CN117590145A - Fault positioning method and system for intelligent power distribution network - Google Patents
Fault positioning method and system for intelligent power distribution network Download PDFInfo
- Publication number
- CN117590145A CN117590145A CN202311431824.XA CN202311431824A CN117590145A CN 117590145 A CN117590145 A CN 117590145A CN 202311431824 A CN202311431824 A CN 202311431824A CN 117590145 A CN117590145 A CN 117590145A
- Authority
- CN
- China
- Prior art keywords
- node
- matrix
- topology
- fault
- running state
- 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
- 238000009826 distribution Methods 0.000 title claims abstract description 77
- 238000000034 method Methods 0.000 title claims abstract description 40
- 239000011159 matrix material Substances 0.000 claims abstract description 109
- 238000013135 deep learning Methods 0.000 claims abstract description 17
- 238000012549 training Methods 0.000 claims description 81
- 239000013598 vector Substances 0.000 claims description 69
- 238000003062 neural network model Methods 0.000 claims description 20
- 238000000605 extraction Methods 0.000 claims description 14
- 238000013527 convolutional neural network Methods 0.000 claims description 13
- 230000006870 function Effects 0.000 claims description 6
- 230000004927 fusion Effects 0.000 claims description 5
- 238000012937 correction Methods 0.000 claims description 4
- 238000010276 construction Methods 0.000 claims description 2
- 238000011156 evaluation Methods 0.000 claims 1
- 238000012545 processing Methods 0.000 description 8
- 230000008439 repair process Effects 0.000 description 7
- 238000010586 diagram Methods 0.000 description 6
- 230000008859 change Effects 0.000 description 5
- 238000013136 deep learning model Methods 0.000 description 5
- 230000003993 interaction Effects 0.000 description 5
- 230000004807 localization Effects 0.000 description 5
- 230000005540 biological transmission Effects 0.000 description 4
- 230000002159 abnormal effect Effects 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 3
- 238000005457 optimization Methods 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 206010000117 Abnormal behaviour Diseases 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 239000000284 extract Substances 0.000 description 2
- 238000011835 investigation Methods 0.000 description 2
- 230000000737 periodic effect Effects 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 238000012300 Sequence Analysis Methods 0.000 description 1
- 230000005856 abnormality Effects 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 238000012790 confirmation Methods 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 230000001934 delay Effects 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000005259 measurement 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
- 238000012806 monitoring device Methods 0.000 description 1
- 238000003012 network analysis Methods 0.000 description 1
- 238000007500 overflow downdraw method Methods 0.000 description 1
- 238000011084 recovery Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 230000003252 repetitive effect Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 230000002441 reversible effect Effects 0.000 description 1
- 230000006641 stabilisation Effects 0.000 description 1
- 238000011105 stabilization Methods 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
- 238000003860 storage Methods 0.000 description 1
- 238000005728 strengthening Methods 0.000 description 1
- 238000000547 structure data Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/08—Locating faults in cables, transmission lines, or networks
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/08—Locating faults in cables, transmission lines, or networks
- G01R31/081—Locating faults in cables, transmission lines, or networks according to type of conductors
- G01R31/086—Locating faults in cables, transmission lines, or networks according to type of conductors in power transmission or distribution networks, i.e. with interconnected conductors
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/08—Locating faults in cables, transmission lines, or networks
- G01R31/088—Aspects of digital computing
-
- 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
- Y04S10/52—Outage or fault management, e.g. fault detection or location
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Theoretical Computer Science (AREA)
- Remote Monitoring And Control Of Power-Distribution Networks (AREA)
Abstract
The invention discloses a fault positioning method and a fault positioning system for an intelligent power distribution network, wherein operation data of each node in a monitored power grid at fault moment are acquired; extracting the running state characteristics of the running data of each node at the fault moment; extracting spatial topological characteristics of each node in the monitored power grid; fusing the running state features and the space topology features by using a network model based on deep learning to obtain a space topology global node running state matrix; and determining whether the node to be evaluated is a fault node or not based on the space topology global node running state matrix. Therefore, the operation data of each node in the monitored power grid at the fault moment can be analyzed by using a deep learning algorithm, and the intelligent fault location of the power distribution network is realized by using the space topological relation among the nodes.
Description
Technical Field
The invention relates to the technical field of intelligent power distribution networks, in particular to a fault positioning method and system for an intelligent power distribution network.
Background
A distribution network is a part of an electrical power system for delivering electrical energy from a power source to an end user. It is typically the last stage in the power system responsible for converting the electrical energy transmitted by the high voltage grid into electrical energy suitable for use by low voltage end users.
In a power distribution network, fault positioning is an important task, and can help an electric power company to discover and repair faults in time, so that safe and stable operation of the power grid is ensured. The main purpose of fault location is to find the fault point quickly so as to take corresponding repair measures, recover power supply and reduce the power failure time and the influence range.
Traditional fault localization relies mainly on manual operations. That is, an experienced engineer is required to interpret the measurement data and calculate the result, and there are problems of complicated operation and high time cost. Therefore, an optimized fault localization method is desired.
