CN114757097B - Line fault diagnosis method and device - Google Patents
Line fault diagnosis method and device Download PDFInfo
- Publication number
- CN114757097B CN114757097B CN202210358432.4A CN202210358432A CN114757097B CN 114757097 B CN114757097 B CN 114757097B CN 202210358432 A CN202210358432 A CN 202210358432A CN 114757097 B CN114757097 B CN 114757097B
- Authority
- CN
- China
- Prior art keywords
- fault
- wave
- recording
- measuring
- feature vector
- 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.)
- Active
Links
- 238000003745 diagnosis Methods 0.000 title claims abstract description 61
- 238000000034 method Methods 0.000 title claims abstract description 35
- 238000004364 calculation method Methods 0.000 claims abstract description 8
- 239000013598 vector Substances 0.000 claims description 164
- 239000011159 matrix material Substances 0.000 claims description 76
- 238000005259 measurement Methods 0.000 claims description 11
- 230000004913 activation Effects 0.000 claims description 9
- 238000010276 construction Methods 0.000 claims description 8
- 238000000354 decomposition reaction Methods 0.000 claims description 8
- 230000001537 neural effect Effects 0.000 claims description 7
- 238000012163 sequencing technique Methods 0.000 claims description 2
- 230000006870 function Effects 0.000 description 22
- 238000004422 calculation algorithm Methods 0.000 description 20
- 238000012545 processing Methods 0.000 description 6
- 238000010586 diagram Methods 0.000 description 5
- 101100377374 Mus musculus Znf568 gene Proteins 0.000 description 4
- 230000008569 process Effects 0.000 description 4
- 238000007792 addition Methods 0.000 description 3
- 230000008901 benefit Effects 0.000 description 3
- 238000004891 communication Methods 0.000 description 3
- 238000004590 computer program Methods 0.000 description 3
- 238000001914 filtration Methods 0.000 description 3
- 238000012549 training Methods 0.000 description 3
- 230000005540 biological transmission Effects 0.000 description 2
- 230000000295 complement effect Effects 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- 230000004075 alteration Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 239000002585 base Substances 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 239000012458 free base Substances 0.000 description 1
- 230000004807 localization Effects 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 230000001052 transient effect Effects 0.000 description 1
- 238000013519 translation Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
- G06F16/367—Ontology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
-
- 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
Abstract
The application discloses a line fault diagnosis method and a device, wherein the line fault diagnosis method comprises the following steps: predicting fault points of the line according to the wave-recording distance-measuring fault data and the knowledge graph; the knowledge graph comprises wave-recording and distance-measuring fault structuring information; sorting the fault points according to the calculation error of each fault point; and determining an actual fault point of the line according to the ordering. The application can solve the problem that the automatic fault diagnosis is difficult to carry out.
Description
Technical Field
The application relates to the technical field of line maintenance, in particular to a line fault diagnosis method and device.
Background
For many years transmission line fault diagnosis and localization has been a concern for power system researchers and power equipment manufacturers. Real-time and accurate fault location plays an important role in rapidly finding fault points, repairing damaged lines and improving the reliability of the system. However, due to the influence of environmental factors and the power system, the fault diagnosis and fault positioning of the power transmission line still need to be further perfected. The accuracy of the current positioning algorithm based on wave recording or protection is low, and because of the huge quantity of fault data, the automatic positioning of fault points has certain difficulty because the requirements of automatic fault diagnosis and accident analysis are difficult to meet.
Disclosure of Invention
The present application has been made to solve the above-mentioned technical problems. The embodiment of the application provides a line fault diagnosis method and device, which can solve the problem that automatic fault diagnosis is difficult to carry out.
According to an aspect of the present application, there is provided a line fault diagnosis method including: predicting fault points of the line according to the wave-recording distance-measuring fault data and the knowledge graph; the knowledge graph comprises wave-recording and distance-measuring fault structuring information; sorting the fault points according to the calculation error of each fault point; and determining an actual fault point of the line according to the ordering.
In one embodiment, the calculated error includes an error compensation value and a confidence level; wherein said sorting said fault points according to the calculated error of each said fault point comprises: calculating the error compensation value and the certainty factor of the fault point; sorting the error compensation values according to the certainty factor; and determining the ordering of the fault points according to the ordering of the error compensation values.
In an embodiment, the wave-recording distance-measuring fault data includes wave-recording distance-measuring fault information and a record sample; wherein said calculating said error compensation value for said fault point and said confidence level comprises: constructing a binary matrix and a multi-value matrix according to the wave-recording distance-measuring fault information set and the recorded sample sequence set; and calculating the error compensation value and the certainty factor of the fault point according to the binary matrix, the multi-value matrix and the knowledge graph.
In an embodiment, the calculating the error compensation value and the certainty factor of the fault point according to the binary matrix, the multi-value matrix and the knowledge graph includes: inputting the single thermal codes of the binary matrix and the multi-value matrix into a neural matrix decomposition model to obtain a binary potential feature vector and a multi-value potential feature vector of the wave-recording ranging fault information, and a binary potential feature vector and a multi-value potential feature vector of the recorded sample sequence; obtaining the potential feature vector of the wave-recording distance-measuring fault information and the potential feature vector of the recorded sample sequence according to the binary potential feature vector and the multi-value potential feature vector of the wave-recording distance-measuring fault information and the binary potential feature vector and the multi-value potential feature vector of the recorded sample sequence; and calculating the error compensation value and the certainty factor of the fault point according to the potential feature vector of the wave-recording distance-measuring fault information and the potential feature vector of the recorded sample sequence.
In an embodiment, the line fault diagnosis method further includes: constructing a target potential vector of the record sample according to the wave-recording distance-measuring fault structural information and the potential feature vector of the record sample sequence; the calculating the error compensation value and the certainty factor of the fault point according to the potential feature vector of the wave-recording distance-measuring fault information and the potential feature vector of the recorded sample sequence includes: and performing inner product operation on the potential feature vector of the wave-recording distance-measuring fault information and the target potential vector of the recorded sample, and calculating the error compensation value and the certainty factor of the fault point.
In an embodiment, the wave-recording ranging fault structuring information comprises a structured embedding vector; the constructing the target potential vector of the record sample according to the wave-recording distance-measuring fault structural information and the potential feature vector of the record sample sequence comprises the following steps:； wherein ,/>A target potential vector representing the recorded sample, < >>A potential feature vector representing the sequence of recorded samples, and (2)>Representing the structured embedding vector.
