs of modern, large-scale power systems. Summary of the Invention

[0004] In response to at least one defect or improvement need in the prior art, the present application provides a method, device and medium for reviewing a power system monitoring information table, aiming to improve the efficiency and accuracy of reviewing a power system monitoring information table.

[0005] To achieve the above objectives, in a first aspect, the present application provides a method for reviewing a power system monitoring information table, comprising:

[0006] Receive input data from the power system monitoring information table, extract power equipment naming information, and obtain prior knowledge and associated information of related power equipment based on the pre-built power domain knowledge graph;

[0007] Encode the text of the power system monitoring information table based on the natural language processing model to generate a text feature vector;

[0008] Use graph neural network to process power system heterogeneous graph and generate graph feature vector;

[0009] Performing multimodal fusion on the text feature vector and the graph feature vector to obtain a comprehensive feature representation;

[0010] Constructing a naming review core network to perform bidirectional feature extraction and long sequence dependency modeling on the comprehensive feature representation;

[0011] The naming of power equipment is evaluated for its standardization through a multi-dimensional scoring index system to obtain the audit results.

[0012] Furthermore, it also includes:

[0013] Based on the professional knowledge of the power field and the naming conventions of power equipment, construct the power field knowledge graph;

[0014] Among them, the power field knowledge graph includes one or more of power equipment type ontology, power equipment related attributes, relationship model between power equipment and power system topology structure information.

[0015] Furthermore, the steps of generating a text feature vector include:

[0016] Encode the text of the power system monitoring information table based on the bidirectional encoder representation model;

[0017] Collect the naming corpus of power equipment and build a special vocabulary for the power field;

[0018] Based on the power sector-specific vocabulary, a masked language model task and a next sentence prediction task are designed to train on the naming rules of power equipment.

[0019] After training, supervised fine-tuning is performed using labeled samples to optimize the model's semantic understanding of terms in the power sector.

[0020] Furthermore, the use of graph neural networks to process heterogeneous graphs of power systems includes:

[0021] Representing the power system as the power system heterogeneous graph, wherein nodes in the power system heterogeneous graph represent power devices, and edges represent physical and / or logical relationships between power devices;

[0022] The R-GCN network is used to model different types of relationships and obtain diverse connection patterns between power devices;

[0023] The GAT network is used to introduce an attention mechanism to dynamically adjust the information aggregation weight according to the importance of the node;

[0024] The node representation is iteratively updated through the multi-layer R-GCN network and GAT network, and the local and global structural information are integrated to form the power equipment node representation including context information.

[0025] Furthermore, the text feature vector and the graph feature vector are multimodally fused to obtain a comprehensive feature representation including:

[0026] Obtain the feature correlation matrix between text features and graph features;

[0027] Designing a gating unit to dynamically adjust the fusion weight according to the feature correlation matrix;

[0028] The text features and graph features are fused through weighted summation or attention mechanism, and the physical location of the power equipment and the system topology are used as auxiliary information to obtain the comprehensive feature representation.

[0029] Furthermore, building a naming audit core network includes:

[0030] Build a Transformer-based encoder-decoder architecture, introduce a multi-head self-attention mechanism in the encoder, and integrate power sector knowledge graph information in the decoder;

[0031] A twin neural network layer is introduced to improve the fitting ability in small sample cases.

[0032] Furthermore, the normative evaluation of the naming of power equipment is carried out through a multi-dimensional scoring index system, including:

[0033] Obtain the semantic compliance score of the power equipment naming and evaluate the semantic consistency between the power equipment naming and the standard terminology through cosine similarity;

[0034] Obtain the structural compliance score of the power equipment naming and evaluate the compliance of the power equipment naming with the regulatory requirements through format matching;

[0035] Obtain the contextual compliance score of the power equipment naming and evaluate the contextual matching of the power equipment naming through the coordination of related sub-items;

[0036] If the score of any dimension is lower than the corresponding preset threshold, the naming of the power equipment is determined to be non-compliant.

[0037] Furthermore, it also includes:

[0038] When it is determined that the naming of the electric equipment is non-compliant, the semantic compliance score, the structural compliance score and the contextual compliance score are retrieved to locate the cause of the non-compliant naming.

[0039] In a second aspect, the present application provides an electronic device comprising at least one processing unit and at least one storage unit, wherein the storage unit stores a computer program, and when the computer program is executed by the processing unit, the processing unit is enabled to perform the steps of the power system monitoring information table audit method described in any one of the aforementioned items.

[0040] In a third aspect, the present application provides a storage medium storing a computer program executable by an access authentication device. When the computer program runs on the access authentication device, the access authentication device is enabled to execute the steps of the power system monitoring information table audit method described in any one of the aforementioned items.

[0041] In general, the above technical solutions conceived by this application can achieve the following beneficial effects compared with the existing technology:

[0042] (1) This application explicitly represents professional knowledge in the power field as a knowledge graph and extracts implicit features through deep learning technology, achieving an organic combination of explicit knowledge and implicit features, thereby improving the accuracy and efficiency of the review.

[0043] (2) This application combines the complementary advantages of graph attention and graph convolution, and simultaneously captures local structural features and global importance relationships through parallel R-GCN networks and GAT networks, thereby improving the accuracy of knowledge extraction.

[0044] (3) This application uses twin neural network technology, which can learn naming conventions from small samples, thereby enhancing the model's adaptability to specific rules of the power system.

[0045] (4) This application introduces a multi-dimensional scoring index system to evaluate naming compliance from three dimensions: semantics, structure, and context, providing a comprehensive audit basis for naming compliance review, thereby improving the accuracy of the review.

[0046] (5) This application can not only provide naming review results, but also locate the reasons for naming non-compliance and provide correction suggestions based on the knowledge graph in the power field, thereby enhancing the interpretability of the model and improving users' understanding and acceptance of the review results. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0048] Figure 1 A core flow chart of a method for reviewing a power system monitoring information table provided in an embodiment of the present application;

[0049] Figure 2 A model architecture diagram corresponding to a method for reviewing a power system monitoring information table provided in an embodiment of the present application;

[0050] Figure 3 A schematic diagram of the knowledge graph architecture in the power field provided in the embodiment of this application;

[0051] Figure 4 A block diagram of an electronic device suitable for implementing the power system monitoring information table audit method described above, provided in an embodiment of the present application. DETAILED DESCRIPTION

[0052] In order to make the purpose, technical solutions and advantages of this application more clearly understood, the present application is further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are merely for the purpose of explaining this application and are not intended to limit this application. In addition, the technical features involved in the various embodiments of this application described below may be combined with each other as long as they do not conflict with each other.

[0053] The terms "first," "second," or "nth" in the specification, claims, or drawings of this application may be used to distinguish different objects or to describe a specific order, depending on the specific scenario. In addition, the terms "including" or "having," and any variations thereof, are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or device that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units that are not listed, or may optionally include other steps or units that are inherent to these processes, methods, products, or devices.

[0054] As described in the background technology section of the specification, the traditional audit method of the power system monitoring information table has many shortcomings: first, there is a lack of in-depth understanding of professional knowledge in the power field; second, it is difficult to handle the complex semantics and structural relationships in the equipment naming; third, it is unable to effectively utilize the topological structure information of the power system. The traditional power system monitoring information table audit mainly relies on manual methods, which is not only inefficient, but also prone to omissions and inconsistencies, and it is difficult to meet the management needs of modern large-scale power systems. In view of this, the present application provides a power system monitoring information table audit method, equipment and medium, the purpose of which is to improve the efficiency and accuracy of the power system monitoring information table audit.

[0055] refer to Figure 1 and Figure 2 An embodiment of the present application provides a method for reviewing a power system monitoring information table. The method for reviewing a power system monitoring information table may specifically include the following steps.

[0056] Step 1: Receive input data from the power system monitoring information table, extract the naming information of power equipment, and obtain prior knowledge and related information of related power equipment based on the pre-built power field knowledge graph.

[0057] In some embodiments, based on the professional knowledge of the power field and the naming conventions of power equipment, a knowledge graph of the power field is pre-built. The constructed knowledge graph of the power field contains the power equipment type ontology, power equipment related attributes, the relationship model between power equipment and the power system topology information. The entities in the knowledge graph of the power field include various types of power equipment, electrical parameters and physical locations, etc. The relationships include "include", "attribute" and "instance". The schematic diagram of the architecture of the knowledge graph of the power field can be referred to Figure 3 .