Disclosure of Invention
The embodiment of the invention provides a fault positioning method and a fault positioning system for an intelligent power distribution network, which are used for acquiring operation data of each node in a monitored power grid at fault moment; extracting the running state characteristics of the running data of each node at the fault moment; extracting spatial topological characteristics of each node in the monitored power grid; fusing the running state features and the space topology features by using a network model based on deep learning to obtain a space topology global node running state matrix; and determining whether the node to be evaluated is a fault node or not based on the space topology global node running state matrix. Therefore, the operation data of each node in the monitored power grid at the fault moment can be analyzed by using a deep learning algorithm, and the intelligent fault location of the power distribution network is realized by using the space topological relation among the nodes.
The embodiment of the invention also provides a fault positioning method of the intelligent power distribution network, which comprises the following steps:
acquiring operation data of each node in a monitored power grid at the fault moment;
extracting the running state characteristics of the running data of each node at the fault moment;
extracting spatial topological characteristics of each node in the monitored power grid;
fusing the running state features and the space topology features by using a network model based on deep learning to obtain a space topology global node running state matrix; and
and determining whether the node to be evaluated is a fault node or not based on the space topology global node running state matrix.
The embodiment of the invention also provides a fault positioning system of the intelligent power distribution network, which comprises the following steps:
the operation data acquisition module is used for acquiring operation data of each node in the monitored power grid at the fault moment;
the running state feature extraction module is used for extracting running state features of the running data of each node at the fault moment;
the space topology feature extraction module is used for extracting the space topology features of each node in the monitored power grid;
the fusion module is used for fusing the running state characteristics and the space topology characteristics by utilizing a network model based on deep learning so as to obtain a space topology global node running state matrix; and
and the fault node judging module is used for determining whether the node to be evaluated is a fault node or not based on the space topology global node running state matrix.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described, it being obvious that the drawings in the following description are only some embodiments of the invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art. In the drawings:
fig. 1 is a flowchart of a fault locating method of an intelligent power distribution network according to an embodiment of the present invention.
Fig. 2 is a schematic diagram of a system architecture of a fault location method for an intelligent power distribution network according to an embodiment of the present invention.
Fig. 3 is a block diagram of a fault location system of an intelligent power distribution network according to an embodiment of the present invention.
Fig. 4 is an application scenario diagram of a fault locating method of an intelligent power distribution network provided in an embodiment of the present invention.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the embodiments of the present invention will be described in further detail with reference to the accompanying drawings. The exemplary embodiments of the present invention and their descriptions herein are for the purpose of explaining the present invention, but are not to be construed as limiting the invention.
As used in the specification and in the claims, the terms "a," "an," "the," and/or "the" are not specific to a singular, but may include a plurality, unless the context clearly dictates otherwise. In general, the terms "comprises" and "comprising" merely indicate that the steps and elements are explicitly identified, and they do not constitute an exclusive list, as other steps or elements may be included in a method or apparatus.
A flowchart is used in the present invention to describe the operations performed by a system according to embodiments of the present invention. It should be understood that the preceding or following operations are not necessarily performed in order precisely. Rather, the various steps may be processed in reverse order or simultaneously, as desired. Also, other operations may be added to or removed from these processes.
Various exemplary embodiments, features and aspects of the disclosure will be described in detail below with reference to the drawings. In the drawings, like reference numbers indicate identical or functionally similar elements. Although various aspects of the embodiments are illustrated in the accompanying drawings, the drawings are not necessarily drawn to scale unless specifically indicated.
The word "exemplary" is used herein to mean "serving as an example, embodiment, or illustration. Any embodiment described herein as "exemplary" is not necessarily to be construed as preferred or advantageous over other embodiments.
The distribution grid is part of an electrical power system, and is primarily responsible for delivering electrical energy from a power source to an end user, typically the last stage in the electrical power system, to convert electrical energy transmitted by a high voltage transmission grid into electrical energy suitable for use by a low voltage end user. The main function of the distribution network is to distribute and transmit electric energy to meet the electricity demand of different users, and the distribution network comprises a series of equipment and components, such as a transformer station, a transformer, a switch device, a cable, a wire and the like, which are connected with each other through the topological structure of the power network to form a complex power transmission network.
In a power distribution network, electrical energy is fed from a high voltage transmission network through a substation to the high voltage side of the power distribution network. In a substation, electrical energy is stepped down through a transformer and then transmitted to the low voltage side of each consumer via switching devices and distribution lines. The power distribution network also bears the tasks of fault detection, fault positioning, fault recovery and the like, and when the power distribution network breaks down, such as short circuit, broken wire and the like, the power interruption and the power supply quality reduction can be caused. In order to ensure safe and stable operation of the power grid, faults need to be found in time and corresponding repair measures are adopted, fault positioning is a key step, and the power grid fault location method can help a power company to quickly find fault points and reduce power failure time and influence range.