In an embodiment, the method comprises performing an inner product operation on the latent feature vector of the recorded ranging fault information and the target latent vector of the recorded sample, and calculating the error compensation value and the error compensation value of the fault pointThe certainty factor includes:, wherein ,/>A potential feature vector representing said wave-recording distance-measuring fault information,>a target potential vector representing the recorded sample, < >>Representing a vector dot product, u represents the wave-recording distance-measuring fault information, and v represents the recorded sample; />, wherein ,/>Representing said error compensation value and said certainty +.>Representing an activation function->Representing a vector dot product.
In an embodiment, the constructing the binary matrix and the multi-valued matrix according to the set of the wave-recording ranging fault information and the set of the recorded sample sequence includes: ； wherein ,/>Representing the binary matrix,/->Representing whether the wave-recording distance-measuring fault information appears in the recorded sample sequence set, wherein m represents the total number of the wave-recording distance-measuring fault information, and n represents the total number of the recorded samples; />； wherein ,/>Representing the multi-valued matrix->And whether recorded data representing the wave-recording distance-measuring fault information is known or not, wherein m represents the total number of the wave-recording distance-measuring fault information, and n represents the total number of the recorded samples.
In one embodiment, the wave-recording distance-measuring fault information comprises 0-1 discrete data, multi-value discrete data and continuous data; the construction of the binary matrix and the multi-value matrix according to the wave-recording distance-measuring fault information set and the recorded sample sequence set comprises the following steps: and constructing the binary matrix and the multi-value matrix according to the 0-1 discrete data, the multi-value discrete data, the continuous data and the record sample sequence set.
According to another aspect of the present application, there is provided a line fault diagnosis apparatus comprising: the prediction module is used for predicting fault points of the line according to the wave-recording distance-measuring fault data and the knowledge graph; the knowledge graph comprises wave-recording and distance-measuring fault structuring information; the sorting module is used for sorting the fault points according to the calculation errors of each fault point; and the determining module is used for determining the actual fault point of the line according to the sequence. According to the line fault diagnosis method and device provided by the application, the actual fault point error compensation value and the certainty factor thereof are predicted according to the unstructured information of the existing wave-recording distance-measuring fault data and the structured power grid line knowledge information in the knowledge graph, and the ranking list of the predicted compensation value is given according to the certainty factor, so that the intelligent fault diagnosis of the power grid line is realized, and the problems that in the prior art, the accuracy of a positioning algorithm is lower and automatic fault diagnosis is difficult to carry out due to huge quantity of fault data can be solved.
Drawings
The above and other objects, features and advantages of the present application will become more apparent by describing embodiments of the present application in more detail with reference to the attached drawings. The accompanying drawings are included to provide a further understanding of embodiments of the application and are incorporated in and constitute a part of this specification, illustrate the application and together with the embodiments of the application, and not constitute a limitation to the application. In the drawings, like reference numerals generally refer to like parts or steps.
Fig. 1 is a schematic diagram of a circuit fault diagnosis model according to an exemplary embodiment of the present application.
Fig. 2 is a flow chart of a line fault diagnosis method according to an exemplary embodiment of the present application.
Fig. 3 is a flow chart of a line fault diagnosis method according to another exemplary embodiment of the present application.
Fig. 4 is a flow chart of a line fault diagnosis method according to another exemplary embodiment of the present application.
Fig. 5 is a schematic structural view of a line fault diagnosis apparatus according to an exemplary embodiment of the present application.
Fig. 6 is a schematic structural view of a line fault diagnosis apparatus according to another exemplary embodiment of the present application.
Fig. 7 is a block diagram of an electronic device according to an exemplary embodiment of the present application.
Detailed Description
Hereinafter, exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. It should be apparent that the described embodiments are only some embodiments of the present application and not all embodiments of the present application, and it should be understood that the present application is not limited by the example embodiments described herein.
When a power grid breaks down or is greatly disturbed, the centralized fault wave recording device and the protection device with the wave recording function (hereinafter referred to as the fault wave recording device) can accurately record wave recording data of a plurality of cycles before and after the fault. Compared to Event (Event) information, fault recording data has many unique advantages, such as: (1) high reliability. The sampling frequency of the fault recorder can generally reach 4kHz or more, and the fault recorder contains more abundant power grid transient information, so that the running condition of the power grid before and after faults can be truly reflected; and (2) good fault tolerance. The fault wave recording device can continuously record waves in a period of time, so that the continuity of wave recording data can be fully utilized to realize fault tolerance; (3) including timing information. The fault recording data can reflect the time section information of the local power system in the longitudinal direction, and also record the time sequence information of continuous change of analog quantity and switching value in the transverse direction, and the time sequence information can reveal more fault information. With the advantages as described above, the fault-recording device is also referred to as a "black box" of the power system. With the continuous development of computer and communication technologies, fault wave recording devices are increasingly commonly applied to power grids. When a power grid fails or is disturbed, the nature of the fault needs to be determined as soon as possible and the fault location needs to be judged so as to restore the power supply as soon as possible. To achieve these objectives, it is necessary to automatically analyze a large amount of fault recording data generated at the time of a fault by a computer system.
Therefore, the line fault diagnosis method provided by the application can be applied to a line fault diagnosis model, can automatically analyze fault recording data and can perform intelligent fault diagnosis. The model can predict the error compensation value of the actual fault point and the certainty factor thereof according to unstructured information of the existing power grid line wave-recording distance-measuring fault data set and by combining with structured power grid line knowledge information in a knowledge graph, and give a ranking list of the predicted compensation value according to the certainty factor, so as to realize intelligent fault diagnosis of the power grid line.
FIG. 1 is a schematic diagram of a circuit fault diagnosis model according to an exemplary embodiment of the present application, as shown in FIG. 1, in which the recording distance fault data may include recording distance fault information and a recording sample, and the recording distance fault information is given as a recording distance fault information setAnd record sample sequence set +.>M and n represent the recording ranging fault information and the total number of recorded samples, respectively. Wherein, the wave-recording distance-measuring fault information can be divided into two types, one type is 0-1 discrete typeData such as weather or rain in weather environment information; the other type is multi-value discrete data and continuous data, such as temperature, wind power, current before failure, wave recording and distance measuring information and the like. Constructing a binary matrix according to the wave-recording distance-measuring fault data of the power grid line , wherein />Meteorological environment information in record-wave distance-measuring fault information set>The +.f in the record sample sequence number set>Appear in the bar record, vice versa>. Similarly, a multi-valued matrix is constructed>, wherein Recorded data representing recorded distance measurement fault information, whereas +.>. By a binary matrix->And multivalue matrix->The one-hot code (independent thermal code) is used as the input of a neural matrix decomposition model, and Embedding adopts normal random initialization to obtain a binary potential feature vector of the wave-recording distance-measuring fault information>And multivalued latent feature vector sum->And recording the binary latent feature vector of the sample sequence +.>And multivalued latent feature vector sum +.>. The addition of the binary and multi-valued potential eigenvectors is complementary to obtain the potential eigenvector representation of the wave-recording ranging fault information: />Recording a potential eigenvector representation of the sample sequence: />。
Defining a knowledge graph containing structural information of wave-recording distance-measuring faultsWherein each triplet represents a head entity +.>And tail entity->There is a relationship->For example (110 kV, voltage class, chato line) triplets describe the objective information that 110kV is the voltage class of Chato line,/A- > and />Respectively representing a set of different entities and a set of different connection relations between the entities, +.>The collection includes line voltage class, line length, line model, etc.