[0058] For example, equipment entities and relationships are extracted from historical monitoring information tables and industry standards to construct a knowledge graph containing relationships such as "substation-container-transformer." This provides structured knowledge support and enhances the standardization and explainability of audits.

[0059] Step 2: Encode the text of the power system monitoring information table based on a natural language processing model to generate a text feature vector. Specifically, the text of the power system monitoring information table is encoded using the BERT model (a bidirectional encoder representation model, a type of Transformer natural language processing model). Fine-tuning is performed on power domain terminology. Simultaneously, semantic embedding is performed on the constructed power domain knowledge graph to form a text feature vector. In some embodiments, more specifically, the following steps are included.

[0060] Step 21: Collect a large amount of naming corpus of power equipment and construct a specific vocabulary in the power field.

[0061] We can collect a corpus of power equipment nomenclature from various sources, including historical power system monitoring data tables, power equipment manuals, and power industry standards and specifications. We then extract specialized terms, abbreviations, model numbers, and other specific expressions to construct a specialized vocabulary for the power sector. This vocabulary includes substation names and the standard nomenclature for various types of power equipment. Power equipment includes electrical equipment (such as transformers, circuit breakers, and disconnectors) and measuring equipment (such as current transformers and voltage transformers). Each entry in the vocabulary contains the standard name, synonyms, abbreviations, and the corresponding equipment type.

[0062] Step 22: Design a masked language model task and a next sentence prediction task, and train them based on the naming rules of power equipment.

[0063] In this step, two tasks are designed based on a specific vocabulary: masked language modeling (MLM) and next sentence prediction (NSP). In the MLM task, 15% of the tokens in the input sequence are randomly masked, and the model is trained to predict these masked tokens. For example, in the MLM task, the "110 kV" in "CT-A-110 kV" is masked, and the model is trained to predict the masked portion. This enhances the model's ability to capture the semantics of power-related terminology, enabling it to better learn the internal structure and composition rules of power equipment naming. In the NSP task, given two sentences, the model is trained to determine whether the second sentence is a true sequel to the first. This helps the model understand the relevance and contextual dependencies between power equipment names. The process uses the standard BERT architecture, initializes the model using a power-domain-specific vocabulary, and trains it on a power system corpus.

[0064] Step 23: Use the labeled power equipment naming samples for supervised fine-tuning to improve the model's ability to understand power field terminology.

[0065] After training on the naming rules for electrical equipment, supervised fine-tuning was performed using a set of manually labeled examples of electrical equipment names. These examples included both correct and non-standard naming patterns, each with a corresponding standardization label. During fine-tuning, the model learned to map input electrical equipment names to standardization scores and identify non-standard components within the names. Fine-tuning employed a cross-entropy loss function and optimized the model parameters using a gradient descent algorithm. The resulting fine-tuned model accurately understood the semantic content of electrical equipment names and identified standardization issues within them.

[0066] Step 24: semantically embed the constructed knowledge graph in the power field.

[0067] Semantic embedding is performed on the power sector knowledge graph. This involves mapping the entities and relationships in the power sector knowledge graph into a low-dimensional vector space, placing semantically similar concepts closer together in the vector space. Semantic embedding can be performed using algorithms such as TransE and RotatE. The resulting vector representation captures the semantic relationships and / or structural features between entities.

[0068] Step 3: Use graph neural networks to process the heterogeneous graph of the power system and generate graph feature vectors. Graph neural networks include R-GCN networks, GAT networks, etc. In some embodiments, specifically, R-GCN networks (Relational Graph Convolutional Network, a graph convolutional network variant specifically designed for processing heterogeneous graph data, mainly used for link prediction and entity classification tasks in knowledge graphs) and GAT networks (Graph Attention Network, a deep learning model for processing graph data, which dynamically adjusts the importance weights between nodes by introducing an attention mechanism to improve the expressiveness and flexibility of the model) are used to process the knowledge graph in the power field. Through a multi-layer attention mechanism, power equipment node information, connection relationships, and feature attributes are extracted to form a structured representation of power equipment. More specifically, the following steps are included.

[0069] Step 31: Represent the power system as a power system heterogeneous graph, wherein nodes in the power system heterogeneous graph represent power devices, and edges represent physical and / or logical relationships between power devices.

[0070] Step 32: Model different types of relationships through the R-GCN network to obtain diverse connection patterns between power devices.

[0071] The R-GCN network is used to process different types of relationships in the power system heterogeneous graph, and its mathematical expression is:

[0072]

[0073] in, Indicates the Nodes in the layer The hidden state of Represents a set of relationship types; Indicates through relationship Connect to a node The set of neighbor nodes of represents the normalization constant; and represents the learnable weight matrix; represents a non-linear activation function.

[0074] Step 33: Use the GAT network to introduce the attention mechanism and dynamically adjust the information aggregation weight according to the importance of the node. Its mathematical expression is:

[0075]

[0076] in, represents the attention coefficient, represents the node For Node The importance of is calculated as follows:

[0077]

[0078] in, represents a learnable attention vector; Represents a vector concatenation operation.

[0079] Step 34: Through iterative updates of the multi-layer R-GCN network and the GAT network, a power equipment node representation containing rich context information is finally formed.

[0080] This application combines the complementary advantages of graph attention and graph convolution. Through the parallel R-GCN network and GAT network, it simultaneously captures local structural features and global importance relationships, thereby improving the accuracy of knowledge extraction. This application can accurately extract complex relationships between power equipment, thereby improving the representation quality of structured features.

[0081] Step 4: Perform multimodal fusion on the text feature vector and the graph feature vector to obtain a comprehensive feature representation. In some embodiments, a gated fusion mechanism is used to perform multimodal fusion on the text feature vector and the graph feature vector, and the physical location information of the power equipment and the system topology information are incorporated to generate a comprehensive feature representation. More specifically, the following steps are included.

[0082] Step 41: Calculate the feature correlation matrix of text features and graph features , the specific formulas include:

[0083]

[0084] in, Represents the text feature matrix; Represents the graph feature matrix; represents the learnable weight matrix.

[0085] Step 42: Design a gate control unit to dynamically adjust the fusion weight. The specific formula includes:

[0086]

[0087]

[0088] in, 、 、 and represents learnable parameters; Represents the sigmoid activation function.

[0089] Step 43: Fuse the text features and graph features through weighted summation or attention mechanism to obtain the fused feature vector , that is, comprehensive feature representation.

[0090]

[0091] in, Represents dot multiplication.

[0092] Step 44: Introduce the physical location of the power equipment and the system topology as auxiliary information to enhance the context understanding capability of the feature representation.

[0093]

[0094] in, Represents location features; Represents topological features; represents a multilayer perceptron; Represents feature concatenation operation; The final comprehensive feature representation.

[0095] This embodiment adaptively assigns a text feature weight of 0.7 and a graph feature weight of 0.3 through a gating unit, thereby enhancing the contextual understanding capability of feature representation, resolving the multimodal feature conflict problem, and improving the representation capability of comprehensive features.

[0096] Step 5: Build a naming verification core network to perform bidirectional feature extraction and long-sequence dependency modeling on the comprehensive feature representation. In some embodiments, the naming verification core network is built based on the Transformer architecture to implement bidirectional feature extraction and long-sequence dependency modeling on the comprehensive feature representation, and to perform deep semantic understanding and normative analysis on the names of power equipment. More specifically, the following steps are included.

[0097] Step 51: Build a Transformer-based encoder-decoder architecture, introduce a multi-head self-attention mechanism in the encoder, integrate power sector knowledge graph information in the decoder, and build a core network for naming verification. The core network for naming verification uses a Transformer-based encoder-decoder architecture. The encoder introduces a multi-head self-attention mechanism to capture long-range dependencies within names, and the decoder integrates knowledge graph information to enhance understanding of naming standards.

[0098] The encoder contains multiple self-attention layers, and the calculation formula for each layer is:

[0099]

[0100] in, 、 and represent query, key, and value matrices respectively; Indicates the dimension of the key.

[0101] The multi-head attention mechanism is implemented in the following way:

[0102]

[0103] in, ; 、 、 and represents the learnable parameter matrix.

[0104] Step 52: Introduce the twin neural network layer to capture more features from small samples.

[0105] That is, the core naming audit network can also include a twin neural network structure to address sample imbalance. The twin neural network consists of two sub-networks with shared weights, which process the input power equipment name and the standard naming template respectively. The standardization of the naming is evaluated by calculating the data output by the two sub-networks. The loss function of the twin neural network adopts contrastive loss:

[0106]

[0107] in, represents a binary label (0 for match, 1 for no match); represents the Euclidean distance between the outputs of two sub-networks; Represents the boundary parameter.

[0108] This application uses twin neural network technology, which can learn naming conventions from small samples, thereby enhancing the model's adaptability to specific rules of the power system and improving the model's generalization ability for naming rare power equipment.

[0109] Step 6: Conduct a normative assessment of the naming of power equipment through a multi-dimensional scoring index system to obtain the audit results.

[0110] In some embodiments, the multi-dimensional scoring index system includes three dimensions: semantic compliance, structural compliance, and contextual compliance. Semantic compliance evaluates the semantic consistency of naming with standard terminology, and the calculation formula is:

[0111]

[0112] in, and The vector representations of power equipment nomenclature and standard terms are respectively; Represents cosine similarity.

[0113] The structural compliance assessment naming format is consistent with the specification requirements. The calculation formula is:

[0114]

[0115] in, and Respectively represent the named Parts and corresponding standard formats; represents the indicator function; Indicates the number of named parts.

[0116] The contextual compliance assessment name is consistent with the power equipment name and other sub-items such as short message, signal type, interval name, etc. in this monitoring information table. The calculation formula is:

[0117]

[0118] in, Indicates the first sub-items; Indicates the number of sub-items; Indicates the threshold for judging whether the name of the power equipment matches this item. If any one of them fails, For example, if the voltage level named "CT-A-110 kV" does not match the topology, the context score is set to zero.

[0119] The scores from multiple dimensions are combined to generate an overall compliance score:

[0120]

[0121] in, The final compliance judgment output is obtained based on the scores of the three dimensions. If the score of any dimension is lower than the corresponding preset threshold, the power equipment naming is judged to be non-compliant.

[0122] This application introduces a multi-dimensional scoring indicator system to evaluate naming compliance from three dimensions: semantics, structure, and context. It comprehensively covers multiple dimensions of naming standards and provides a comprehensive audit basis for naming compliance review, thereby improving the accuracy of the review and reducing the missed detection rate.

[0123] Step 7: When it is determined that the naming of the power equipment is not compliant, retrieve the semantic compliance score , Structural Compliance Score and contextual compliance scores , locate the reasons for the naming non-compliance and provide correction suggestions based on the power field knowledge graph.

[0124] For example, if the structure score is low, a prompt "The name is missing the voltage level field" will be displayed, and the standard format "CT-A-110 kV" will be recommended. This application not only provides naming review results, but also locates the reasons for non-compliant naming based on the power industry knowledge graph and provides correction suggestions. This enhances the model's interpretability, helps users quickly correct errors, and improves user understanding and acceptance of the review results.

[0125] This application explicitly represents professional knowledge in the power field as a knowledge graph, and extracts implicit features through deep learning technology, realizing the organic combination of explicit knowledge and implicit features, thereby improving the accuracy and efficiency of the review.

[0126] Figure 4 The block diagram of an electronic device suitable for implementing the above-described power system monitoring information table auditing method according to an embodiment of the present application is schematically shown. Figure 4 The electronic device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0127] like Figure 4 As shown, the electronic device 1000 described in this embodiment includes: a processor 1001, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 1002 or a program loaded from a storage part 1008 into a random access memory (RAM) 1003. The processor 1001 may, for example, include a general-purpose microprocessor (such as a CPU), an instruction set processor and / or a related chipset and / or a special-purpose microprocessor (for example, an application-specific integrated circuit (ASIC)), etc. The processor 1001 may also include on-board memory for caching purposes. The processor 1001 may include a single processing unit or multiple processing units for executing different actions of the power system monitoring information table audit method process according to an embodiment of the present application.

[0128] Various programs and data required for the operation of the electronic device 1000 are stored in RAM 1003. The processor 1001, ROM 1002 and RAM 1003 are connected to each other via a bus 1004. The processor 1001 performs various operations of the power system monitoring information table audit method process according to the embodiment of the present application by executing the programs in ROM 1002 and / or RAM 1003. It should be noted that the program can also be stored in one or more memories other than ROM 1002 and RAM 1003. The processor 1001 can also perform various operations of the power system monitoring information table audit method process according to the embodiment of the present application by executing the programs stored in the one or more memories.

[0129] According to an embodiment of the present application, electronic device 1000 may further include an input / output (I / O) interface 1005, which is also connected to bus 1004. Electronic device 1000 may further include one or more of the following components connected to I / O interface 1005: an input portion 1006 including a keyboard, mouse, etc.; an output portion 1007 including devices such as a cathode ray tube (CRT), liquid crystal display (LCD), and speakers; a storage portion 1008 including a hard disk; and a communication portion 1009 including a network interface card such as a LAN card or modem. Communication portion 1009 performs communication processing via a network such as the Internet. A drive 1010 is also connected to I / O interface 1005 as needed. Removable media 1011, such as a magnetic disk, optical disk, magneto-optical disk, semiconductor memory, etc., is installed in drive 1010 as needed, so that computer programs read from the removable media can be installed into storage portion 1008 as needed.

[0130] According to the power system monitoring information table audit method process of the embodiment of the present application, it can be implemented as a computer software program. For example, the embodiment of the present application includes a computer program product, which includes a computer program carried on a computer-readable storage medium, and the computer program contains a program code for executing the power system monitoring information table audit method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 1009, and / or installed from the removable medium 1011. When the computer program is executed by the processor 1001, the above-mentioned functions defined in the system of the embodiment of the present application are executed. According to the embodiment of the present application, the systems, devices, means, modules and / or units described above can be implemented by computer program modules.

[0131] Embodiments of the present application also provide a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments, or may exist independently without being incorporated into the device / apparatus / system. The computer-readable storage medium carries one or more programs, and when the one or more programs are executed, the steps of the power system monitoring information table audit method according to the embodiments of the present application can be implemented.

[0132] According to an embodiment of the present application, the computer-readable storage medium may be a non-volatile computer-readable storage medium, for example, it may include but is not limited to: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In an embodiment of the present application, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, apparatus, or device. For example, according to an embodiment of the present application, the computer-readable storage medium may include one or more memories other than the ROM 1002 and / or RAM 1003 described above.

[0133] It should be noted that the functional modules in the various embodiments of the present application can be integrated into a processing module, or each module can exist physically separately, or two or more modules can be integrated into one module. The above-mentioned integrated modules can be implemented in the form of hardware or in the form of software functional modules. If the integrated modules are implemented in the form of software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, in essence, or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product.

[0134] The flowcharts and / or block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present application. In this regard, each box in the flowchart and / or block diagram can represent a module, a program segment or a part of code, and the part of the above-mentioned module, program segment or code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in an order different from that marked in the accompanying drawings. It should also be noted that each box in the block diagram or flowchart, and the combination of the boxes in the block diagram or flowchart, can be implemented by a special hardware-based system that performs the specified function or operation, or can be implemented by a combination of special hardware and computer instructions.

[0135] Those skilled in the art will appreciate that the features described in the various embodiments and / or claims of this application may be combined and / or coupled in various ways, even if such combinations and / or couplings are not explicitly described in this application. In particular, without departing from the spirit and teachings of this application, the technical features described in the various embodiments and / or claims of this application may be combined and / or coupled in various ways, and all such combinations and / or couplings fall within the scope of this application.

[0136] Although the present application has been shown and described with reference to certain exemplary embodiments thereof, it will be understood by those skilled in the art that various changes in form and details may be made to the present application without departing from the spirit and scope of the present application as defined by the appended claims and their equivalents. Therefore, the scope of the present application should not be limited to the above-described embodiments, but should be determined not only by the appended claims but also by the equivalents of the appended claims.

Claims

1. A method for reviewing a power system monitoring information table, characterized in that: include: Receive input data from the power system monitoring information table, extract power equipment naming information, and obtain prior knowledge and associated information of related power equipment based on the pre-built power domain knowledge graph; Encode the text of the power system monitoring information table based on the natural language processing model to generate a text feature vector; Use graph neural network to process power system heterogeneous graph and generate graph feature vector; Performing multimodal fusion on the text feature vector and the graph feature vector to obtain a comprehensive feature representation; Constructing a naming review core network to perform bidirectional feature extraction and long sequence dependency modeling on the comprehensive feature representation; The naming of power equipment is evaluated for its standardization through a multi-dimensional scoring index system to obtain the audit results.

2. The power system monitoring information table auditing method according to claim 1, characterized in that: Also includes: Based on the professional knowledge of the power field and the naming conventions of power equipment, construct the power field knowledge graph; Among them, the power field knowledge graph includes one or more of power equipment type ontology, power equipment related attributes, relationship model between power equipment and power system topology structure information.

3. The power system monitoring information table auditing method according to claim 1, characterized in that: The steps to generate text feature vectors include: Encode the text of the power system monitoring information table based on the bidirectional encoder representation model; Collect the naming corpus of power equipment and build a special vocabulary for the power field; Based on the power sector-specific vocabulary, a masked language model task and a next sentence prediction task are designed to train on the naming rules of power equipment. After training, supervised fine-tuning is performed using labeled samples to optimize the model's semantic understanding of terms in the power sector.

4. The power system monitoring information table auditing method according to claim 1, characterized in that: Using graph neural networks to process heterogeneous graphs of power systems includes: Representing the power system as the power system heterogeneous graph, wherein nodes in the power system heterogeneous graph represent power devices, and edges represent physical and / or logical relationships between power devices; The R-GCN network is used to model different types of relationships and obtain diverse connection patterns between power devices; The GAT network is used to introduce an attention mechanism to dynamically adjust the information aggregation weight according to the importance of the node; The node representation is iteratively updated through the multi-layer R-GCN network and GAT network, and the local and global structural information are integrated to form the power equipment node representation including context information.

5. The power system monitoring information table auditing method according to claim 1, characterized in that: Performing multimodal fusion on the text feature vector and the graph feature vector to obtain a comprehensive feature representation includes: Obtain the feature correlation matrix between text features and graph features; Designing a gating unit to dynamically adjust the fusion weight according to the feature correlation matrix; The text features and graph features are fused through weighted summation or attention mechanism, and the physical location of the power equipment and the system topology are used as auxiliary information to obtain the comprehensive feature representation.

6. The power system monitoring information table auditing method according to claim 1, characterized in that: Building a naming audit core network includes: Build a Transformer-based encoder-decoder architecture, introduce a multi-head self-attention mechanism in the encoder, and integrate power sector knowledge graph information in the decoder; A twin neural network layer is introduced to improve the fitting ability in small sample cases.

7. The power system monitoring information table auditing method according to claim 1, characterized in that: The standardized evaluation of power equipment naming is carried out through a multi-dimensional scoring index system, including: Obtain the semantic compliance score of the power equipment naming and evaluate the semantic consistency between the power equipment naming and the standard terminology through cosine similarity; Obtain the structural compliance score of the power equipment naming and evaluate the compliance of the power equipment naming with the regulatory requirements through format matching; Obtain the contextual compliance score of the power equipment naming and evaluate the contextual matching of the power equipment naming through the coordination of related sub-items; If the score of any dimension is lower than the corresponding preset threshold, the naming of the power equipment is determined to be non-compliant.

8. The power system monitoring information table auditing method according to claim 7, characterized in that: Also includes: When it is determined that the naming of the electric equipment is non-compliant, the semantic compliance score, the structural compliance score and the contextual compliance score are retrieved to locate the cause of the non-compliant naming.

9. An electronic device, characterized in that: It includes at least one processing unit and at least one storage unit, wherein the storage unit stores a computer program, and when the computer program is executed by the processing unit, the processing unit is able to perform the steps of the power system monitoring information table audit method described in any one of claims 1-8.

10. A storage medium, characterized in that: It stores a computer program that can be executed by an access authentication device. When the computer program runs on the access authentication device, the access authentication device can execute the steps of the power system monitoring information table audit method according to any one of claims 1 to 8.

Images