Further, fault location is a very important task in a power distribution network, and its main purpose is to quickly find a fault point, so as to take corresponding repair measures, restore power supply, reduce outage time and influence range, and the accuracy and efficiency of fault location are critical to ensure safe and stable operation of the power distribution network. In a distribution network, faults may include short circuits, broken lines, equipment faults, etc., when a fault occurs, the power supply will be interrupted or affected, causing inconvenience and loss to the user. Thus, finding a fault in time and locating the fault point accurately is an urgent task for the electric utility company.
Traditional fault locating methods rely mainly on manual operation. When customers encounter an interruption in power supply or other abnormal situation, they report a fault to the utility company, which typically includes information such as the location and time of the power outage; the staff of the electric company can carry out preliminary fault confirmation according to the user report, and the staff can further communicate with the user, know the specific condition of the fault and verify the information in the report; engineers go to the fault site for line investigation based on the fault report and the confirmed information, and they check equipment and lines related to the fault for possible fault signs such as burnt cables, fuse tripping, etc.; during the course of line investigation, engineers will gradually narrow down the fault, they will examine adjacent line segments, excluding the parts that are working properly, to determine the area where the fault point is located. Engineers may use measuring instruments and test equipment to measure and test the line to obtain more fault information, for example, they may measure parameters such as current, voltage, resistance, etc. to help determine the location of the fault point; finally, engineers determine the approximate location of the fault point based on the collected information and experience, which may require comprehensive consideration of factors such as the physical state of the line, the operation of the equipment, user reports, etc.; once the point of failure is determined, the engineer may take corresponding repair measures, which may include replacing the failed device, repairing the cable, resetting the fuse, etc. After the repair is completed, the power supply will resume.
Conventional fault localization methods have some limitations and challenges. First, they rely on manual operation and empirical judgment, the accuracy being affected by the skill and level of experience of the operator; second, these methods generally require significant time and human resources, especially in complex failure situations; in addition, when a fault occurs in a place where the underground cable or the like is not easily overhauled, the fault location is more difficult.
Accordingly, in the present application, an optimized fault localization method is provided.
In one embodiment of the present invention, fig. 1 is a flowchart of a fault locating method of an intelligent power distribution network provided in the embodiment of the present invention. Fig. 2 is a schematic diagram of a system architecture of a fault location method for an intelligent power distribution network according to an embodiment of the present invention. As shown in fig. 1 and fig. 2, a fault locating method for an intelligent power distribution network according to an embodiment of the present invention includes: 110, acquiring operation data of each node in the monitored power grid at the fault moment; 120, extracting the operation state characteristics of the operation data of each node at the fault moment; 130, extracting the spatial topological characteristics of each node in the monitored power grid; 140, fusing the running state features and the space topology features by using a network model based on deep learning to obtain a space topology global node running state matrix; and 150, determining whether the node to be evaluated is a fault node or not based on the space topology global node running state matrix.
In step 110, it is ensured that accurate and comprehensive operational data of the individual nodes in the monitored power network at the time of the fault are acquired, which involves sensors and monitoring devices installed in the power network, and a corresponding data acquisition and storage system. Which selects suitable sensors to monitor the operating state of the grid, such as current sensors, voltage sensors, etc., arranges the sensors reasonably to cover the entire grid, and ensures the accuracy and reliability of the sensors. And establishing a corresponding data acquisition system to ensure that the operation data of each node can be acquired in real time. At the same time, a data storage system is built for subsequent data analysis and processing.
In the step 120, useful operating state characteristics are extracted from the acquired operating data, and these characteristics may reflect information such as power load, voltage stability, frequency change, etc. of the node. Wherein, selecting proper characteristics to describe the operation state of the node, such as average voltage, current fluctuation degree, power factor and the like, the characteristic selection needs to comprehensively consider the information quantity of the characteristics and the contribution degree to fault location. Based on the selected features, corresponding feature extraction methods are designed, which may involve signal processing, statistical analysis, etc. techniques to extract useful features from the raw operational data.
In the step 130, the topology structure of the monitored power grid is analyzed, and the spatial topology features of each node are extracted, which features may reflect the connection relationship between the nodes and the structural characteristics of the power grid. The connection mode and the topology characteristics between the nodes, such as the neighbor nodes of the nodes, the degree of the nodes and the like, are determined by analyzing the connection relation and the topology structure of the power grid. Corresponding feature extraction methods are designed to extract useful features from the topology. Methods used include graph theory analysis, network analysis, and the like.
In the step 140, the operation state features and the spatial topology features are fused by using the network model of deep learning, so as to obtain a spatial topology global node operation state matrix, where the spatial topology global node operation state matrix can provide operation state information of the node under the global view. Wherein a suitable deep learning network model, such as a convolutional neural network, a graph neural network, etc., is selected and model training is performed, and the training data may include operational data, operational state characteristics, and spatial topology characteristics at the time of failure. And designing a proper feature fusion method, combining the running state features and the space topology features, and inputting the running state features and the space topology features into a network model for training and prediction.
In the step 150, a space topology global node operation state matrix is used to determine whether the node to be evaluated is a fault node, which can be judged by comparing the operation state of the node to be evaluated with the operation states of the normal node and the known fault node. According to specific fault judgment criteria, a judgment method is designed to determine whether the node to be evaluated is a fault node, and the method relates to threshold setting of running state characteristics, an abnormality detection method and the like. And outputting information of whether the node to be evaluated is a fault node, possible fault types and positions and the like according to the judging result.