In the structured knowledge feature extraction stage, a power grid line wave-recording distance-measuring fault data set and a Freebase knowledge graph are taken as input, a power grid line related entity construction sub-knowledge graph is extracted, and a knowledge graph embedding method (Xavier-TransR) is used for searching a potential vector representation of structured knowledge from a sub-graph, so that a power grid line knowledge representation with knowledge perception is obtained. Then, introducing structural knowledge features into a collaborative filtering algorithm to enrich semantic information of a record sample sequence, and finally obtaining target potential vector representation of the record sample fused with the structural knowledge：/>That is, the final recorded sample sequence represents the potential eigenvector of the fault information to be recorded with the distance measurement>And finally record sample sequence representation +.>Performing an inner product operation, wherein the formula is defined as: />，/>The vector dot product is represented, u represents the wave-recording distance-measuring fault information, v represents the recorded sample, and the final prediction formula is as follows: />，/>Representing error compensation value and certainty, +.>Representing activation functionsCount (n)/(l)>Representing a vector dot product, wherein +_ >Using sigmoid function as activation function of output layer, the output predicted value is in [0-1 ]]The linear characteristic and the nonlinear characteristic can be well combined.
The model takes matrix decomposition thought as a basis, wave-recording and distance-measuring fault information and finally records potential feature vectors of sample sequences as input, acquires shallow features of the wave-recording and distance-measuring fault information by dot product operation, and then gives a binary matrixMultiple value matrix->And knowledge pattern->And predicting an actual fault point error compensation value of the power grid line and the certainty factor thereof, and giving a ranking list of the predicted compensation value according to the certainty factor to realize intelligent fault diagnosis of the power grid line.
Fig. 2 is a flow chart of a line fault diagnosis method according to an exemplary embodiment of the present application, and as shown in fig. 2, the line fault diagnosis method includes:
step 100: and predicting fault points of the line according to the wave-recording distance-measuring fault data and the knowledge graph.
The knowledge graph comprises wave-recording and distance-measuring fault structuring information.
The wave-recording distance-measuring fault data and the knowledge graph can be input in advance, for example, the wave-recording distance-measuring fault data can comprise wave-recording distance-measuring fault information and recorded samples, and the wave-recording distance-measuring fault information set is given And record sample sequence set +.>M and n represent the recording ranging fault information and the total number of recorded samples, respectively. The knowledge graph can be obtained from the existing knowledge graph, and the knowledge graph is a structured semantic knowledge base and is used for rapidly describing concepts and interrelationships of the concepts in the physical world. The knowledge graph is converted into simple and clear triples of entities, relations and entities by effectively processing, processing and integrating the data of the complicated documents, and finally a large amount of knowledge is aggregated, so that the quick response and reasoning of the knowledge are realized. For example, a knowledge graph containing structured information of wave-recording distance-measuring faults can be definedWherein each triplet represents a head entity +.>And tail entity->There is a relationship->. For example (110 kV, voltage class, chato line) triplets describe the objective information that 110kV is the voltage class of Chato line,/A-> and />Respectively representing a set of different entities and a set of different connection relations between the entities, +.>The collection includes line voltage class, line length, line model, etc.
The wave-recording distance-measuring fault data contains unstructured information such as pictures, audios, videos and texts, and the knowledge graph contains structured power grid line knowledge information such as a relational database. Therefore, the wave-recording distance-measuring fault data and the knowledge graph need to be processed so as to predict the fault point of the line.
Step 200: and sequencing the fault points according to the calculation error of each fault point.
The wave-recording distance-measuring fault data and the knowledge graph can be processed in the model and then used as input to obtain potential feature vectors of wave-recording distance-measuring fault information and potential vectors of a recorded sample sequence, and then the potential feature vectors are subjected to inner product so as to predict an actual fault point error compensation value and the certainty factor thereof, and a ranking list of the predicted compensation value is given according to the certainty factor.
Step 300: and determining the actual fault point of the line according to the sequence.
In the generation stage of the ranking list of the predicted compensation values, a Top-N compensation list of the error predicted compensation values of the actual fault points is generated by using the obtained wave-recording and distance-measuring fault information and the inner product represented by the potential vector of the wave-recording and distance-measuring fault record sample, and comprehensive selection is carried out according to the absolute value of the error predicted compensation values of the actual fault points in the list and the certainty factor of the absolute value of the error predicted compensation values of the actual fault points, so that the actual fault points are selected from the predicted fault points, and intelligent fault diagnosis of the power grid line is realized.
According to the line fault diagnosis method provided by the application, the actual fault point error compensation value and the certainty factor thereof are predicted according to the unstructured information of the existing wave-recording distance-measuring fault data and the structured power grid line knowledge information in the knowledge graph, and the ranking list of the predicted compensation value is given according to the certainty factor, so that the intelligent fault diagnosis of the power grid line is realized, and the problems that in the prior art, the accuracy of a positioning algorithm is low and the automatic fault diagnosis is difficult to carry out due to the huge quantity of fault data can be solved.
FIG. 3 is a flow chart of a line fault diagnosis method according to another exemplary embodiment of the present application, wherein the calculated error includes an error compensation value and a certainty factor as shown in FIG. 3; the step 200 may include:
step 210: and calculating an error compensation value and certainty of the fault point.
The error compensation is to artificially create a new original error to counteract the original error which is a problem at present, and the two errors should be equal in size and opposite in direction as much as possible, so that the purposes of reducing the error and improving the precision are achieved. Therefore, by calculating the error compensation value and the certainty of the predicted fault point, the accuracy of the currently predicted fault point can be judged, and the final actual fault point can be determined according to the accuracy of the predicted fault point.
Step 220: the error compensation values are ordered according to certainty.
In the generation stage of the ranking list of the predicted compensation values, a Top-N compensation list of the error predicted compensation values of the actual fault points is generated by using the obtained wave-recording and distance-measuring fault information and the inner product represented by the potential vector of the wave-recording and distance-measuring fault record sample, and comprehensive selection is carried out according to the absolute value of the error predicted compensation values of the actual fault points in the list and the certainty factor of the absolute value of the error predicted compensation values of the actual fault points, so that the actual fault points are selected from the predicted fault points, and intelligent fault diagnosis of the power grid line is realized.
Step 230: and determining the sequence of the fault points according to the sequence of the error compensation values.
And comprehensively selecting according to the absolute value of the error prediction compensation value of the actual fault point in the list and the certainty factor of the absolute value, selecting the fault point with higher precision from the predicted fault points as the actual fault point, and realizing intelligent fault diagnosis of the power grid line.
Fig. 4 is a flowchart of a line fault diagnosis method according to another exemplary embodiment of the present application, where, as shown in fig. 4, the wave-recording ranging fault data may include wave-recording ranging fault information and a record sample; the step 210 may include:
step 211: and constructing a binary matrix and a multi-value matrix according to the wave-recording distance-measuring fault information set and the recorded sample sequence set.
Given wave-recording distance-measuring fault information setAnd recording a set of sample sequencesM and n represent the total number of recording distance measurement fault information and recording samples, respectively, wherein the recording distance measurement fault information can be divided into two types,the data is 0-1 discrete data, such as weather and rain weather in weather environment information; the other type is multi-value discrete data and continuous data, such as temperature, wind power, current before failure, wave recording and distance measuring information and the like. According to the wave-recording distance-measuring fault data of the power grid line, constructing a binary matrix +. >, wherein />Meteorological environment information in record-wave distance-measuring fault information set>The +.f in the record sample sequence number set>Appear in the bar record, vice versa>. Similarly, a multi-valued matrix is constructed>, wherein />Recorded data representing recorded distance measurement fault information, whereas +.>。
Step 212: and calculating error compensation values and certainty of the fault points according to the binary matrix, the multi-value matrix and the knowledge graph.
By giving a binary matrixMultiple value matrix->And knowledge graph, predicting the error compensation value of the actual fault point of the power grid line and the certainty factor thereof, and giving out the pre-prediction according to the certainty factorAnd measuring a ranking list of the compensation values to realize intelligent fault diagnosis of the power grid line.
In one embodiment, the step 212 may include:
inputting the single thermal codes of the binary matrix and the multi-value matrix into a neural matrix decomposition model to obtain the binary potential feature vector and the multi-value potential feature vector of the wave-recording and distance-measuring fault information, and recording the binary potential feature vector and the multi-value potential feature vector of the sample sequence.
In a binary matrixAnd multivalue matrix->The one-hot code (independent thermal code) is used as the input of a neural matrix decomposition model, and Embedding adopts normal random initialization to obtain a binary potential feature vector of the wave-recording distance-measuring fault information >And multivalued latent feature vector sum->And recording the binary latent feature vector of the sample sequence +.>And multivalued latent feature vector sum +.>。
And obtaining the potential feature vector of the wave-recording distance-measuring fault information and the potential feature vector of the recorded sample sequence according to the binary potential feature vector and the multi-value potential feature vector of the wave-recording distance-measuring fault information and the binary potential feature vector and the multi-value potential feature vector of the recorded sample sequence.
The addition of the binary and multi-valued potential eigenvectors is complementary to obtain the potential eigenvector representation of the wave-recording ranging fault information:recording a potential eigenvector representation of the sample sequence: />。
And calculating error compensation values and certainty of the fault points according to the potential feature vectors of the wave-recording distance-measuring fault information and the potential feature vectors of the recorded sample sequences.
And carrying out inner product operation on the potential feature vector of the wave-recording distance-measuring fault information and the potential vector of the recorded sample sequence, so as to predict an actual fault point error compensation value and the certainty factor thereof.
In an embodiment, the line fault diagnosis method may further include:
and constructing a target potential vector of the record sample according to the wave-recording distance-measuring fault structural information and the potential feature vector of the record sample sequence.
In order to construct a final potential vector representation of the recording sample of the wave-recording distance measurement fault, namely, a target potential vector of the recording sample sequence, a structured embedded vector and an unstructured wave-recording distance measurement fault information feature vector in a knowledge graph can be integrated together, a combined learning framework is provided, and an embedded vector representation of the intelligent fault diagnosis model of the power grid line is obtained simultaneously through a combined learning knowledge graph embedding and a loss function of a collaborative filtering algorithm. That is, in the collaborative filtering algorithm, the structural knowledge features are introduced to enrich the semantic information of the record sample sequence, and finally the target potential vector representation of the record sample after the structural knowledge is fused is obtained：/>, wherein ,/>Target potential vector representing recorded sample, +.>Potential eigenvectors representing recorded sample sequences, +.>Representing the structured embedding vector.
The calculating the error compensation value and the certainty factor of the fault point according to the potential feature vector of the wave-recording ranging fault information and the potential feature vector of the recorded sample sequence may include:
and carrying out inner product operation on the potential feature vector of the wave-recording ranging fault information and the target potential vector of the recorded sample, and calculating an error compensation value and certainty factor of the fault point.
Performing inner product operation on the potential feature vector of the wave-recording ranging fault information and the potential vector of the recorded sample sequence, so as to predict an actual fault point error compensation value and the certainty factor, wherein the formula is defined as follows:，/>potential eigenvectors representing wave-recording ranging fault information, < >>Target potential vector representing recorded sample, +.>Representing a vector dot product, u representing the recording ranging fault information, v representing the recording sample, and a prediction formula is defined as follows: />，Representing error compensation value and certainty, +.>Representing an activation function->Representing a vector dot productWherein->Using sigmoid function as activation function of output layer, the output predicted value is in [0-1 ]]The linear characteristic and the nonlinear characteristic can be well combined.
In one embodiment, the wave-recording ranging fault-structured information includes a structured embedded vector; the constructing a target potential vector of the record sample according to the wave-recording ranging fault structural information and the potential feature vector of the record sample sequence may include:； wherein ,/>Target potential vector representing recorded sample, +.>Potential eigenvectors representing recorded sample sequences, +.>Representing the structured embedding vector.
For extraction of structural features, the TransR algorithm can map entities and relationships to different semantic spaces based on a relationship-specific matrix, as compared with embedding the entities and relationships into the same vector space Is more reasonable. First, use the knowledge triad +.>Describing a fact, followed by +.A.A. based on the basic principle of the translation algorithm>Optimizing by an optimizing method, and learning the embedded vector of each entity and relation. Use-> and />Representation-> and />Is a vector of embedding; />Representation-> and />In relation space->Is included in the projection vector. The knowledge triples describing the factsThe credibility scoring formula is defined as follows:
wherein ,for relation->Describing a projective transformation of the relationship, projecting the entity from the entity vector space in d-dimension to the relationship vector space in k-dimension. Its function value->The lower the score value, the more realistic the triplet defines the objective fact.
According to the prior study, when the variance of each layer output of the network model is equal as much as possibleWhen the information in the network has better mobility. Thus, the Xavier-based TransR algorithm is used in the present application. When initializing embedded vector in algorithm, parameters are made and />Obeys the following normal distribution
wherein , and />Embedding the model into the layer->Layer input and->Number of features of the layer output.
Therefore, in the training process of the model applied by the application, the algorithm comprehensively considers the relative sequence between the positive sample triplet and the negative sample triplet, and in order to distinguish the two, the distinguishing degree is defined based on the loss function of the BPR algorithm:
wherein ,for entity-relationship quadruple set, +.>To construct a negative sample triplet, one entity in the positive sample triplet may be randomly replaced to constructManufacturing; />Is a sigmoid function. The algorithm may tend to 0 in the training process, in order to avoid such a situation, normalization processing is required before each update of the embedded vectors of the entity and the relation, and the algorithm is used for k-dimensional input vector +>The transformation is as follows:, wherein ,/>Is a minimum value, avoiding the denominator being 0. To this end, for any project entity +.>An embedding vector can be obtained based on the Xavier-TransR algorithm>. Different from a loss function based on an edge (margin-based) in a basic TransR algorithm, the method calculates the pairing-wise ordering probability of the entity-relation triples in the knowledge graph by using a sigmoid function, assigns related parameters based on an Xavier initialization mode, and can effectively improve the expression capability of the model and obtain high-quality semantic representation from the knowledge graph.
In one embodiment, performing an inner product operation on the latent feature vector of the wave-recording ranging fault information and the target latent vector of the recorded sample, calculating the error compensation value and the certainty factor of the fault point may include:
wherein ,representing fault information of wave recording and distance measuringLatent feature vector +_>Target potential vector representing recorded sample, +.>Representing a vector dot product, u representing the recording ranging fault information, v representing a recording sample;
wherein ,representing error compensation value and certainty, +.>Representing an activation function->Representing a vector dot product.
The sigmoid function can be used as an activation function of an output layer, and the output predicted value is in the range of 0-1]The linear characteristic and the nonlinear characteristic can be well combined.
In an embodiment, the step 211 may include:
； wherein ,/>Representing a binary matrix>Indicating whether the wave-recording distance-measuring fault information appears in the recorded sample sequence set, m indicating the total number of the wave-recording distance-measuring fault information, and n indicating the total number of the recorded samples.
； wherein ,/>Representing a multi-valued matrix +_>Whether recorded data representing the recording ranging fault information is known or not, m represents the total number of recording ranging fault information, and n represents the total number of recorded samples.
Given wave-recording distance-measuring fault information setAnd recording a set of sample sequencesM and n respectively represent the total number of the wave-recording distance-measuring fault information and the recorded samples, wherein the wave-recording distance-measuring fault information can be divided into two types, and one type is 0-1 discrete data, such as weather and rain weather in meteorological environment information; the other type is multi-value discrete data and continuous data, such as temperature, wind power, current before failure, wave recording and distance measuring information and the like. According to the wave-recording distance-measuring fault data of the power grid line, constructing a binary matrix +. >, wherein />Meteorological environment information in record-wave distance-measuring fault information set>The +.f in the record sample sequence number set>Appear in the bar record, vice versa>. Similarly, a multi-valued matrix is constructed>, wherein />Recorded data representing recorded distance measurement fault information, whereas +.>。
In one embodiment, the wave-recording distance-measuring fault information comprises 0-1 discrete data, multi-value discrete data and continuous data; the step 211 may include:
and constructing a binary matrix and a multi-value matrix according to the 0-1 discrete data, the multi-value discrete data, the continuous data and the recorded sample sequence set.
The wave-recording distance-measuring fault information can be divided into two types, wherein one type is 0-1 discrete data, such as weather and rain weather in weather environment information; the other type is multi-value discrete data and continuous data, such as temperature, wind power, current before failure, wave recording and distance measuring information and the like. And constructing a binary matrix and a multi-value matrix according to the wave-recording and distance-measuring fault data of the power grid line, acquiring shallow characteristics of wave-recording and distance-measuring fault information through dot product operation, finally predicting an actual fault point error compensation value and the certainty factor of the error compensation value of the power grid line, and giving a ranking list of the predicted compensation value through the certainty factor to realize intelligent fault diagnosis of the power grid line.
The line fault diagnosis method provided by the application is applied to a line fault diagnosis model to realize the prediction of the error compensation value of the actual fault point and the certainty factor thereof, and a ranking list of the prediction compensation value is given according to the certainty factor, so that the intelligent fault diagnosis of the power grid line is realized, and in the model construction process, a loss function is a crucial component part of the prediction model construction and is related to the performance of a prediction algorithm. The loss function is essentially an objective optimization function of the prediction model, and can be defined according to the wave-recording distance-measuring fault information and the recorded sample sequence. Therefore, in order to optimize the line fault diagnosis model, the wave-recording ranging fault information is reconstructed based on the model loss function in the BPR algorithm. It is generally considered that, compared to the unknown recorded ranging fault information and recorded sample sequence, the BPR algorithm needs to assign a higher predicted value to the known recorded ranging fault information and recorded sample sequence, and the loss function is defined as follows:
wherein ,the method comprises the steps of recording wave-recording and distance-measuring fault information and recording a triplet set of a sample sequence, namely a training data set; />Representing known (positive example) recorded distance measurement fault information +.>And recording the sample sequence +.>A set of interactions between- >It means that unknown (negative sample) wave-recording distance-measuring fault information +.>And recording the sample sequence +.>A set of interactions between; />Representing a sigmoid function. Finally, the combination typeAnd (d) theMinimizing the model loss function below, thereby learningTo optimal model parameters:
wherein ,is a parameter set of the model; in order to avoid the over fitting phenomenon of the model in the optimization process in the calculation, the parameter set is required to be +.>L2 regularization is performed, < >>Is a coefficient of a regular term.
Fig. 5 is a schematic structural view of a line fault diagnosis apparatus according to an exemplary embodiment of the present application, and as shown in fig. 5, the line fault diagnosis apparatus 8 includes: the prediction module 81 is configured to predict a fault point of the line according to the wave-recording ranging fault data and the knowledge graph; the knowledge graph comprises wave-recording and distance-measuring fault structuring information; a ranking module 82, configured to rank the fault points according to the calculation error of each fault point; and a determining module 83 for determining an actual fault point of the line according to the ordering.
According to unstructured information of existing wave-recording distance-measuring fault data, the line fault diagnosis device 8 provided by the application predicts the error compensation value of an actual fault point and the certainty factor of the error compensation value according to the structured power grid line knowledge information in the knowledge graph, gives a ranking list of the predicted compensation value according to the certainty factor, realizes intelligent fault diagnosis of the power grid line, and can solve the problems that in the prior art, the accuracy of a positioning algorithm is low and automatic fault diagnosis is difficult due to huge quantity of fault data.
Fig. 6 is a schematic structural diagram of a line fault diagnosis apparatus according to another exemplary embodiment of the present application, and as shown in fig. 6, the sorting module 82 may include: a calculating unit 821 for calculating error compensation value and certainty factor of the fault point; a sorting unit 822 for sorting the error compensation values according to the certainty factor; a determining unit 823 for determining the order of the fault points according to the order of the error compensation values.
In an embodiment, as shown in fig. 6, the computing unit 821 may include: a constructing subunit 8211, configured to construct a binary matrix and a multi-value matrix according to the set of wave-recording ranging fault information and the set of recorded sample sequences; the calculating subunit 8212 is configured to calculate an error compensation value and a certainty factor of the fault point according to the binary matrix, the multi-valued matrix and the knowledge graph.
In an embodiment, as shown in fig. 6, the computing subunit 8212 may be further configured to: inputting the single thermal codes of the binary matrix and the multi-value matrix into a neural matrix decomposition model to obtain a binary potential feature vector and a multi-value potential feature vector of the wave-recording and distance-measuring fault information, and recording the binary potential feature vector and the multi-value potential feature vector of the sample sequence; obtaining a potential feature vector of the wave-recording distance-measuring fault information and a potential feature vector of the recorded sample sequence according to the binary potential feature vector and the multi-value potential feature vector of the wave-recording distance-measuring fault information and the binary potential feature vector and the multi-value potential feature vector of the recorded sample sequence; and calculating error compensation values and certainty of the fault points according to the potential feature vectors of the wave-recording distance-measuring fault information and the potential feature vectors of the recorded sample sequences.
In an embodiment, as shown in fig. 6, the line fault diagnosis apparatus 8 may be configured to: constructing a target potential vector of a record sample according to the wave-recording and distance-measuring fault structuring information and the potential feature vector of the record sample sequence; wherein the computing subunit 8212 may be further configured to: and carrying out inner product operation on the potential feature vector of the wave-recording ranging fault information and the target potential vector of the recorded sample, and calculating an error compensation value and certainty factor of the fault point.
In an embodiment, as shown in fig. 6, the computing subunit 8212 may be further configured to:； wherein ,/>Target potential vector representing recorded sample, +.>Potential eigenvectors representing recorded sample sequences, +.>Representing the structured embedding vector.
In an embodiment, as shown in fig. 6, the computing subunit 8212 may be further configured to:, wherein ,/>A potential feature vector representing said wave-recording distance-measuring fault information,>a target potential vector representing the recorded sample, < >>Representing a vector dot product, u represents the wave-recording distance-measuring fault information, and v represents the recorded sample; />, wherein ,/>Representing said error compensation value and said certainty +. >Representing an activation function->Representing a vector dot product.
In an embodiment, as shown in fig. 6, the above-mentioned construction subunit 8211 may be further configured to:； wherein ,/>Representing a binary matrix>Representing whether wave-recording distance-measuring fault information appears in a recorded sample sequence set, wherein m represents the total number of wave-recording distance-measuring fault information, and n represents the total number of recorded samples; />； wherein ,/>Representing a multi-valued matrix +_>Whether recorded data representing the recording ranging fault information is known or not, m represents the total number of recording ranging fault information, and n represents the total number of recorded samples.
In an embodiment, as shown in fig. 6, the above-mentioned construction subunit 8211 may be further configured to: and constructing a binary matrix and a multi-value matrix according to the 0-1 discrete data, the multi-value discrete data, the continuous data and the recorded sample sequence set.
Next, an electronic device according to an embodiment of the present application is described with reference to fig. 7. The electronic device may be either or both of the first device and the second device, or a stand-alone device independent thereof, which may communicate with the first device and the second device to receive the acquired input signals therefrom.
Fig. 7 illustrates a block diagram of an electronic device according to an embodiment of the application.
As shown in fig. 7, the electronic device 10 includes one or more processors 11 and a memory 12.
The processor 11 may be a Central Processing Unit (CPU) or other form of processing unit having data processing and/or instruction execution capabilities, and may control other components in the electronic device 10 to perform desired functions.
Memory 12 may include one or more computer program products that may include various forms of computer-readable storage media, such as volatile memory and/or non-volatile memory. The volatile memory may include, for example, random Access Memory (RAM) and/or cache memory (cache), and the like. The non-volatile memory may include, for example, read Only Memory (ROM), hard disk, flash memory, and the like. One or more computer program instructions may be stored on the computer readable storage medium that may be executed by the processor 11 to implement the line fault diagnosis method and/or other desired functions of the various embodiments of the present application described above. Various contents such as an input signal, a signal component, a noise component, and the like may also be stored in the computer-readable storage medium.
In one example, the electronic device 10 may further include: an input device 13 and an output device 14, which are interconnected by a bus system and/or other forms of connection mechanisms (not shown).
When the electronic device is a stand-alone device, the input means 13 may be a communication network connector for receiving the acquired input signals from the first device and the second device.
In addition, the input device 13 may also include, for example, a keyboard, a mouse, and the like.
The output device 14 may output various information to the outside, including the determined distance information, direction information, and the like. The output means 14 may include, for example, a display, speakers, a printer, and a communication network and remote output devices connected thereto, etc.
Of course, only some of the components of the electronic device 10 that are relevant to the present application are shown in fig. 7 for simplicity, components such as buses, input/output interfaces, etc. are omitted. In addition, the electronic device 10 may include any other suitable components depending on the particular application.
The computer program product may write program code for performing operations of embodiments of the present application in any combination of one or more programming languages, including an object oriented programming language such as Java, C++ or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computing device, partly on the user's device, as a stand-alone software package, partly on the user's computing device, partly on a remote computing device, or entirely on the remote computing device or server.
The computer readable storage medium may employ any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may include, for example, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or a combination of any of the foregoing. More specific examples (a non-exhaustive list) of the readable storage medium would include the following: an electrical connection having one or more wires, a portable disk, a hard disk, random Access Memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
The foregoing description has been presented for purposes of illustration and description. Furthermore, this description is not intended to limit embodiments of the application to the form disclosed herein. Although a number of example aspects and embodiments have been discussed above, a person of ordinary skill in the art will recognize certain variations, modifications, alterations, additions, and subcombinations thereof.
Claims (6)
1. A line fault diagnosis method, comprising:
Predicting fault points of the line according to the wave-recording distance-measuring fault data and the knowledge graph; the knowledge graph comprises wave-recording distance-measuring fault structural information, the wave-recording distance-measuring fault data comprises wave-recording distance-measuring fault information and a record sample, and a wave-recording distance-measuring fault information set is givenAnd record sample sequence set +.>Wherein m and n respectively represent the recording distance measurement fault information and the total number of recording samples;
sorting the fault points according to the calculation error of each fault point; wherein the calculated error includes an error compensation value and a certainty factor; and
determining an actual fault point of the line according to the ordering;
wherein said sorting said fault points according to the calculated error of each said fault point comprises:
calculating the error compensation value and the certainty factor of the fault point;
sorting the error compensation values according to the certainty factor; and
determining the order of the fault points according to the order of the error compensation values;
wherein said calculating said error compensation value for said fault point and said confidence level comprises:
constructing a binary matrix and a multi-value matrix according to the wave-recording distance-measuring fault information set and the recorded sample sequence set;
Calculating the error compensation value and the certainty factor of the fault point according to the binary matrix, the multi-value matrix and the knowledge graph;
the calculating the error compensation value and the certainty factor of the fault point according to the binary matrix, the multi-value matrix and the knowledge graph comprises:
inputting the single thermal codes of the binary matrix and the multi-value matrix into a neural matrix decomposition model to obtain a binary potential feature vector and a multi-value potential feature vector of the wave-recording ranging fault information, and a binary potential feature vector and a multi-value potential feature vector of the recorded sample sequence;
obtaining the potential feature vector of the wave-recording distance-measuring fault information and the potential feature vector of the recorded sample sequence according to the binary potential feature vector and the multi-value potential feature vector of the wave-recording distance-measuring fault information and the binary potential feature vector and the multi-value potential feature vector of the recorded sample sequence;
constructing a target potential vector of the record sample according to the wave-recording distance-measuring fault structural information and the potential feature vector of the record sample sequence;
and performing inner product operation on the potential feature vector of the wave-recording distance-measuring fault information and the target potential vector of the recorded sample, and calculating the error compensation value and the certainty factor of the fault point.
2. The line fault diagnosis method according to claim 1, wherein the wave-recording ranging fault structuring information comprises a structured embedding vector; the constructing the target potential vector of the record sample according to the wave-recording distance-measuring fault structural information and the potential feature vector of the record sample sequence comprises the following steps:
； wherein ,/>A target potential vector representing the recorded sample, < >>A potential feature vector representing the sequence of recorded samples, and (2)>Representing the structured embedding vector.
3. The line fault diagnosis method according to claim 2, wherein the performing an inner product operation on the latent feature vector of the wave-recording ranging fault information and the target latent vector of the record sample, and calculating the error compensation value and the certainty factor of the fault point includes:
wherein ,a potential feature vector representing said wave-recording distance-measuring fault information,>a target potential vector representing the recorded sample, < >>Representing a vector dot product, u represents the wave-recording distance-measuring fault information, and v represents the recorded sample;
wherein ,representing said error compensation value and said certainty +.>Representing an activation function- >Representing a vector dot product.
4. The line fault diagnosis method according to claim 1, wherein constructing a binary matrix and a multi-valued matrix from the set of wave-recording ranging fault information and the set of recorded sample sequences comprises:
； wherein ,/>Representing the binary matrix,/->Representing whether the wave-recording distance-measuring fault information appears in the recorded sample sequence set, wherein m represents the total number of the wave-recording distance-measuring fault information, and n represents the total number of the recorded samples;
； wherein ,/>Representing the multi-valued matrix->And whether recorded data representing the wave-recording distance-measuring fault information is known or not, wherein m represents the total number of the wave-recording distance-measuring fault information, and n represents the total number of the recorded samples.
5. The line fault diagnosis method according to claim 1, wherein the wave-recording distance-measuring fault information includes 0-1 discrete data, multi-value discrete data, and continuous data; the construction of the binary matrix and the multi-value matrix according to the wave-recording distance-measuring fault information set and the recorded sample sequence set comprises the following steps:
and constructing the binary matrix and the multi-value matrix according to the 0-1 discrete data, the multi-value discrete data, the continuous data and the record sample sequence set.
6. A line fault diagnosis apparatus, comprising:
the prediction module is used for predicting fault points of the line according to the wave-recording distance-measuring fault data and the knowledge graph; wherein the knowledge graph comprises wave recording and distance measuring fault structural information, and the wave recording and measuring comprises the steps ofThe distance fault data comprises wave-recording distance-measuring fault information and record samples, and the wave-recording distance-measuring fault information set is givenAnd recording a set of sample sequencesWherein m and n respectively represent the recording distance measurement fault information and the total number of recording samples;
the sorting module is used for sorting the fault points according to the calculation errors of each fault point; wherein the calculated error includes an error compensation value and a certainty factor; and
the determining module is used for determining the actual fault point of the line according to the sequence;
wherein, the sequencing module includes:
the calculating unit is used for calculating the error compensation value and the certainty factor of the fault point;
the sorting unit is used for sorting the error compensation values according to the certainty factor;
a determining unit, configured to determine a ranking of the fault points according to the error compensation value ranking;
wherein the computing unit includes:
The construction subunit is used for constructing a binary matrix and a multi-value matrix according to the wave-recording distance-measuring fault information set and the recorded sample sequence set;
the calculating subunit is used for calculating the error compensation value and the certainty factor of the fault point according to the binary matrix, the multi-value matrix and the knowledge graph;
wherein the computing subunit is further configured to:
inputting the single thermal codes of the binary matrix and the multi-value matrix into a neural matrix decomposition model to obtain a binary potential feature vector and a multi-value potential feature vector of the wave-recording ranging fault information, and a binary potential feature vector and a multi-value potential feature vector of the recorded sample sequence;
obtaining the potential feature vector of the wave-recording distance-measuring fault information and the potential feature vector of the recorded sample sequence according to the binary potential feature vector and the multi-value potential feature vector of the wave-recording distance-measuring fault information and the binary potential feature vector and the multi-value potential feature vector of the recorded sample sequence;
constructing a target potential vector of the record sample according to the wave-recording distance-measuring fault structural information and the potential feature vector of the record sample sequence;
And performing inner product operation on the potential feature vector of the wave-recording distance-measuring fault information and the target potential vector of the recorded sample, and calculating the error compensation value and the certainty factor of the fault point.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210358432.4A CN114757097B (en) | 2022-04-07 | 2022-04-07 | Line fault diagnosis method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210358432.4A CN114757097B (en) | 2022-04-07 | 2022-04-07 | Line fault diagnosis method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114757097A CN114757097A (en) | 2022-07-15 |
| CN114757097B true CN114757097B (en) | 2023-09-26 |
Family
ID=82328754
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210358432.4A Active CN114757097B (en) | 2022-04-07 | 2022-04-07 | Line fault diagnosis method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114757097B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115639801A (en) * | 2022-10-10 | 2023-01-24 | 合肥合锻智能制造股份有限公司 | Fault diagnosis and analysis decision platform based on multiple intelligent agents |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111209472A (en) * | 2019-12-24 | 2020-05-29 | 中国铁道科学研究院集团有限公司电子计算技术研究所 | Railway accident fault association and accident fault reason analysis method and system |
| CN111680820A (en) * | 2020-05-08 | 2020-09-18 | 国网河南省电力公司电力科学研究院 | Distributed photovoltaic power station fault diagnosis method and device |
| CN112149898A (en) * | 2020-09-21 | 2020-12-29 | 广东电网有限责任公司清远供电局 | Fault rate prediction model training method, fault rate prediction method and related device |
| CN113283027A (en) * | 2021-05-20 | 2021-08-20 | 南京航空航天大学 | Mechanical fault diagnosis method based on knowledge graph and graph neural network |
| CN113553420A (en) * | 2020-04-22 | 2021-10-26 | 国网电力科学研究院有限公司 | Power grid fault handling regulation recommendation method and system based on knowledge graph |
| CN114091912A (en) * | 2021-11-19 | 2022-02-25 | 王坤 | Method for analyzing topological transaction of medium-voltage power grid by using knowledge graph |
-
2022
- 2022-04-07 CN CN202210358432.4A patent/CN114757097B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111209472A (en) * | 2019-12-24 | 2020-05-29 | 中国铁道科学研究院集团有限公司电子计算技术研究所 | Railway accident fault association and accident fault reason analysis method and system |
| CN113553420A (en) * | 2020-04-22 | 2021-10-26 | 国网电力科学研究院有限公司 | Power grid fault handling regulation recommendation method and system based on knowledge graph |
| CN111680820A (en) * | 2020-05-08 | 2020-09-18 | 国网河南省电力公司电力科学研究院 | Distributed photovoltaic power station fault diagnosis method and device |
| CN112149898A (en) * | 2020-09-21 | 2020-12-29 | 广东电网有限责任公司清远供电局 | Fault rate prediction model training method, fault rate prediction method and related device |
| CN113283027A (en) * | 2021-05-20 | 2021-08-20 | 南京航空航天大学 | Mechanical fault diagnosis method based on knowledge graph and graph neural network |
| CN114091912A (en) * | 2021-11-19 | 2022-02-25 | 王坤 | Method for analyzing topological transaction of medium-voltage power grid by using knowledge graph |
Non-Patent Citations (5)
| Title |
|---|
| Knowledge Graph Embedding Based Collaborative Filtering;Yuhang Zhang 等;《IEEE Access 》;全文 * |
| 基于电网多元信息知识图谱的故障处置研究及应用;李金星 等;《电力信息与通信技术》;全文 * |
| 基于知识图嵌入的协同过滤推荐算法;张屹晗 等;《计算机应用研究》;第38卷(第12期);第1-8页 * |
| 基于重要度排序与全寿命周期的配网故障治理措施优化方法;郭刚 等;《电力大数据》;全文 * |
| 融合知识图谱与卷积神经网络的电压暂降分类识别;白浩 等;《电力信息与通信技术》;全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114757097A (en) | 2022-07-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111143226B (en) | Automatic test method and device, computer readable storage medium and electronic equipment | |
| CN113884961A (en) | SOC calibration method, modeling apparatus, computer device, and medium | |
| CN113780003B (en) | Cross-modal enhancement method for space-time data variable-division encoding and decoding | |
| CN111177216A (en) | Association rule generation method and device for behavior characteristics of comprehensive energy consumer | |
| CN114757097B (en) | Line fault diagnosis method and device | |
| CN115757103A (en) | Neural network test case generation method based on tree structure | |
| Fu et al. | MCA-DTCN: A novel dual-task temporal convolutional network with multi-channel attention for first prediction time detection and remaining useful life prediction | |
| CN111950646A (en) | Hierarchical knowledge model construction method and target identification method for electromagnetic image | |
| CN111161238A (en) | Image quality evaluation method and device, electronic device, and storage medium | |
| WO2023000725A1 (en) | Named entity identification method and apparatus for electric power measurement, and computer device | |
| CN116861269A (en) | Multi-source heterogeneous data fusion and analysis method in engineering field | |
| CN110045691A (en) | A kind of multitasking fault monitoring method of multi-source heterogeneous big data | |
| CN115880472A (en) | Intelligent diagnosis and analysis system for electric power infrared image data | |
| CN115391523A (en) | Wind power plant multi-source heterogeneous data processing method and device | |
| CN114595787A (en) | Recommendation model training method, recommendation device, medium and equipment | |
| CN114282657A (en) | Market data long-term prediction model training method, device, equipment and storage medium | |
| CN114021619A (en) | Classification method and device based on graph structure evolution | |
| CN110826616A (en) | Information processing method and device, electronic equipment and storage medium | |
| CN117421386B (en) | GIS-based spatial data processing method and system | |
| US20230022253A1 (en) | Fast and accurate prediction methods and systems based on analytical models | |
| CN117151247B (en) | Method, apparatus, computer device and storage medium for modeling machine learning task | |
| CN113378560B (en) | Test report intelligent diagnosis analysis method based on natural language processing | |
| CN117591813B (en) | Complex equipment fault diagnosis method and system based on multidimensional features | |
| CN112712197B (en) | Data processing method, device, equipment and storage medium | |
| CN114842251A (en) | Training method and device of image classification model, image processing method and device and computing equipment |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |