CN117114657A - Fault information early warning system and method based on power equipment inspection knowledge graph - Google Patents
Fault information early warning system and method based on power equipment inspection knowledge graph Download PDFInfo
- Publication number
- CN117114657A CN117114657A CN202311371006.5A CN202311371006A CN117114657A CN 117114657 A CN117114657 A CN 117114657A CN 202311371006 A CN202311371006 A CN 202311371006A CN 117114657 A CN117114657 A CN 117114657A
- Authority
- CN
- China
- Prior art keywords
- power equipment
- early warning
- information
- knowledge graph
- nodes
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000007689 inspection Methods 0.000 title claims abstract description 70
- 238000000034 method Methods 0.000 title claims abstract description 20
- 238000012512 characterization method Methods 0.000 claims abstract description 41
- 238000013528 artificial neural network Methods 0.000 claims abstract description 28
- 238000010276 construction Methods 0.000 claims abstract description 12
- 230000005540 biological transmission Effects 0.000 claims abstract description 7
- 239000000284 extract Substances 0.000 claims abstract description 4
- 230000007246 mechanism Effects 0.000 claims description 32
- 239000013598 vector Substances 0.000 claims description 24
- 238000013527 convolutional neural network Methods 0.000 claims description 15
- 230000006870 function Effects 0.000 claims description 13
- 238000001914 filtration Methods 0.000 claims description 9
- 238000013507 mapping Methods 0.000 claims description 9
- 230000004913 activation Effects 0.000 claims description 8
- 238000005516 engineering process Methods 0.000 claims description 7
- 230000008878 coupling Effects 0.000 claims description 6
- 238000010168 coupling process Methods 0.000 claims description 6
- 238000005859 coupling reaction Methods 0.000 claims description 6
- 239000011159 matrix material Substances 0.000 claims description 6
- 238000012216 screening Methods 0.000 claims description 6
- 238000009792 diffusion process Methods 0.000 claims description 4
- 230000003993 interaction Effects 0.000 claims description 4
- 230000001902 propagating effect Effects 0.000 claims description 4
- 238000000605 extraction Methods 0.000 claims description 3
- 230000008569 process Effects 0.000 claims description 3
- 230000000644 propagated effect Effects 0.000 abstract description 3
- 230000010365 information processing Effects 0.000 abstract description 2
- 230000009466 transformation Effects 0.000 description 11
- 238000004891 communication Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 238000012549 training Methods 0.000 description 2
- 210000004556 brain Anatomy 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000007480 spreading Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/20—Administration of product repair or maintenance
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/048—Activation functions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/02—Knowledge representation; Symbolic representation
- G06N5/022—Knowledge engineering; Knowledge acquisition
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- General Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- Evolutionary Computation (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Human Resources & Organizations (AREA)
- Economics (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Molecular Biology (AREA)
- Strategic Management (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Water Supply & Treatment (AREA)
- Public Health (AREA)
- Quality & Reliability (AREA)
- Operations Research (AREA)
- Entrepreneurship & Innovation (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The invention belongs to the technical field of power inspection information processing, and particularly relates to a fault information early warning system and method based on an inspection knowledge graph of power equipment. The system extracts relevant corpus of the power equipment from a corpus of the power equipment through a knowledge graph construction module and constructs a patrol knowledge graph of the power equipment; the information transmission module based on the iterative characterization enhanced graph convolution neural network utilizes the constructed power equipment inspection knowledge graph, and combines the iterative characterization enhanced graph convolution neural network to enable fault information of the power equipment to be transmitted among different equipment; and the early warning information identification module carries out fault judgment on the propagated information and carries out early warning. The invention enables the early warning information to be transmitted among different power equipment nodes, and accurate judgment is carried out in fewer power equipment samples.
Description
Technical Field
The invention belongs to the technical field of power inspection information processing, and particularly relates to a fault information early warning system and method based on an inspection knowledge graph of power equipment.