Through the steps, the accuracy and the efficiency of fault positioning can be improved, the power failure time and the influence range are reduced, more comprehensive power grid running state information is provided, and the quick response and the fault repair of an electric power company are facilitated. Meanwhile, a large amount of operation data and topology information can be fully utilized by utilizing the deep learning network model, and the accuracy and the automation degree of fault judgment are improved.
Aiming at the technical problems, the technical concept of the method is to analyze the operation data of each node in the monitored power grid at the fault moment by using a deep learning algorithm and realize the intelligent fault positioning of the power distribution network by using the space topological relation among each node.
Based on the above, in the technical scheme of the application, operation data of each node in the monitored power grid at the fault moment is firstly obtained. Here, the operation data at the fault time includes parameters such as voltage, current, power and the like of each node in the power grid, and these parameters reflect the operation state and fault information of each node at the fault time.
It should be appreciated that the operational status characteristics of the failed node and the normal node exist in the operational data of each node at the time of the failure. The operational status of a failed node is typically characterized by abnormal changes in parameters such as voltage, current, power, etc., such as voltage drops, current increases, power fluctuations, etc. The operating state characteristics of the normal node are represented by changes in the normal range of parameters such as voltage, current, power, etc., for example, voltage stabilization, current balance, power constancy, etc. Therefore, in the technical scheme of the application, the operation data of each node at the fault moment passes through the full connection layer to obtain the sequence of the node operation state coding vector. That is, the operational state characteristics of the respective nodes at the time of failure are extracted.
In a specific embodiment of the present application, extracting the operation state characteristics of the operation data of each node at the fault moment includes: the operation data of each node at the fault moment passes through a full connection layer to obtain a sequence of node operation state coding vectors; and taking the sequence of the node running state coding vectors as the running state characteristics.
It should be understood that, through the fully connected layer, the original multidimensional operation data can be converted into a sequence of node operation state encoding vectors, so as to extract key feature information, reduce the dimension of the data to a dimension more suitable for processing, further reduce the complexity of the data, and improve the efficiency of subsequent processing. The full connection layer can learn the association and interaction between the nodes, and potential relations between the nodes, such as similarity, dependency and influence between the nodes, can be captured through processing the node operation data by the full connection layer, so that the node operation state characteristics can be described more accurately.
After the node running state code vectors are serialized, they can be further analyzed and processed as running state features to extract patterns and trends in the sequence, and further reveal the features of dynamic changes and faults between nodes, for example, a time sequence analysis method can be used to detect periodic changes or abnormal behaviors. The node running state coding vector sequence is used as a running state characteristic, so that the input requirement of a deep learning model can be better met, the deep learning model generally has certain requirements on the dimension and the shape of input data, and the running state coding vector sequence can be more conveniently input into the deep learning model for training and prediction.
And then, extracting the spatial topological characteristics of each node in the monitored power grid. Here, the spatial topological relation between the nodes is considered to represent the interconnection condition and the distance relation of the nodes in the power grid. When a fault occurs in the power grid, abnormal changes are not only generated in the fault node, the position of the fault point can influence the operation state of surrounding nodes, the operation state of the surrounding nodes can influence the operation state of nodes at a farther position, and a dynamic ripple effect is generated. It should be appreciated that capturing spatial topological features of individual nodes in the detected grid may characterize node-to-node associations to some extent.
In a specific example of the present application, extracting spatial topology features of each node in the monitored power grid includes: constructing a topology matrix among all nodes in the monitored power grid, wherein the values of all positions on non-diagonal positions in the topology matrix are used for representing the distance between the two corresponding nodes; the topological matrix passes through a topological feature extractor based on a convolutional neural network model to obtain a spatial topological feature matrix; and taking the space topology feature matrix as the space topology feature.
The distance between the corresponding two nodes can be represented by a numerical value on the off-diagonal position of the topology matrix to capture the spatial relationship and distance information between the nodes. For example, the distance between nodes may reflect transmission delays in the grid, cable lengths, etc., which are important for fault localization and analysis. The topological feature extractor based on the convolutional neural network model can learn the topological structure and the association among the nodes, and can extract the representative spatial topological feature by inputting the topological matrix into the topological feature extractor. These features may reflect the connection, dependency and interaction relationships between nodes, helping to better understand the structure and operational state of the power grid.
The topology matrix can be converted into a space topology feature matrix through the topology feature extractor, the space topology feature matrix can be regarded as an abstract representation of the power grid topology structure, numerical values among the space topology feature matrix reflect the topology relation and the features among the nodes, and the space topology feature matrix can serve as a part of the running state features to provide more comprehensive information describing the power grid. The topological feature extractor based on the convolutional neural network model is suitable for processing matrix data, can fully utilize structural information in a topological matrix, can better adapt to the complexity of a power grid topological structure, and extracts topological features which are meaningful for fault positioning. Meanwhile, the deep learning model also has good generalization capability, and can process power grid topologies of different scales and structures.
And then, the space topology characteristic matrix and the sequence of the node running state coding vectors pass through a graph neural network model to obtain a space topology global node running state matrix. Here, the graph neural network model is a deep learning model capable of effectively processing graph structure data, and features and information of nodes can be extracted by using relationships among the nodes in the graph. In the technical scheme of the application, the graph neural network model can fuse the sequence of the space topology characteristic matrix and the node running state coding vector, so that a space topology global node running state matrix is obtained, and the running state and the mutual influence of each node in the power grid at the fault moment can be reflected by the matrix. That is, the space topology global node running state matrix can represent the running state of each node and aggregate the space topology association and the space-time dependency relationship between each node.
In a specific embodiment of the present application, the fusing the operation state features and the spatial topology features to obtain a spatial topology global node operation state matrix by using a network model based on deep learning includes: and the space topology feature matrix and the sequence of the node running state coding vectors are processed through a graph neural network model to obtain the space topology global node running state matrix.
Through the graph neural network model, the space topological feature matrix and the sequence of the node running state coding vectors can be fused, so that the space topological structure of the power grid is combined with the running state information of the nodes, and the running state of the power grid is more comprehensively described. By fusing the two kinds of information, richer and accurate characteristics can be extracted, and the method is helpful for better understanding and analyzing the overall operation condition of the power grid. The graph neural network model can learn global features and correlations among nodes, and through taking a sequence of a space topological feature matrix and node running state coding vectors as input of the graph neural network model, interaction and influence relations among the nodes can be learned in the whole power grid range, so that the graph neural network model is beneficial to revealing global modes, abnormal behaviors and fault propagation paths among the nodes, and more accurate information is provided for fault positioning and analysis. The sequence of the node running state coding vectors can reflect the dynamic change and time sequence relation of the nodes, and the sequences can be modeled and analyzed through a graph neural network model, so that the dynamic change and time sequence relation among the nodes is captured, the periodic change is detected, the future state change is predicted, and more accurate fault positioning and prediction capability is provided. By fusing space topology and node state information and modeling and analyzing by using a graph neural network model, the accuracy of fault positioning can be improved. The space topology global node operation state matrix can comprehensively consider the structural characteristics of the power grid, the operation states of the nodes and the interaction relation among the nodes, so that the position and the influence range of the fault can be more accurately determined, the power failure time and the influence range are reduced, and the safe and stable operation of the power distribution network is ensured.
Further, extracting row vectors corresponding to the nodes to be evaluated from the space topology global node running state matrix to obtain semantic feature vectors of the node running state to be evaluated; and the semantic feature vector of the operation state of the node to be evaluated passes through a classifier to obtain a classification result, wherein the classification result is used for indicating whether the node to be evaluated is a fault node or not.
In a specific embodiment of the present application, determining whether the node to be evaluated is a failure node based on the spatial topology global node operation state matrix includes: extracting row vectors corresponding to the nodes to be evaluated from the space topology global node running state matrix to obtain semantic feature vectors of the node running state to be evaluated; and passing the operation state semantic feature vector of the node to be evaluated through a classifier to obtain a classification result, wherein the classification result is used for indicating whether the node to be evaluated is a fault node or not.
In an embodiment of the present application, the fault locating method of the smart distribution network further includes a training step: training the full connection layer, the topological feature extractor based on the convolutional neural network model, the graph neural network model and the classifier; wherein the training step comprises: acquiring training data, wherein the training data comprises training operation data of each training node in a monitored power grid at the fault moment, and whether the node to be evaluated is a true value of the fault node or not; training operation data of each training node at the fault moment passes through the full connection layer to obtain a sequence of training node operation state coding vectors; performing feature distribution correction on the sequence of the training node running state coding vector to obtain a corrected sequence of the running state coding vector; constructing a training topology matrix among all training nodes in the monitored power grid, wherein values of all positions on non-diagonal positions in the training topology matrix are used for representing distances between two corresponding training nodes; passing the training topological matrix through the topological feature extractor based on the convolutional neural network model to obtain a training space topological feature matrix; the training space topology feature matrix and the sequence of the corrected node running state coding vector are passed through the graph neural network model to obtain a training space topology global node running state matrix; performing feature distribution correction on the training space topology global node operation state matrix to obtain a corrected space topology global node operation state matrix; extracting row vectors corresponding to the training nodes to be evaluated from the corrected space topology global node running state matrix to obtain semantic feature vectors of the running states of the training nodes to be evaluated; passing the semantic feature vector of the operation state of the training node to be evaluated through a classifier to obtain a classification loss function value; and training the full connection layer, the convolutional neural network model-based topological feature extractor, the graph neural network model, and the classifier with the classification loss function values.
In the technical scheme of the application, each training node running state code vector in the sequence of training node running state code vectors expresses data semantic code features of running data of corresponding nodes, so that after the training space topology feature matrix and the sequence of training node running state code vectors pass through a graph neural network model, topology association expression of the data semantic code features under a distribution distance topology in a node global space distribution domain can be further obtained, however, in consideration of expression independence of each node running state semantic feature vector, such as a row feature vector, of the training space topology global node running state matrix, the expression of the training space topology global node running state matrix as a whole for the data semantic code topology association features of each individual node may be unbalanced, and therefore the expression effect of the training node running state semantic feature vector to be evaluated, which is obtained by extracting a row vector corresponding to a node to be evaluated from the training space topology global node running state matrix, is affected.
Here, the applicant of the present application finds that such imbalance is related to a large extent to the feature expression scale, i.e. the feature vector corresponds to the data semantic coding feature scale of an individual node, and the topology association scale of the data semantic coding feature distribution in the node global spatial distribution domain of the feature matrix between the individual feature vectors, e.g. it can be understood that the more unbalanced the scale distribution corresponding to an individual node is relative to the topology association scale, the more unbalanced the overall expression of the spatial topology global node operation state matrix is.
Thus, preferably, for each training node operational state encoding vector in the sequence of training node operational state encoding vectors, e.g. denoted V i And the training space topology global node operational state matrix, e.g., denoted as M, is optimized, expressed as: optimizing each training node running state coding vector and the training space topology global node running state matrix in the sequence of training node running state coding vectors according to the following optimization formula; wherein, the optimization formula is:
wherein V is i Each training node operational state code vector in the sequence of training node operational state code vectors, M is the training space topology global node operational state matrix, L is the training node operational state code vector V i Length v of (v) ij Is the running state coding vector V of the training node i Is used to determine the characteristic value of the (j) th characteristic value,representing the training node running state code vector V i Is the square of the two norms of (S) is the scale of the training space topology global node operational state matrix, i.e. width times height, and +.>The square, m of the Frobenius norm representing the training space topology global node operational state matrix i,j Is the characteristic value, w, of each position of the running state matrix of the global node of the training space topology 1i Is the weight coefficient of the training node running state coding vector, w 2 The weight coefficient of the modified space topology global node running state matrix; and, with the weight coefficient w 1i Weighting the running state coding vector of the training node and using the weight coefficient w 2 And weighting the operation state matrix of the corrected space topology global node.
Here, the optimization can perform correlation constraint of a multi-level distribution structure on the feature probability density distribution in the high-dimensional feature space based on the feature scale through a tail distribution strengthening mechanism of a quasi-standard cauchy distribution type, so that the probability density distribution of the high-dimensional features with different scales is uniformly unfolded in the whole probability density space, and probability density convergence heterogeneity caused by feature scale deviation is compensated. Thus, at each iteration of the training process, the weight w is used 1i Weighting the running state coding vector of the training node and using the weight w 2 The method comprises the steps of weighting a corrected space topology global node running state matrix obtained from an optimized sequence of training node running state coding vectors, and improving the expression convergence of the corrected space topology global node running state matrix in a probability density domain, so that the accuracy of a classification result obtained by a classifier through the training to-be-evaluated node running state semantic feature vector obtained by extracting row vectors corresponding to-be-evaluated nodes from the corrected space topology global node running state matrix is improved.
In summary, the fault positioning method of the intelligent power distribution network based on the embodiment of the invention is explained, which utilizes a deep learning algorithm to analyze the operation data of each node in the monitored power distribution network at the fault moment, and utilizes the space topological relation among each node to realize the intelligent fault positioning of the power distribution network.
In one embodiment of the present invention, fig. 3 is a block diagram of a fault location system of an intelligent power distribution network according to an embodiment of the present invention. As shown in fig. 3, a fault location system 200 of a smart distribution network according to an embodiment of the present invention includes: an operation data obtaining module 210, configured to obtain operation data of each node in the monitored power grid at a fault moment; an operation state feature extraction module 220, configured to extract operation state features of operation data of each node at a fault moment; a spatial topology feature extraction module 230, configured to extract spatial topology features of each node in the monitored power grid; a fusion module 240, configured to fuse the operation state feature and the spatial topology feature by using a network model based on deep learning to obtain a spatial topology global node operation state matrix; and a fault node judging module 250, configured to determine whether the node to be evaluated is a fault node based on the spatial topology global node operation state matrix.
In the fault location system of the smart distribution network, the operation state feature extraction module includes: the full connection unit is used for enabling the operation data of each node at the fault moment to pass through the full connection layer to obtain a sequence of node operation state coding vectors; and an operation state feature generation unit, configured to take the sequence of the node operation state encoding vectors as the operation state feature.
In the fault location system of the smart distribution network, the spatial topological feature extraction module includes: the topology matrix construction unit is used for constructing a topology matrix among all nodes in the monitored power grid, wherein the values of all positions on the non-diagonal positions in the topology matrix are used for representing the distance between the two corresponding nodes; the topological feature extraction unit is used for enabling the topological matrix to pass through a topological feature extractor based on a convolutional neural network model to obtain a spatial topological feature matrix; and a spatial topological feature generation unit, configured to use the spatial topological feature matrix as the spatial topological feature.
In the fault location system of the smart distribution network, the fusion module is configured to: and the space topology feature matrix and the sequence of the node running state coding vectors are processed through a graph neural network model to obtain the space topology global node running state matrix.
Here, it will be understood by those skilled in the art that the specific functions and operations of the respective units and modules in the fault location system of the smart distribution network described above have been described in detail in the above description of the fault location method of the smart distribution network with reference to fig. 1 to 2, and thus, repetitive descriptions thereof will be omitted.
As described above, the fault location system 200 for a smart distribution network according to an embodiment of the present invention may be implemented in various terminal devices, for example, a server for fault location of a smart distribution network, etc. In one example, the fault location system 200 of the smart distribution network according to an embodiment of the present invention may be integrated into the terminal device as a software module and/or hardware module. For example, the fault location system 200 of the smart distribution network may be a software module in the operating system of the terminal device or may be an application developed for the terminal device; of course, the fault location system 200 of the smart distribution network may also be one of a plurality of hardware modules of the terminal device.
Alternatively, in another example, the fault location system 200 of the smart distribution network and the terminal device may be separate devices, and the fault location system 200 of the smart distribution network may be connected to the terminal device through a wired and/or wireless network and transmit the interaction information according to an agreed data format.
Fig. 4 is an application scenario diagram of a fault locating method of an intelligent power distribution network provided in an embodiment of the present invention. As shown in fig. 4, in this application scenario, first, operation data of each node in the monitored power grid at the time of failure is acquired (for example, C as illustrated in fig. 4); the acquired operational data is then input to a server (e.g., S as illustrated in fig. 4) deployed with a fault location algorithm for the smart distribution network, wherein the server is capable of processing the operational data based on the fault location algorithm for the smart distribution network to determine whether the node to be evaluated is a faulty node.
The foregoing description of the embodiments has been provided for the purpose of illustrating the general principles of the invention, and is not meant to limit the scope of the invention, but to limit the invention to the particular embodiments, and any modifications, equivalents, improvements, etc. that fall within the spirit and principles of the invention are intended to be included within the scope of the invention.
Claims (10)
1. The fault positioning method for the intelligent power distribution network is characterized by comprising the following steps of:
acquiring operation data of each node in a monitored power grid at the fault moment;
extracting the running state characteristics of the running data of each node at the fault moment;
extracting spatial topological characteristics of each node in the monitored power grid;
fusing the running state features and the space topology features by using a network model based on deep learning to obtain a space topology global node running state matrix; and
and determining whether the node to be evaluated is a fault node or not based on the space topology global node running state matrix.
2. The fault location method of a smart distribution network according to claim 1, wherein extracting the operation state characteristics of the operation data of each node at the fault moment comprises:
the operation data of each node at the fault moment passes through a full connection layer to obtain a sequence of node operation state coding vectors; and
and taking the sequence of the node running state coding vectors as the running state characteristics.
3. The fault location method of a smart distribution network according to claim 2, wherein extracting spatial topology features of each node in the monitored power network comprises:
constructing a topology matrix among all nodes in the monitored power grid, wherein the values of all positions on non-diagonal positions in the topology matrix are used for representing the distance between the two corresponding nodes;
the topological matrix passes through a topological feature extractor based on a convolutional neural network model to obtain a spatial topological feature matrix; and
and taking the space topological feature matrix as the space topological feature.
4. A fault locating method for a smart distribution network according to claim 3, wherein fusing the operational state features and the spatial topology features to obtain a spatial topology global node operational state matrix using a deep learning based network model comprises:
and the space topology feature matrix and the sequence of the node running state coding vectors are processed through a graph neural network model to obtain the space topology global node running state matrix.
5. The fault location method of a smart distribution network of claim 4, wherein determining whether the node under evaluation is a faulty node based on the spatial topology global node operation state matrix comprises:
extracting row vectors corresponding to the nodes to be evaluated from the space topology global node running state matrix to obtain semantic feature vectors of the node running state to be evaluated; and
and the semantic feature vector of the operation state of the node to be evaluated passes through a classifier to obtain a classification result, wherein the classification result is used for indicating whether the node to be evaluated is a fault node or not.
6. The fault location method of intelligent distribution network according to claim 5, further comprising the training step of: training the full connection layer, the topological feature extractor based on the convolutional neural network model, the graph neural network model and the classifier;
wherein the training step comprises:
acquiring training data, wherein the training data comprises training operation data of each training node in a monitored power grid at the fault moment, and whether the node to be evaluated is a true value of the fault node or not;
training operation data of each training node at the fault moment passes through the full connection layer to obtain a sequence of training node operation state coding vectors;
performing feature distribution correction on the sequence of the training node running state coding vector to obtain a corrected sequence of the running state coding vector;
constructing a training topology matrix among all training nodes in the monitored power grid, wherein values of all positions on non-diagonal positions in the training topology matrix are used for representing distances between two corresponding training nodes;
passing the training topological matrix through the topological feature extractor based on the convolutional neural network model to obtain a training space topological feature matrix;
the training space topology feature matrix and the sequence of the corrected node running state coding vector are passed through the graph neural network model to obtain a training space topology global node running state matrix;
performing feature distribution correction on the training space topology global node operation state matrix to obtain a corrected space topology global node operation state matrix;
extracting row vectors corresponding to the training nodes to be evaluated from the corrected space topology global node running state matrix to obtain semantic feature vectors of the running states of the training nodes to be evaluated;
passing the semantic feature vector of the operation state of the training node to be evaluated through a classifier to obtain a classification loss function value; and
training the full connection layer, the topological feature extractor based on the convolutional neural network model, the graph neural network model and the classifier with the classification loss function values.
7. A fault location system for an intelligent distribution network, comprising:
the operation data acquisition module is used for acquiring operation data of each node in the monitored power grid at the fault moment;
the running state feature extraction module is used for extracting running state features of the running data of each node at the fault moment;
the space topology feature extraction module is used for extracting the space topology features of each node in the monitored power grid;
the fusion module is used for fusing the running state characteristics and the space topology characteristics by utilizing a network model based on deep learning so as to obtain a space topology global node running state matrix; and
and the fault node judging module is used for determining whether the node to be evaluated is a fault node or not based on the space topology global node running state matrix.
8. The fault location system of a smart distribution network of claim 7, wherein the operational state feature extraction module comprises:
the full connection unit is used for enabling the operation data of each node at the fault moment to pass through the full connection layer to obtain a sequence of node operation state coding vectors; and
and the running state characteristic generating unit is used for taking the sequence of the node running state coding vectors as the running state characteristic.
9. The fault location system of a smart distribution network of claim 8, wherein the spatial topology feature extraction module comprises:
the topology matrix construction unit is used for constructing a topology matrix among all nodes in the monitored power grid, wherein the values of all positions on the non-diagonal positions in the topology matrix are used for representing the distance between the two corresponding nodes;
the topological feature extraction unit is used for enabling the topological matrix to pass through a topological feature extractor based on a convolutional neural network model to obtain a spatial topological feature matrix; and
and the space topology feature generation unit is used for taking the space topology feature matrix as the space topology feature.
10. The fault location system of a smart distribution network of claim 9, wherein the fusion module is configured to:
and the space topology feature matrix and the sequence of the node running state coding vectors are processed through a graph neural network model to obtain the space topology global node running state matrix.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311431824.XA CN117590145A (en) | 2023-10-31 | 2023-10-31 | Fault positioning method and system for intelligent power distribution network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311431824.XA CN117590145A (en) | 2023-10-31 | 2023-10-31 | Fault positioning method and system for intelligent power distribution network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117590145A true CN117590145A (en) | 2024-02-23 |
Family
ID=89912426
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311431824.XA Pending CN117590145A (en) | 2023-10-31 | 2023-10-31 | Fault positioning method and system for intelligent power distribution network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117590145A (en) |
-
2023
- 2023-10-31 CN CN202311431824.XA patent/CN117590145A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11674994B2 (en) | Distribution fault location using graph neural network with both node and link attributes | |
| US10886736B2 (en) | Post-disaster topology detection and energy flow recovery in power distribution network | |
| Peppanen et al. | Distribution system low-voltage circuit topology estimation using smart metering data | |
| CN110470951B (en) | Active power distribution network fault diagnosis method based on PMU information and Petri network | |
| Gilanifar et al. | Multi-task logistic low-ranked dirty model for fault detection in power distribution system | |
| CN103018592A (en) | Fault diagnosis method of traction transformer based on model diagnosis | |
| CN101907868A (en) | Intelligent trouble diagnosis method for tractive power supply system and system thereof | |
| CN113169709A (en) | Solar power generation control system and method based on machine learning | |
| Kezunovic | Translational knowledge: From collecting data to making decisions in a smart grid | |
| CN110412417B (en) | Micro-grid data fault diagnosis method based on intelligent power monitoring instrument | |
| CN113220751A (en) | Metering system and evaluation method for multi-source data state quantity | |
| CN115453267A (en) | Fault diagnosis system for electric power information system | |
| CN103324858A (en) | Three-phase load flow state estimation method of power distribution network | |
| Bai et al. | Frequency response of pv inverters toward high renewable penetrated distribution networks | |
| CN113987724A (en) | Power grid risk identification method and system based on topology analysis | |
| CN116400172A (en) | Cloud-edge cooperative power distribution network fault detection method and system based on random matrix | |
| Marashi et al. | Quantification and analysis of interdependency in cyber-physical systems | |
| CN117590145A (en) | Fault positioning method and system for intelligent power distribution network | |
| CN115663886A (en) | Fault identification processing method and system for distributed power distribution network | |
| CN115684829A (en) | Power secondary circuit fault detection early warning method and system | |
| CN115563786A (en) | Electrochemical energy storage power station fault scene reconstruction method | |
| Kotsalos et al. | On the development of a framework for the advanced monitoring of LV grids | |
| CN113591937B (en) | Power system key node identification method based on local distance coding | |
| Pani et al. | A Graph-Theoretic Approach to Assess the Power Grid Vulnerabilities to Transmission Line Outages | |
| Yuanhai | Robust Estimators in Modern Powe1· System Networks |
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 |