Background
The reliability and stability degree of the transformer side equipment serving as a core pivot point of the power grid directly determine whether the power grid runs reliably or not. And analyzing the power accident data related to the power transformation field, and finding out related accidents to the aspects of misoperation and untimely inspection of related personnel. The power transformation equipment is characterized in that the power transformation equipment is provided with a plurality of power transformation equipment types, the power transformation equipment types are distributed unevenly, the power transformation equipment types are lack of overall information, and equipment faults cannot be foreseen timely, so that the power transformation equipment is low in power transformation equipment type and the power transformation equipment type is heavy in workload of related personnel. It is urgent to improve the efficiency and quality of inspection by accelerating the intelligent and digital transformation of the power grid infrastructure so as to reduce the dependence on inspection personnel, and ensure the safe, reliable and stable operation of the power grid.
One of the important directions of intelligent and digital transformation of the power grid infrastructure is to automatically early warn equipment, and many related researches are carried out for effectively, efficiently and automatically spreading information on a knowledge graph network. In recent years, inspired by the biological brain technology, many researchers train the neural network by combining prior knowledge in the knowledge graph, and obtain better effects. Therefore, combining knowledge in the knowledge graph enables the early warning model to accurately judge in fewer power equipment samples, which is a significant scientific problem.
Disclosure of Invention
The invention aims to provide a fault information early warning system and method based on a power equipment inspection knowledge graph, which can enable early warning information to be transmitted among different power equipment nodes and accurately judge in fewer power equipment samples.
The fault information early warning system based on the power equipment inspection knowledge graph comprises a knowledge graph construction module facing the power equipment, an information transmission module based on an iterative characterization enhanced graph convolution neural network and an early warning information identification module; the knowledge graph construction module extracts relevant corpus of the power equipment from the corpus of the power equipment and constructs a patrol knowledge graph of the power equipment; the information transmission module utilizes the constructed power equipment inspection knowledge graph and combines the iteration characterization enhancement graph convolution neural network so that fault information of the power equipment can be transmitted among different equipment; and the early warning information identification module carries out fault judgment on the propagated information and carries out early warning.
Further preferably, the information transmission module based on the iterative characterization enhanced graph convolution neural network firstly adopts an adjacency matrix to model the power equipment inspection knowledge graph, and the power equipment with connection is represented in the power equipment inspection knowledge graph 1, otherwise, the power equipment inspection knowledge graph is 0; the information propagation module adopts self-attention and cross-attention mechanisms to iteratively update the information of nodes in the graph; dynamically updating the representation of the early warning state of the power equipment by using the iterative representation enhancement graph convolutional neural network; the iterative characterization enhancement map convolutional neural network (Att-GRN) takes word vectors of all early warning information as input, performs context information modeling of all power equipment nodes and relations thereof by using self-attention and cross-attention mechanisms through the iterative characterization enhancement map convolutional neural network, and outputs the feature representation of each power equipment node after enhancement.
Further preferably, the knowledge graph construction module facing the power equipment constructs knowledge in a semi-automatic mode; combining the structural characteristics of the electric power inspection corpus, firstly constructing an entity dictionary comprising equipment, signal lamps, voltage and temperature; designing relationships among entities according to the entity dictionary, wherein the relationships among the entities comprise subordinate relationships and position coupling relationships of the power equipment; secondly, automatically finding new entities from the corpus according to the distance of word vectors among the entities; then, filtering the discovered entities by using an entity disambiguation technology, and manually screening; and finally, taking the power equipment node as a central node, taking the power equipment attribute node as a common node, and constructing a complete power equipment inspection knowledge graph by taking the relation between the power equipment and the power equipment attribute as an edge.
Further preferably, the iterative token enhancement graph convolution neural network includes a self-attention based node token enhancement module and a cross-attention based edge token enhancement module, the features being updated in an iterative manner.
Further preferably, the fault information early warning method based on the power equipment inspection knowledge graph comprises the following steps:
step 1, constructing a power equipment inspection knowledge graph in a semi-automatic mode;
step 2, utilizing iteration characterization enhancement map convolutional neural network (Att-GRN) based on self-attention and cross-attention to transmit early warning information;
modeling the power equipment inspection knowledge graph by adopting an adjacency matrix, and representing the power equipment with connection in the power equipment inspection knowledge graph 1, or else, 0;
step 2.2, adopting a self-attention mechanism to transmit context information among different nodes, thereby dynamically adjusting key information in the nodes, and filtering noise to enhance the characteristic representation of the nodes; the self-attention mechanism dynamically allocates importance weights of different positions in the node; the mathematical expression is as follows:
；
；
wherein X represents the feature representation of the node after the self-attention mechanism update, O represents the feature representation of the node before the update, Q, K, V represents the mapping of the feature representation of each node into a linear network of query vectors, key vectors, value vectors, d k For the dimension represented by the characteristics of the nodes, softmax is the activation function of the self-attention mechanism, and T represents the transpose; FFN denotes a feed forward network with two full connection layers and residual connection; MH represents multi-headed attention; SA represents the self-attention mechanism;
step 2.3, adopting a cross attention mechanism to propagate context information between different nodes and edges associated with the nodes, and increasing interaction between the nodes and the edges so as to enhance characteristic representation of the nodes and the edges; the mathematical expression is as follows:
；
；
；
；
wherein,representing node characteristics after the cross-attention mechanism is updated, Y representing the initial characteristics of the edge, ++>Representing the updated edge feature of the cross-attention mechanism,/-for>Represents a cross-attention module for updating node characteristics,representing a cross attention module for updating edge features,/->Representing a feed forward network with two full connection layers and residual connections for node feature mapping; />Representing a feed forward network with two full connection layers and residual connections for edge feature mapping;
step 2.4, propagating early warning information of the power equipment by using the iterative characterization enhancement map convolutional neural network; the iterative characterization enhancement graph convolution neural network takes word vectors of all early warning information as input, propagates the early warning information to all nodes and edges associated with the nodes through the characteristic of contextual information diffusion of the iterative characterization enhancement graph convolution neural network, and finally obtains the characteristic representation after the relationship between each power equipment node and the power equipment node is enhanced through iterative updating;
and 3, judging the output information of the iterative characterization enhanced graph convolution neural network by utilizing an early warning information identification module, judging which power equipment needs early warning, and outputting an early warning result.
Further preferably, the construction process of the power equipment inspection knowledge graph is as follows:
step 1.1, combining the structural characteristics of an inspection corpus of electric equipment, firstly constructing an entity dictionary comprising equipment, signal lamps, voltage and temperature;
step 1.2, designing relationships among entities according to the entity dictionary, wherein the relationships among the entities comprise subordinate relationships and position coupling relationships of the power equipment;
step 1.3, extracting an alternative entity from the power equipment inspection corpus by using an entity extraction technology;
step 1.4, automatically finding a new entity with the cosine distance of the word vector of the manually selected entity within a set threshold range from the power equipment inspection corpus according to the cosine distance of the word vector between the candidate entity and the manually selected entity;
step 1.5, filtering the discovered new entity by using an entity disambiguation technology, and manually screening;
and step 1.6, the nodes comprise power equipment nodes and power equipment attribute nodes, the power equipment nodes are taken as central nodes, the power equipment attribute nodes are common nodes, and the connection between the power equipment and the power equipment attribute is taken as an edge to construct a complete power equipment inspection knowledge graph.
Further preferably, step 3 normalizes the output of each node using a Sigmoid activation function and trains the iterative representation enhancement graph convolutional neural network using a binary cross entropy loss function to calculate the loss.
The invention provides a nonvolatile computer storage medium, wherein computer executable instructions are stored in the computer storage medium, and the computer executable instructions can execute the fault information early warning method based on the power equipment inspection knowledge graph.
The present invention provides an electronic device including: the system comprises at least one processor and a memory in communication connection with the at least one processor, wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute a fault information early warning method based on a power equipment inspection knowledge graph.
According to the invention, the power equipment inspection knowledge graph is constructed, then the early warning knowledge is transmitted based on semantic information of the power equipment, and the power equipment fault judgment is performed by combining the knowledge graph with the graph convolution. The invention adopts an iterative characterization enhanced graph convolutional neural network (Att-GRN), the network obtains better characterization of each node in the graph by using a self-attention mechanism, then the information interaction of the nodes and the edges is carried out by using a cross-attention mechanism, and early warning information is transmitted to the whole network through graph convolution.
Drawings
Fig. 1 is a block diagram of a fault information early warning system based on a power equipment inspection knowledge graph.
FIG. 2 is a schematic diagram of the reasoning process of the iterative characterization enhancement map convolutional neural network.
Detailed Description
The invention is further elucidated in the following in connection with the accompanying drawings and examples.
Referring to fig. 1, the fault information early warning system based on the power equipment inspection knowledge graph provided by the embodiment comprises a knowledge graph construction module facing the power equipment, an information propagation module based on an iterative characterization enhanced graph convolution neural network and an early warning information identification module; the knowledge graph construction module extracts relevant corpus of the power equipment from the corpus of the power equipment and constructs a patrol knowledge graph of the power equipment; the information transmission module utilizes the constructed power equipment inspection knowledge graph and combines the iteration characterization enhancement graph convolution neural network so that fault information of the power equipment can be transmitted among different equipment; and the early warning information identification module carries out fault judgment on the propagated information and carries out early warning.
In this embodiment, the knowledge graph construction module facing the power equipment constructs knowledge in a semi-automatic manner; combining the structural characteristics of the electric power inspection corpus, firstly constructing an entity dictionary comprising equipment, signal lamps, voltage and temperature; designing relationships among entities according to the entity dictionary, wherein the relationships among the entities comprise subordinate relationships and position coupling relationships of the power equipment; secondly, automatically finding new entities from the corpus according to the distance of word vectors among the entities; then, filtering the discovered entities by using an entity disambiguation technology, and manually screening; and finally, taking the power equipment node as a central node, taking the power equipment attribute node as a common node, and constructing a complete power equipment inspection knowledge graph by taking the relation between the power equipment and the power equipment attribute as an edge.
In this embodiment, the information propagation module of the iterative characterization enhancement graph convolution neural network first uses an adjacency matrix to model a power equipment inspection knowledge graph, and represents the power equipment with connection in a power equipment inspection knowledge graph 1, otherwise, the power equipment inspection knowledge graph is 0; the information propagation module dynamically updates the information of nodes and edges in the graph by adopting a self-attention and cross-attention mechanism; propagating early warning state information of the power equipment by using the iterative characterization enhancement graph convolutional neural network; the iterative characterization enhancement graph convolution neural network takes word vectors of all early warning information as input, propagates information to all power equipment nodes through the knowledge diffusion characteristics of the iterative characterization enhancement graph convolution neural network, and finally outputs the characteristic representation of each power equipment node.
In this embodiment, the early warning information identifying module normalizes the output of each node using a Sigmoid activation function. And calculating the loss by adopting a binary cross entropy loss function and training.
Referring to fig. 2, the fault information early warning method based on the power equipment inspection knowledge graph of the embodiment includes the following steps:
step 1, constructing a power equipment inspection knowledge graph in a semi-automatic mode;
step 1.1, combining the structural characteristics of an inspection corpus of electric equipment, firstly constructing an entity dictionary comprising equipment, signal lamps, voltage and temperature;
step 1.2, designing relationships among entities according to the entity dictionary, wherein the relationships among the entities comprise subordinate relationships and position coupling relationships of the power equipment;
step 1.3, extracting an alternative entity from the power equipment inspection corpus by using an entity extraction technology;
step 1.4, automatically finding a new entity with the cosine distance of the word vector of the manually selected entity within a set threshold range from the power equipment inspection corpus according to the cosine distance of the word vector between the candidate entity and the manually selected entity;
step 1.5, filtering the discovered new entity by using an entity disambiguation technology, and manually screening;
and step 1.6, the nodes comprise power equipment nodes and power equipment attribute nodes, the power equipment nodes are taken as central nodes, the power equipment attribute nodes are common nodes, and the connection between the power equipment and the power equipment attribute is taken as an edge to construct a complete power equipment inspection knowledge graph.
Step 2, utilizing iteration characterization enhancement map convolution neural network (Att-GRN) to transmit early warning information;
modeling the power equipment inspection knowledge graph by adopting an adjacency matrix, and representing the power equipment with connection in the power equipment inspection knowledge graph 1, or else, 0;
step 2.2, adopting a self-attention mechanism to transmit context information among different nodes, thereby dynamically adjusting key information in the nodes, and filtering noise to enhance the characteristic representation of the nodes; the self-attention mechanism dynamically allocates importance weights of different positions in the node; the mathematical expression is as follows:
；
；
wherein X represents the characteristic representation of the node after the self-attention mechanism is updated, O represents the characteristic representation of the node before the update,q, K, V each represents a linear network mapping the feature representation of each node into a query vector, a key vector, a value vector, d k For the dimension represented by the characteristics of the nodes, softmax is the activation function of the self-attention mechanism, and T represents the transpose; FFN denotes a feed forward network with two full connection layers and residual connection; MH represents multi-headed attention; SA represents the self-attention mechanism; multi-headed attention is a stacked multi-layered self-attention mechanism.
Step 2.3, adopting a cross attention mechanism to propagate context information between different nodes and edges associated with the nodes, and increasing interaction between the nodes and the edges so as to enhance characteristic representation of the nodes and the edges, thereby eliminating noise information to a certain extent; the mathematical expression is as follows:
；
；
；
；
wherein,representing node characteristics after the cross-attention mechanism is updated, Y representing the initial characteristics of the edge, ++>Representing the updated edge feature of the cross-attention mechanism,/-for>Represents a cross-attention module for updating node characteristics,representation for further useCross attention module of new edge feature, +.>Representing a feed forward network with two full connection layers and residual connections for node feature mapping; />Representing a feed forward network with two full connection layers and residual connections for edge feature mapping;
step 2.4, propagating early warning information of the power equipment by using the iterative characterization enhancement map convolutional neural network; the iterative characterization enhancement graph convolution neural network takes word vectors of all early warning information as input, propagates the early warning information to all nodes and edges associated with the nodes through the characteristic of contextual information diffusion of the iterative characterization enhancement graph convolution neural network, and finally obtains the characteristic representation after the relationship between each power equipment node and the power equipment node is enhanced through iterative updating;
and 3, judging the output information of the iterative characterization enhanced graph convolution neural network by utilizing an early warning information identification module, judging which power equipment needs early warning, and outputting an early warning result: and normalizing the output of each node by using a Sigmoid activation function, and training the iterative characterization enhancement graph convolutional neural network by adopting a binary cross entropy loss function to calculate loss. And when the test is carried out, the power equipment with the normalized output value exceeding the threshold value of 0.5 is used as the power equipment needing early warning.
The specific mathematical expression of the binary cross entropy Loss function Loss is as follows:
；
where N represents the number of nodes in the graph,for the i-th authentic tag,/->Predicting labels for ith nodeAnd normalizing the probability distribution of the node prediction labels through a Sigmoid activation function.
The accuracy is adopted as an evaluation index of the method, and specifically, the number of samples for judging and early warning is divided by the total number of test samples.
In another embodiment, a non-volatile computer storage medium is provided, where the computer storage medium stores computer executable instructions that can execute the fault information early warning method based on the power equipment inspection knowledge graph in any of the above embodiments.
The present embodiment provides an electronic device including: the system comprises at least one processor and a memory in communication connection with the at least one processor, wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute a fault information early warning method based on a power equipment inspection knowledge graph.
The above-described specific embodiments further illustrate the objects, technical solutions and technical effects of the present invention in detail. It should be understood that the foregoing is only illustrative of the present invention and is not intended to limit the scope of the invention, and that all equivalent changes and modifications that may be made by those skilled in the art without departing from the spirit and principles of the invention shall fall within the scope of the invention.
Claims (8)
1. The fault information early warning system based on the power equipment inspection knowledge graph is characterized by comprising a knowledge graph construction module facing the power equipment, an information propagation module based on an iterative characterization enhanced graph convolution neural network and an early warning information recognition module; the knowledge graph construction module extracts relevant corpus of the power equipment from the corpus of the power equipment and constructs a patrol knowledge graph of the power equipment; the information transmission module utilizes the constructed power equipment inspection knowledge graph and combines the iteration characterization enhancement graph convolution neural network so that fault information of the power equipment can be transmitted among different power equipment nodes; the early warning information identification module carries out fault judgment on the information after propagation and carries out early warning;
the information transmission module based on the iterative characterization enhanced graph convolution neural network firstly adopts an adjacency matrix to model a power equipment inspection knowledge graph, and the power equipment with connection is represented in a power equipment inspection knowledge graph 1, otherwise, the power equipment inspection knowledge graph is 0; the information propagation module adopts self-attention and cross-attention mechanisms to iteratively update the information of nodes and edges in the graph; dynamically updating the representation of the early warning state of the power equipment by using the iterative representation enhancement graph convolutional neural network; the iterative characterization enhancement graph convolution neural network takes word vectors of all early warning information as input, performs context information modeling of all power equipment nodes and relations thereof by using self-attention and cross-attention mechanisms through the iterative characterization enhancement graph convolution neural network, and outputs the feature representation of each power equipment node after enhancement.
2. The fault information early warning system based on the power equipment inspection knowledge graph according to claim 1, wherein the power equipment-oriented knowledge graph construction module constructs knowledge in a semi-automatic mode; combining the structural characteristics of the electric power inspection corpus, firstly constructing an entity dictionary comprising equipment, signal lamps, voltage and temperature; designing relationships among entities according to the entity dictionary, wherein the relationships among the entities comprise subordinate relationships and position coupling relationships of the power equipment; secondly, automatically finding new entities from the corpus according to the distance of word vectors among the entities; then, filtering the discovered entities by using an entity disambiguation technology, and manually screening; and finally, taking the power equipment node as a central node, taking the power equipment attribute node as a common node, and constructing a complete power equipment inspection knowledge graph by taking the relation between the power equipment and the power equipment attribute as an edge.
3. The power equipment inspection knowledge graph-based fault information pre-warning system of claim 1, wherein the iterative characterization enhancement graph convolution neural network comprises a self-attention-based node characterization enhancement module and a cross-attention-based edge characterization enhancement module, wherein features are updated in an iterative manner.
4. The fault information early warning method based on the power equipment inspection knowledge graph is characterized by comprising the following steps of:
step 1, constructing a power equipment inspection knowledge graph in a semi-automatic mode;
step 2, utilizing iteration characterization enhancement map convolutional neural network propagation early warning information based on self-attention and cross-attention;
modeling the power equipment inspection knowledge graph by adopting an adjacency matrix, and representing the power equipment with connection in the power equipment inspection knowledge graph 1, or else, 0;
step 2.2, adopting a self-attention mechanism to transmit context information among different nodes, thereby dynamically adjusting key information in the nodes, and filtering noise to enhance the characteristic representation of the nodes; the self-attention mechanism dynamically allocates importance weights of different positions in the node; the mathematical expression is as follows:
；
；
wherein X represents the feature representation of the node after the self-attention mechanism update, O represents the feature representation of the node before the update, Q, K, V represents the mapping of the feature representation of each node into a linear network of query vectors, key vectors, value vectors, d k For the dimension represented by the characteristics of the nodes, softmax is the activation function of the self-attention mechanism, and T represents the transpose; FFN denotes a feed forward network with two full connection layers and residual connection; MH represents multi-headed attention; SA represents the self-attention mechanism;
step 2.3, adopting a cross attention mechanism to propagate context information between different nodes and edges associated with the nodes, and increasing interaction between the nodes and the edges so as to enhance characteristic representation of the nodes and the edges; the mathematical expression is as follows:
；
；
；
；
wherein,representing node characteristics after the cross-attention mechanism is updated, Y representing the initial characteristics of the edge, ++>Representing the updated edge feature of the cross-attention mechanism,/-for>Representing a cross attention module for updating node characteristics,/->Representing a cross attention module for updating edge features,/->Representing a feed forward network with two full connection layers and residual connections for node feature mapping; />Representation for edgesA feed-forward network with two full connection layers and residual connection for feature mapping;
step 2.4, propagating early warning information of the power equipment by using the iterative characterization enhancement map convolutional neural network; the iterative characterization enhancement graph convolution neural network takes word vectors of all early warning information as input, propagates the early warning information to all nodes and edges associated with the nodes through the characteristic of contextual information diffusion of the iterative characterization enhancement graph convolution neural network, and finally obtains the characteristic representation after the relationship between each power equipment node and each power equipment node is enhanced through iterative updating;
and 3, judging the output information of the iterative characterization enhanced graph convolution neural network by utilizing an early warning information identification module, judging which power equipment needs early warning, and outputting an early warning result.
5. The fault information early warning method based on the power equipment inspection knowledge graph according to claim 4, wherein the construction process of the power equipment inspection knowledge graph is as follows:
step 1.1, combining the structural characteristics of an inspection corpus of electric equipment, firstly constructing an entity dictionary comprising equipment, signal lamps, voltage and temperature;
step 1.2, designing relationships among entities according to the entity dictionary, wherein the relationships among the entities comprise subordinate relationships and position coupling relationships of the power equipment;
step 1.3, extracting an alternative entity from the power equipment inspection corpus by using an entity extraction technology;
step 1.4, automatically finding a new entity with the cosine distance of the word vector of the manually selected entity within a set threshold range from the power equipment inspection corpus according to the cosine distance of the word vector between the candidate entity and the manually selected entity;
step 1.5, filtering the discovered new entity by using an entity disambiguation technology, and manually screening;
and step 1.6, the nodes comprise power equipment nodes and power equipment attribute nodes, the power equipment nodes are taken as central nodes, the power equipment attribute nodes are common nodes, and the connection between the power equipment and the power equipment attribute is taken as an edge to construct a complete power equipment inspection knowledge graph.
6. The fault information early warning method based on the power equipment inspection knowledge graph according to claim 4, wherein the step 3 is characterized in that the output of each node is normalized by using a Sigmoid activation function, and the iterative characterization enhancement graph convolutional neural network is trained by calculating loss by adopting a binary cross entropy loss function.
7. A non-volatile computer storage medium storing computer executable instructions for performing the fault information pre-warning method based on the power equipment inspection knowledge graph of any one of claims 5 to 6.
8. An electronic device, comprising: the system comprises at least one processor and a memory communicatively connected with the at least one processor, wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the fault information early warning method based on the power equipment inspection knowledge graph according to any one of claims 5-6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311371006.5A CN117114657A (en) | 2023-10-23 | 2023-10-23 | Fault information early warning system and method based on power equipment inspection knowledge graph |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311371006.5A CN117114657A (en) | 2023-10-23 | 2023-10-23 | Fault information early warning system and method based on power equipment inspection knowledge graph |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117114657A true CN117114657A (en) | 2023-11-24 |
Family
ID=88800501
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311371006.5A Pending CN117114657A (en) | 2023-10-23 | 2023-10-23 | Fault information early warning system and method based on power equipment inspection knowledge graph |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117114657A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117540175A (en) * | 2024-01-09 | 2024-02-09 | 海纳云物联科技有限公司 | Model training method, prediction method, device, equipment and medium |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114168740A (en) * | 2021-10-11 | 2022-03-11 | 国网天津市电力公司电力科学研究院 | Transformer concurrent fault diagnosis method based on graph convolution neural network and knowledge graph |
| CN114266301A (en) * | 2021-12-16 | 2022-04-01 | 郑州轻工业大学 | Intelligent power equipment fault prediction method based on graph convolution neural network |
| CN114758152A (en) * | 2022-04-25 | 2022-07-15 | 东南大学 | Feature matching method based on attention mechanism and neighborhood consistency |
| CN115171091A (en) * | 2022-09-06 | 2022-10-11 | 浩鲸云计算科技股份有限公司 | Meter identification method for substation inspection |
| CN115409122A (en) * | 2022-09-14 | 2022-11-29 | 中国电力科学研究院有限公司 | Method, system, equipment and medium for analyzing concurrent faults of power transformation equipment |
| CN115659279A (en) * | 2022-11-08 | 2023-01-31 | 国网浙江省电力有限公司杭州市富阳区供电公司 | Multi-mode data fusion method based on image-text interaction |
| CN116108835A (en) * | 2023-01-13 | 2023-05-12 | 大连大学 | Entity alignment method integrating iterative relation graph reasoning and attribute semantic embedding |
| CN116186350A (en) * | 2023-04-23 | 2023-05-30 | 浙江大学 | Power transmission line engineering searching method and device based on knowledge graph and topic text |
| CN116226735A (en) * | 2023-01-16 | 2023-06-06 | 国网吉林省电力有限公司建设分公司 | Substation equipment fault diagnosis method and system based on deep learning and knowledge graph |
-
2023
- 2023-10-23 CN CN202311371006.5A patent/CN117114657A/en active Pending
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114168740A (en) * | 2021-10-11 | 2022-03-11 | 国网天津市电力公司电力科学研究院 | Transformer concurrent fault diagnosis method based on graph convolution neural network and knowledge graph |
| CN114266301A (en) * | 2021-12-16 | 2022-04-01 | 郑州轻工业大学 | Intelligent power equipment fault prediction method based on graph convolution neural network |
| CN114758152A (en) * | 2022-04-25 | 2022-07-15 | 东南大学 | Feature matching method based on attention mechanism and neighborhood consistency |
| CN115171091A (en) * | 2022-09-06 | 2022-10-11 | 浩鲸云计算科技股份有限公司 | Meter identification method for substation inspection |
| CN115409122A (en) * | 2022-09-14 | 2022-11-29 | 中国电力科学研究院有限公司 | Method, system, equipment and medium for analyzing concurrent faults of power transformation equipment |
| CN115659279A (en) * | 2022-11-08 | 2023-01-31 | 国网浙江省电力有限公司杭州市富阳区供电公司 | Multi-mode data fusion method based on image-text interaction |
| CN116108835A (en) * | 2023-01-13 | 2023-05-12 | 大连大学 | Entity alignment method integrating iterative relation graph reasoning and attribute semantic embedding |
| CN116226735A (en) * | 2023-01-16 | 2023-06-06 | 国网吉林省电力有限公司建设分公司 | Substation equipment fault diagnosis method and system based on deep learning and knowledge graph |
| CN116186350A (en) * | 2023-04-23 | 2023-05-30 | 浙江大学 | Power transmission line engineering searching method and device based on knowledge graph and topic text |
Non-Patent Citations (1)
| Title |
|---|
| 杨玉亭;冯林;代磊超;苏菡;: "面向上下文注意力联合学习网络的方面级情感分类模型", 模式识别与人工智能, no. 08 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117540175A (en) * | 2024-01-09 | 2024-02-09 | 海纳云物联科技有限公司 | Model training method, prediction method, device, equipment and medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111581396B (en) | Event graph construction system and method based on multi-dimensional feature fusion and dependency syntax | |
| US20230031738A1 (en) | Taxpayer industry classification method based on label-noise learning | |
| CN112711953A (en) | Text multi-label classification method and system based on attention mechanism and GCN | |
| CN113191148B (en) | Rail transit entity identification method based on semi-supervised learning and clustering | |
| CN109492230B (en) | Method for extracting insurance contract key information based on interested text field convolutional neural network | |
| CN112016313B (en) | Spoken language element recognition method and device and warning analysis system | |
| WO2023241272A1 (en) | Method for automatically generating concrete dam defect image description on basis of graph attention network | |
| CN117114657A (en) | Fault information early warning system and method based on power equipment inspection knowledge graph | |
| CN116089873A (en) | Model training method, data classification and classification method, device, equipment and medium | |
| CN110222192A (en) | Corpus method for building up and device | |
| CN115359799A (en) | Speech recognition method, training method, device, electronic equipment and storage medium | |
| CN113434683A (en) | Text classification method, device, medium and electronic equipment | |
| CN116383399A (en) | Event public opinion risk prediction method and system | |
| CN110826315B (en) | Method for identifying timeliness of short text by using neural network system | |
| CN115757775A (en) | Text implication-based triggerless text event detection method and system | |
| CN115240145A (en) | Method and system for detecting illegal operation behaviors based on scene recognition | |
| CN113886602A (en) | Multi-granularity cognition-based domain knowledge base entity identification method | |
| Mai et al. | Nested relation extraction via self-contrastive learning guided by structure and semantic similarity | |
| CN115952290B (en) | Case characteristic labeling method, device and equipment based on active learning and semi-supervised learning | |
| CN115878815B (en) | Legal document judgment result prediction method, legal document judgment result prediction device and storage medium | |
| CN113284498B (en) | Client intention identification method and device | |
| CN117591657B (en) | Intelligent dialogue management system and method based on AI | |
| CN111639483B (en) | Evaluation aspect determining method and device | |
| CN117725172A (en) | Multi-intention recognition supporting large model QA question-answering method and device | |
| CN116226711A (en) | Case dispute focus identification method and device based on multi-feature fusion |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |