CN117454148A - Power system state estimation method, device, electronic equipment and storage medium - Google Patents
Power system state estimation method, device, electronic equipment and storage medium Download PDFInfo
- Publication number
- CN117454148A CN117454148A CN202311420489.3A CN202311420489A CN117454148A CN 117454148 A CN117454148 A CN 117454148A CN 202311420489 A CN202311420489 A CN 202311420489A CN 117454148 A CN117454148 A CN 117454148A
- Authority
- CN
- China
- Prior art keywords
- state estimation
- power system
- node
- data
- input
- 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
- 238000000034 method Methods 0.000 title claims abstract description 57
- 239000011159 matrix material Substances 0.000 claims abstract description 81
- 238000012549 training Methods 0.000 claims abstract description 30
- 238000002347 injection Methods 0.000 claims description 28
- 239000007924 injection Substances 0.000 claims description 28
- 238000004590 computer program Methods 0.000 claims description 16
- 230000006870 function Effects 0.000 claims description 15
- 230000005856 abnormality Effects 0.000 claims description 11
- 238000012423 maintenance Methods 0.000 claims description 4
- 238000013528 artificial neural network Methods 0.000 abstract description 5
- 230000007246 mechanism Effects 0.000 abstract description 5
- 238000010586 diagram Methods 0.000 description 11
- 238000004891 communication Methods 0.000 description 8
- 238000012545 processing Methods 0.000 description 8
- 239000013598 vector Substances 0.000 description 7
- 230000008569 process Effects 0.000 description 5
- 230000003287 optical effect Effects 0.000 description 3
- 239000000243 solution Substances 0.000 description 3
- 238000013527 convolutional neural network Methods 0.000 description 2
- 238000012937 correction Methods 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000009466 transformation Effects 0.000 description 2
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 230000004913 activation Effects 0.000 description 1
- 230000002776 aggregation Effects 0.000 description 1
- 238000004220 aggregation Methods 0.000 description 1
- 238000003491 array Methods 0.000 description 1
- 238000013473 artificial intelligence Methods 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000001413 cellular effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 239000011521 glass Substances 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 239000004973 liquid crystal related substance Substances 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 238000010248 power generation Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 230000001953 sensory effect Effects 0.000 description 1
- 238000000547 structure data Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J3/00—Circuit arrangements for ac mains or ac distribution networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
-
- 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
- G06N3/042—Knowledge-based neural networks; Logical representations of neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/0464—Convolutional networks [CNN, ConvNet]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/047—Probabilistic or stochastic networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/048—Activation functions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/0985—Hyperparameter optimisation; Meta-learning; Learning-to-learn
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/06—Electricity, gas or water supply
-
- H—ELECTRICITY
- H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
- H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
- H02J2203/00—Indexing scheme relating to details of circuit arrangements for AC mains or AC distribution networks
- H02J2203/20—Simulating, e g planning, reliability check, modelling or computer assisted design [CAD]
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Business, Economics & Management (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Economics (AREA)
- Water Supply & Treatment (AREA)
- Human Resources & Organizations (AREA)
- Probability & Statistics with Applications (AREA)
- Power Engineering (AREA)
- Bioinformatics & Computational Biology (AREA)
- Public Health (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The embodiment of the invention discloses a power system state estimation method, a device, electronic equipment and a storage medium. The method comprises the following steps: acquiring node input characteristics of each node in a target power system, and forming a characteristic input matrix based on the node input characteristics; inputting the characteristic input matrix into a pre-trained state estimation graph network model to obtain state estimation data of each node in a target power system; the state estimation graph network model consists of a graph attention network and a feature learning unit and is obtained through training of a system topology structure of a target power system. According to the technical scheme provided by the embodiment of the invention, the graph neural network and the attention mechanism can be introduced into the state estimation network model, and the state estimation data can be obtained under the condition that the structural information between each node and each line in the target power system is fully considered, so that the accuracy of the obtained state estimation data is higher, and the generalization and the robustness are stronger.
Description
Technical Field
The embodiment of the invention relates to the technical field of power, in particular to a power system state estimation method, a device, electronic equipment and a storage medium.
Background
The new energy power generation scale is gradually enlarged, the structure is gradually complicated, the load fluctuation condition of the power system is more frequent, and the challenge is brought to the safe operation of the power system. It is therefore important to ensure that the power system is safe to operate by estimating the state of the cable system to provide effective, accurate underlying data.
In the prior art, a state estimation mode based on data driving such as a fully connected neural network and a convolutional neural network is adopted to perform state estimation on each node of a power system. However, in view of increasingly complicated power grid topology and diversified elements, the existing state estimation mode based on data driving only can mine potential relation between system measurement information and state quantity, and structural information between power grid nodes and lines is not considered, so that the robustness and generalization of the prior art for multi-topology scenes are poor, the situation that topology changes occur in actual operation of a power system cannot be adapted, and estimation results are low in accuracy and effectiveness.
Disclosure of Invention
The embodiment of the invention provides a power system state estimation method, a device, electronic equipment and a storage medium, so as to achieve the purposes of higher accuracy and stronger generalization of state estimation data.
According to an aspect of the present invention, there is provided a power system state estimation method including:
acquiring node input characteristics of each node in a target power system, and forming a characteristic input matrix based on each node input characteristic;
inputting the characteristic input matrix into a pre-trained state estimation graph network model to obtain state estimation data of each node in the target power system;
the state estimation graph network model consists of a graph attention network and a feature learning unit and is obtained through training of a system topology structure of the target power system.
According to another aspect of the present invention, there is provided a power system state estimation apparatus including:
the node input characteristic acquisition module is used for acquiring node input characteristics of all nodes in the target power system and forming a characteristic input matrix based on all the node input characteristics;
the characteristic input matrix input module is used for inputting the characteristic input matrix into a pre-trained state estimation graph network model to obtain state estimation data of each node in the target power system;
the state estimation graph network model consists of a graph attention network and a feature learning unit and is obtained through training of a system topology structure of the target power system.
According to another aspect of the present invention, there is provided an electronic apparatus including:
at least one processor; and
a memory communicatively coupled to the at least one processor; wherein,
the memory stores a computer program executable by the at least one processor to enable the at least one processor to perform the power system state estimation method according to any one of the embodiments of the present invention.
According to another aspect of the present invention, there is provided a computer readable storage medium storing computer instructions for causing a processor to implement the power system state estimation method according to any one of the embodiments of the present invention when executed.
According to the technical scheme, the node input characteristics of all nodes in the target power system are obtained, and a characteristic input matrix is formed based on the node input characteristics; inputting the characteristic input matrix into a pre-trained state estimation graph network model to obtain state estimation data of each node in a target power system; in this embodiment, the state estimation data is obtained through a state estimation graph network model, where the state estimation graph network model is composed of a graph attention network and a feature learning unit, and is obtained through system topology training of a target power system. Therefore, the graph neural network and the attention mechanism are introduced into the state estimation network model, and the state estimation data can be obtained under the condition that the structural information between each node and the line in the target power system is fully considered, so that the accuracy of the obtained state estimation data is higher, and the generalization and the robustness are stronger.
It should be understood that the description in this section is not intended to identify key or critical features of the embodiments of the invention or to delineate the scope of the invention. Other features of the present invention will become apparent from the description that follows.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly described below, and it is apparent that the drawings in the following description are only some embodiments of the present invention, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
Fig. 1 is a flowchart of a power system state estimation method according to an embodiment of the present invention;
FIG. 2 is a flow chart of another power system state estimation method provided in accordance with an embodiment of the present invention;
FIG. 3 is a schematic diagram of a structure of an attention-based state estimation graph network to be trained according to an embodiment of the present invention;
fig. 4 is a schematic structural diagram of a power system state estimation device according to an embodiment of the present invention;
fig. 5 is a schematic structural diagram of an electronic device implementing a power system state estimation method according to an embodiment of the present invention.
Detailed Description
In order that those skilled in the art will better understand the present invention, a technical solution in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in which it is apparent that the described embodiments are only some embodiments of the present invention, not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the present invention without making any inventive effort, shall fall within the scope of the present invention.
It should be noted that the terms "first," "second," and the like in the description and the claims of the present invention and the above figures are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate such that the embodiments of the invention described herein may be implemented in sequences other than those illustrated or otherwise described herein. Furthermore, the terms "comprises," "comprising," and "includes," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
Fig. 1 is a flowchart of a power system state estimation method according to an embodiment of the present invention. The embodiment is applicable to the situation that the state of each node in the power system is estimated, and the method can be executed by a power system state estimation device, and the power system state estimation device can be implemented in a form of hardware and/or software.
As shown in fig. 1, the method of this embodiment may specifically include:
s110, acquiring node input characteristics of each node in the target power system, and forming a characteristic input matrix based on the node input characteristics.
The node of the target power system may be a power station in the target power system, the node input feature may include active injection power and/or reactive injection power, and the feature input matrix is a power matrix formed by splicing the active injection power and/or the reactive injection power.
In this embodiment, the node input features include active injection power and reactive injection power; the method for obtaining the node input characteristics of each node in the target power system and forming the characteristic input matrix based on the node input characteristics can be as follows: acquiring active injection power and reactive injection power of each node in a target power system; and performing splicing operation on the active injection power and the reactive injection power of each node to form a characteristic input matrix.
Specifically, each node in the target power system can be connected with a sensor, and the active injection power and the reactive injection power corresponding to each node can be determined through data acquired by the sensor. And splicing the active injection power and the reactive injection power of each node to obtain a characteristic input matrix.
By way of example, if the target power system includes 39 nodes, a matrix Pinj (39,1) composed of active injection power of each Node and a matrix Qinj (39,1) composed of reactive injection power of each Node can be obtained, and by splicing the matrix Pinj (39,1) and the matrix Qinj (39,1), a feature input matrix node_input (39,2) is obtained; the channel direction of the characteristic input matrix is 2 dimensions, and the characteristic direction is 39 dimensions.
S120, inputting the characteristic input matrix into a pre-trained state estimation graph network model to obtain state estimation data of each node in the target power system.
The state estimation graph network model consists of a graph attention network and a feature learning unit and is obtained through training of a system topology structure of a target power system.
For example, the state estimation graph network model may be composed of one graph attention network (Graph Attention Network, GAT) and two feature learning units (leannablelaylayer) in series.
In a specific implementation, the feature input matrix may be input into a state estimation graph network model, and the output result of the state estimation graph network model is determined as state estimation data of each node in the target power system. Wherein the state estimation data comprises a voltage amplitude estimate and/or a voltage phase angle estimate.
Optionally, the method for inputting the feature input matrix into the pre-trained state estimation graph network model to obtain the state estimation data of each node in the target power system may be: and inputting the characteristic input matrix into a pre-trained state estimation graph network model to obtain the voltage amplitude estimation value and/or the voltage phase angle estimation value of each node in the target power system.
Specifically, the characteristic input matrix is input into a pre-trained state estimation graph network model, and a voltage amplitude matrix and/or a voltage phase angle matrix corresponding to the target power system are obtained through input. Wherein the dimensions of the voltage magnitude matrix and the voltage phase angle matrix may be determined by the number of nodes in the target power system. For example, if the number of nodes in the target power system is 39, the state estimation graph network model may output a voltage magnitude matrix V m (39,1) and a voltage phase angle matrix V a (39,1); the voltage estimated value of each node can be determined through the voltage amplitude matrix, and the voltage phase angle estimated value of each node can be determined through the voltage phase angle matrix.
According to the method and the device, the voltage amplitude estimated value and/or the voltage phase angle estimated value in the target power system are obtained, so that the states of all nodes of the target power system in operation are effectively estimated, and the safety of the target power system in operation is improved conveniently.
In order to better ensure the working stability and safety of the power system, after obtaining the state estimation data of each node in the target power system, the method further comprises the following steps: determining whether the state estimation data is within a preset normal data range; if not, generating state abnormality prompt information, and sending the state abnormality prompt information to the operation and maintenance terminal.
Specifically, when determining whether the state estimation data is within the preset normal data range, the voltage amplitude estimation value and the voltage phase angle estimation value corresponding to each node in the target power system can be judged, and if the voltage amplitude estimation value or the voltage phase angle estimation value of a certain node exceeds the preset normal data range, the state estimation data is determined not to be within the preset normal data range. And if the voltage amplitude estimated value and the voltage phase angle estimated value of each node are both in the preset normal data range, determining that the state estimated data are in the preset normal range.
Further, if the state estimation data is not within the preset normal data range, it is indicated that the target power system is in a normal working state, and the state estimation data can be recorded. If the state estimation data is not in the preset normal data range, it is indicated that the working state of the node in the target power system may be abnormal, and in order to avoid the safety problem caused by the working abnormality, the state abnormality prompt information may be generated based on at least one of the voltage amplitude estimation value, the voltage phase angle estimation value and the node information of the corresponding node which are not in the preset normal data range, and the state abnormality information may be sent to the operation and maintenance terminal, so that the staff may find the abnormality in time and process in time, thereby avoiding the safety problem.
According to the technical scheme, the node input characteristics of all nodes in the target power system are obtained, and a characteristic input matrix is formed based on the node input characteristics; inputting the characteristic input matrix into a pre-trained state estimation graph network model to obtain state estimation data of each node in a target power system; in this embodiment, the state estimation data is obtained through a state estimation graph network model, where the state estimation graph network model is composed of a graph attention network and a feature learning unit, and is obtained through system topology training of a target power system. Therefore, the graph neural network and the attention mechanism are introduced into the state estimation network model, and the state estimation data can be obtained under the condition that the structural information between each node and the line in the target power system is fully considered, so that the accuracy of the obtained state estimation data is higher, and the generalization and the robustness are stronger.
Fig. 2 is a flowchart of another power system state estimation method according to an embodiment of the present invention. Optionally, before inputting the feature input matrix into the pre-trained state estimation graph network model, the method further comprises: acquiring historical input characteristics and historical state real data corresponding to a target power system, and generating a sample data set based on the historical input characteristics and the corresponding historical state real data; and training the state estimation graph network to be trained based on the attention based on the sample data set to obtain a state estimation graph network model. Wherein, the explanation of the same or corresponding terms as the above embodiments is not repeated herein. As shown in fig. 2, the method includes:
s210, acquiring historical input characteristics and historical state real data corresponding to the target power system, and generating a sample data set based on the historical input characteristics and the corresponding historical state real data.
The historical input features are, for example, active and/or reactive injection power of the target power system over a historical period of time; the historical state real data are the voltage amplitude and/or the voltage phase angle actually output by the target power system corresponding to each historical input characteristic.
In the specific implementation, a mode of acquiring equal time intervals can be adopted to acquire the history input characteristics and the corresponding history state real data in the history time period, and a sample data set is obtained by the history input characteristics and the history state real data corresponding to each acquisition time; and a random acquisition mode can be adopted to acquire a preset number of historical input features and corresponding historical state real data, so that a sample data set is formed.
S220, training the state estimation graph network based on the attention to be trained based on the sample data set to obtain a state estimation graph network model.
An exemplary structure diagram of the attention-based state estimation graph network to be trained is shown in fig. 3, and the attention-based state estimation graph network to be trained is composed of one graph attention network (Graph Attention Network, GAT) and two feature learning units (leannablelayersyer) connected in series.
In a specific implementation, training a state estimation graph network based on attention to be trained based on a sample data set, and obtaining a state estimation graph network model includes: respectively inputting a plurality of historical input features in the sample data set into a state estimation graph network to be trained based on attention, and outputting a plurality of historical state estimation data; determining a total loss value based on the corresponding historical state real data and the historical state estimation data for each historical input feature so as to respectively correct model parameters in the attention-based state estimation graph network based on the total loss value; and taking convergence of a loss function in the state estimation graph network based on the attention as a training target to obtain a state estimation graph network model.
It should be noted that the sample data set includes a plurality of history input features, and history state real data corresponding to each history input feature. When the attention-based state estimation graph network to be trained is trained, a plurality of history input features can be respectively input into the attention-based state estimation graph network to be trained, so that history state estimation data corresponding to each history input feature is obtained.
And comparing the historical state real data with the historical state estimation data to determine a loss value. For example, a mean square error loss function may be used to determine the total loss value. And correcting model parameters in the attention-based state estimation graph network through the obtained total loss value, converging a loss function in the attention-based state estimation graph network to serve as a training target, determining model parameters meeting the training target, and obtaining a state estimation graph network model based on the determined model parameters.
The model is trained through the loss function, and accuracy and effectiveness of training results are improved.
In a specific implementation, determining the total loss value based on the corresponding historical state real data and the historical state estimation data includes: determining a voltage amplitude loss value based on the corresponding historical voltage amplitude real data and historical voltage amplitude estimation data; determining a voltage phase angle loss value based on the corresponding historical voltage phase angle real data and historical voltage phase angle estimation data; the total loss value is determined based on the voltage amplitude loss value, the voltage phase angle loss value, and a preset coefficient.
Specifically, the voltage amplitude Loss value loss_v is determined by means of a mean square error Loss function MSELoss (), historical voltage amplitude real data and historical voltage amplitude estimated data m The method comprises the steps of carrying out a first treatment on the surface of the Determining a voltage phase angle Loss value loss_V through a mean square error Loss function MSELess (), historical voltage phase angle real data and historical voltage phase angle estimation data a The total Loss value Loss calculation formula may be:
Loss＝Loss_V m +λ*Loss_V a
wherein lambda is a preset coefficient, and Loss_V can be controlled by the preset coefficient a And (5) performing constraint. Illustratively, λ may be set to 2, and the super-parameters during training are set to: the learning rate was set to 0.001, using Adam optimizer.
In the embodiment, the total loss value is determined by determining the voltage amplitude loss value and the voltage phase angle loss value, and the influence of the model training result on the voltage amplitude and the voltage phase angle is considered, so that the accuracy of the model obtained by training is improved comprehensively.
In this embodiment, before training the Attention-based state estimation graph network to be trained, an Attention-based state estimation graph network (Attention-based State Estimation Graph Network, ASEG-Net) needs to be built. The attention-based state estimation graph network comprises a graph attention network and a plurality of characteristic learning units which are connected in series; the number of feature learning units may be 2, for example.
In order to better help understand the attention-based state estimation graph network, each network link of the graph attention network and feature learning will be described.
1. For a graph attention network, each network link includes: and constructing a node adjacency matrix according to the system topology structure, and initializing a weight matrix. For each node, N first-order adjacent points are found out according to the introduced adjacent matrix, wherein N is the number of the first-order adjacent points of the node, namely the corresponding degree of the node. And multiplying the feature vectors of the node and the adjacent nodes by weight matrixes needed to participate in training through a graph attention network to obtain linear transformation vectors.
In addition, attention mechanisms are introduced to the drawing attention, which can assign a different attention score to each neighbor point, thereby identifying more important neighbor points. The method comprises the following specific steps: the obtained linear transformation vectors are spliced and processed through a max () function to obtain initial attention coefficients, after the attention coefficients are processed through a LeakyReLU () function and a Softmax () function, normalized attention coefficients are obtained, the attention coefficients reflect different influences of different neighbors on the node, and an attention coefficient matrix can be obtained by combining all the attention coefficients.
Multiplying the attention coefficient matrix by the linear variable vector spliced matrix, and then obtaining target features through sigma () activation function, wherein the target features represent feature vectors updated by each node according to neighbor feature information of the node and adjacent points thereof. And parallelizing all the nodes, so that after the power matrix is injected into the node input GAT module, the node characteristic diagram with embedded characteristics can be obtained.
2. For the feature learning units, each feature learning unit is composed of an autonomously designed graph feature learning graph convolutional neural network (Learnable Graph Convolutional Networks, LGCN), a node feature self-learning subunit. The input matrix of the LGCN comprises a node characteristic diagram and a node adjacency matrix; the output matrix has 1: the learned node characteristic diagram is the state estimation data of the power system in the application scene. Each network link of the feature learning unit comprises the following parts.
Constructing a neighbor node matrix: for each node, N first-order adjacent points are found out according to the introduced adjacent matrix, wherein N is the number of the first-order adjacent points of the node, namely the corresponding degree of the node; and obtaining a matrix containing all neighbor characteristic information of each node, namely a neighbor node matrix.
And carrying out zero padding and eigenvector splicing on the neighbor node matrix to obtain tensors corresponding to the neighbor node matrix, and realizing the process of processing the graph structure data into grid-like data.
Drawing convolution operation: and performing convolution operation on the tensor, then splicing along the channel dimension to obtain processed feature vectors of each node after feature aggregation and feature filtering, and splicing the processed feature vectors of each node to obtain an output matrix of the LGCN.
After obtaining an output matrix of the LGCN, firstly using 1 Multi-Layer Perceptron (MLP) to learn in a characteristic dimension direction, adding elements of the obtained intermediate input characteristic diagram and the output matrix output by the LGCN module, finally using 1 MLP Layer to learn in a channel dimension direction, and performing dimension reduction processing on the obtained matrix to obtain a 2-dimensional matrix, namely a voltage amplitude matrix V m And a voltage phase angle matrix V a . Thus, the estimation of the voltage amplitude and the voltage phase angle of each node in the power system is realized; and by adding a residual structure, layer normalization and Dropout layer in the network link, the network convergence is accelerated, the network prediction accuracy is improved, and the network overfitting condition is slowed down.
S230, acquiring node input characteristics of all nodes in the target power system, and forming a characteristic input matrix based on the node input characteristics.
S240, inputting the characteristic input matrix into a pre-trained state estimation graph network model to obtain state estimation data of each node in the target power system.
The state estimation graph network model consists of a graph attention network and a feature learning unit and is obtained through training of a system topology structure of a target power system.
Fig. 4 is a schematic structural diagram of a power system state estimation device according to an embodiment of the present invention, where the device is configured to perform the power system state estimation method according to any of the foregoing embodiments. The device belongs to the same inventive concept as the power system state estimation method of each of the above embodiments, and reference may be made to the above embodiments of the power system state estimation method for details that are not described in detail in the embodiments of the power system state estimation device. As shown in fig. 4, the apparatus includes:
the node input feature acquisition module 10 is configured to acquire node input features of each node in the target power system, and form a feature input matrix based on the node input features;
the feature input matrix input module 11 is configured to input a feature input matrix into a state estimation graph network model that is trained in advance, so as to obtain state estimation data of each node in the target power system;
the state estimation graph network model consists of a graph attention network and a feature learning unit and is obtained through training of a system topology structure of a target power system.
On the basis of any optional technical scheme in the embodiment of the invention, optional node input features comprise active injection power and reactive injection power; the node input feature acquisition module 10 includes:
the power acquisition unit is used for acquiring active injection power and reactive injection power of each node in the target power system;
and the splicing unit is used for carrying out splicing operation on the active injection power and the reactive injection power of each node to form a characteristic input matrix.
On the basis of any optional technical scheme in the embodiment of the invention, optionally, the state estimation data comprises a voltage amplitude estimation value and/or a voltage phase angle estimation value;
the feature input matrix input module 11 includes:
the characteristic input unit is used for inputting the characteristic input matrix into the pre-trained state estimation graph network model to obtain the voltage amplitude estimation value and/or the voltage phase angle estimation value of each node in the target power system.
On the basis of any optional technical scheme in the embodiment of the invention, the method further comprises the following steps:
the sample data set generation module is used for acquiring historical input characteristics and historical state real data corresponding to the target power system before inputting the characteristic input matrix into the state estimation graph network model which is trained in advance, and generating a sample data set based on the historical input characteristics and the corresponding historical state real data;
and the network training module is used for training the state estimation graph network based on the attention to be trained based on the sample data set to obtain a state estimation graph network model.
On the basis of any optional technical scheme in the embodiment of the invention, an optional network training module comprises:
the estimating data output unit is used for respectively inputting a plurality of historical input features in the sample data set into the attention-based state estimating graph network to be trained and outputting a plurality of historical state estimating data;
a model parameter correction unit for determining a total loss value based on the corresponding historical state real data and the historical state estimation data for each historical input feature, so as to respectively correct the model parameters in the attention-based state estimation graph network based on the total loss value;
and the state estimation graph network generation unit is used for converging a loss function in the state estimation graph network based on the attention as a training target to obtain a state estimation graph network model.
On the basis of any optional technical scheme in the embodiment of the invention, an optional model parameter correction unit comprises:
the voltage amplitude loss value determining subunit is used for determining a voltage amplitude loss value based on corresponding historical voltage amplitude real data and historical voltage amplitude estimation data;
a voltage phase angle loss value determining subunit, configured to determine a voltage phase angle loss value based on the corresponding historical voltage phase angle real data and historical voltage phase angle estimation data;
and the total loss value determining subunit is used for determining the total loss value based on the voltage amplitude loss value, the voltage phase angle loss value and the preset coefficient.
On the basis of any optional technical scheme in the embodiment of the invention, the method further comprises the following steps:
the state abnormality prompt information sending module is used for determining whether the state estimation data are in a preset normal data range after obtaining the state estimation data of each node in the target power system; if not, generating state abnormality prompt information, and sending the state abnormality prompt information to the operation and maintenance terminal.
According to the technical scheme, the node input characteristics of all nodes in the target power system are obtained, and a characteristic input matrix is formed based on the node input characteristics; inputting the characteristic input matrix into a pre-trained state estimation graph network model to obtain state estimation data of each node in a target power system; in this embodiment, the state estimation data is obtained through a state estimation graph network model, where the state estimation graph network model is composed of a graph attention network and a feature learning unit, and is obtained through system topology training of a target power system. Therefore, the graph neural network and the attention mechanism are introduced into the state estimation network model, and the state estimation data can be obtained under the condition that the structural information between each node and the line in the target power system is fully considered, so that the accuracy of the obtained state estimation data is higher, and the generalization and the robustness are stronger.
It should be noted that, in the embodiment of the power system state estimation device, each unit and module included are only divided according to the functional logic, but not limited to the above-mentioned division, so long as the corresponding functions can be implemented; in addition, the specific names of the functional units are also only for distinguishing from each other, and are not used to limit the protection scope of the present invention.
Fig. 5 is a schematic structural diagram of an electronic device implementing a power system state estimation method according to an embodiment of the present invention. Electronic devices are intended to represent various forms of digital computers, such as laptops, desktops, workstations, personal digital assistants, servers, blade servers, mainframes, and other appropriate computers. Electronic equipment may also represent various forms of mobile devices, such as personal digital processing, cellular telephones, smartphones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions, are meant to be exemplary only, and are not meant to limit implementations of the inventions described and/or claimed herein.
As shown in fig. 5, the electronic device 20 includes at least one processor 21, and a memory, such as a Read Only Memory (ROM) 22, a Random Access Memory (RAM) 23, etc., communicatively connected to the at least one processor 21, wherein the memory stores a computer program executable by the at least one processor, and the processor 21 can perform various suitable actions and processes according to the computer program stored in the Read Only Memory (ROM) 22 or the computer program loaded from the storage unit 28 into the Random Access Memory (RAM) 23. In the RAM23, various programs and data required for the operation of the electronic device 20 may also be stored. The processor 21, the ROM22 and the RAM23 are connected to each other via a bus 24. An input/output (I/O) interface 25 is also connected to bus 24.
Various components in the electronic device 20 are connected to the I/O interface 25, including: an input unit 26 such as a keyboard, a mouse, etc.; an output unit 27 such as various types of displays, speakers, and the like; a storage unit 28 such as a magnetic disk, an optical disk, or the like; and a communication unit 29 such as a network card, modem, wireless communication transceiver, etc. The communication unit 29 allows the electronic device 20 to exchange information/data with other devices via a computer network, such as the internet, and/or various telecommunication networks.
The processor 21 may be a variety of general and/or special purpose processing components having processing and computing capabilities. Some examples of processor 21 include, but are not limited to, a Central Processing Unit (CPU), a Graphics Processing Unit (GPU), various specialized Artificial Intelligence (AI) computing chips, various processors running machine learning model algorithms, digital Signal Processors (DSPs), and any suitable processor, controller, microcontroller, etc. The processor 21 performs the various methods and processes described above, such as the method power system state estimation method.
In some embodiments, the power system state estimation method may be implemented as a computer program tangibly embodied on a computer-readable storage medium, such as the storage unit 28. In some embodiments, part or all of the computer program may be loaded and/or installed onto the electronic device 20 via the ROM22 and/or the communication unit 29. When the computer program is loaded into the RAM23 and executed by the processor 21, one or more steps of the power system state estimation method described above may be performed. Alternatively, in other embodiments, the processor 21 may be configured to perform the power system state estimation method by any other suitable means (e.g., by means of firmware).
Various implementations of the systems and techniques described here above may be implemented in digital electronic circuitry, integrated circuit systems, field Programmable Gate Arrays (FPGAs), application Specific Integrated Circuits (ASICs), application Specific Standard Products (ASSPs), systems On Chip (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and/or combinations thereof. These various embodiments may include: implemented in one or more computer programs, the one or more computer programs may be executed and/or interpreted on a programmable system including at least one programmable processor, which may be a special purpose or general-purpose programmable processor, that may receive data and instructions from, and transmit data and instructions to, a storage system, at least one input device, and at least one output device.
A computer program for carrying out methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general purpose computer, special purpose computer, or other programmable data processing apparatus, such that the computer programs, when executed by the processor, cause the functions/acts specified in the flowchart and/or block diagram block or blocks to be implemented. The computer program may execute entirely on the machine, partly on the machine, as a stand-alone software package, partly on the machine and partly on a remote machine or entirely on the remote machine or server.
In the context of the present invention, a computer-readable storage medium may be a tangible medium that can contain, or store a computer program for use by or in connection with an instruction execution system, apparatus, or device. The computer readable storage medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. Alternatively, the computer readable storage medium may be a machine readable signal medium. More specific examples of a machine-readable storage medium would include an electrical connection based on one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
To provide for interaction with a user, the systems and techniques described here can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to a user; and a keyboard and a pointing device (e.g., a mouse or a trackball) through which a user can provide input to the electronic device. Other kinds of devices may also be used to provide for interaction with a user; for example, feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form, including acoustic input, speech input, or tactile input.
The systems and techniques described here can be implemented in a computing system that includes a background component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front-end component (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such background, middleware, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: local Area Networks (LANs), wide Area Networks (WANs), blockchain networks, and the internet.
The computing system may include clients and servers. The client and server are typically remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other. The server can be a cloud server, also called a cloud computing server or a cloud host, and is a host product in a cloud computing service system, so that the defects of high management difficulty and weak service expansibility in the traditional physical hosts and VPS service are overcome.
It should be appreciated that various forms of the flows shown above may be used to reorder, add, or delete steps. For example, the steps described in the present invention may be performed in parallel, sequentially, or in a different order, so long as the desired results of the technical solution of the present invention are achieved, and the present invention is not limited herein.
The above embodiments do not limit the scope of the present invention. It will be apparent to those skilled in the art that various modifications, combinations, sub-combinations and alternatives are possible, depending on design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the scope of the present invention.
Claims (10)
1. A method for estimating a state of an electrical power system, comprising:
acquiring node input characteristics of each node in a target power system, and forming a characteristic input matrix based on each node input characteristic;
inputting the characteristic input matrix into a pre-trained state estimation graph network model to obtain state estimation data of each node in the target power system;
the state estimation graph network model consists of a graph attention network and a feature learning unit and is obtained through training of a system topology structure of the target power system.
2. The method of claim 1, wherein the node input features include active and reactive injection power;
the obtaining node input features of each node in the target power system, forming a feature input matrix based on each node input feature, includes:
acquiring active injection power and reactive injection power of each node in a target power system;
and performing splicing operation on the active injection power and the reactive injection power of each node to form the characteristic input matrix.
3. The method according to claim 1, wherein the state estimation data comprises a voltage amplitude estimate and/or a voltage phase angle estimate;
the step of inputting the characteristic input matrix into a pre-trained state estimation graph network model to obtain state estimation data of each node in the target power system, comprising the following steps:
and inputting the characteristic input matrix into a pre-trained state estimation graph network model to obtain voltage amplitude estimation values and/or voltage phase angle estimation values of all nodes in the target power system.
4. The method of claim 1, further comprising, prior to said inputting the feature input matrix into a pre-trained state estimation graph network model:
acquiring historical input characteristics and historical state real data corresponding to the target power system, and generating a sample data set based on the historical input characteristics and the corresponding historical state real data;
training a state estimation graph network based on attention to be trained based on a sample data set to obtain the state estimation graph network model.
5. The method of claim 4, wherein training the attention-based state estimation graph network to be trained based on the sample dataset to obtain the state estimation graph network model comprises:
respectively inputting a plurality of historical input features in the sample data set into a state estimation graph network to be trained based on attention, and outputting a plurality of historical state estimation data;
for each history input feature, determining a total loss value based on the corresponding history state real data and history state estimation data, so as to respectively correct model parameters in the attention-based state estimation graph network based on the total loss value;
and taking convergence of a loss function in the state estimation graph network based on the attention as a training target to obtain the state estimation graph network model.
6. The method of claim 5, wherein the determining a total loss value based on the corresponding historical state real data and the historical state estimation data comprises:
determining a voltage amplitude loss value based on the corresponding historical voltage amplitude real data and historical voltage amplitude estimation data;
determining a voltage phase angle loss value based on the corresponding historical voltage phase angle real data and historical voltage phase angle estimation data;
and determining the total loss value based on the voltage amplitude loss value, the voltage phase angle loss value and a preset coefficient.
7. The method according to claim 1, wherein after obtaining the state estimation data of each node in the target power system, further comprising:
determining whether the state estimation data is within a preset normal data range;
if not, generating state abnormality prompting information, and sending the state abnormality prompting information to the operation and maintenance terminal.
8. An electric power system state estimation device, characterized by comprising:
the node input characteristic acquisition module is used for acquiring node input characteristics of all nodes in the target power system and forming a characteristic input matrix based on all the node input characteristics;
the characteristic input matrix input module is used for inputting the characteristic input matrix into a pre-trained state estimation graph network model to obtain state estimation data of each node in the target power system;
the state estimation graph network model consists of a graph attention network and a feature learning unit and is obtained through training of a system topology structure of the target power system.
9. An electronic device, the electronic device comprising:
at least one processor; and
a memory communicatively coupled to the at least one processor; wherein,
the memory stores a computer program executable by the at least one processor to enable the at least one processor to perform the power system state estimation method of any one of claims 1-7.
10. A computer readable storage medium storing computer instructions for causing a processor to implement the power system state estimation method of any one of claims 1-7 when executed.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311420489.3A CN117454148A (en) | 2023-10-30 | 2023-10-30 | Power system state estimation method, device, electronic equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311420489.3A CN117454148A (en) | 2023-10-30 | 2023-10-30 | Power system state estimation method, device, electronic equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117454148A true CN117454148A (en) | 2024-01-26 |
Family
ID=89581350
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311420489.3A Pending CN117454148A (en) | 2023-10-30 | 2023-10-30 | Power system state estimation method, device, electronic equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117454148A (en) |
-
2023
- 2023-10-30 CN CN202311420489.3A patent/CN117454148A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN115147687A (en) | Student model training method, device, equipment and storage medium | |
| CN116307215A (en) | Load prediction method, device, equipment and storage medium of power system | |
| CN113965313B (en) | Model training method, device, equipment and storage medium based on homomorphic encryption | |
| CN115421044A (en) | Lithium ion battery state determination method and device, electronic equipment and storage medium | |
| CN117332897A (en) | Artificial intelligent driven new energy small time scale power interpolation integrated prediction method | |
| CN113393034A (en) | Electric quantity prediction method of online self-adaptive OSELM-GARCH model | |
| CN117454148A (en) | Power system state estimation method, device, electronic equipment and storage medium | |
| CN114372036B (en) | State estimation method, device, equipment and computer storage medium for power system | |
| CN115952916A (en) | Artificial intelligence-based wind power prediction error correction method, device and equipment | |
| CN113313261B (en) | Function processing method and device and electronic equipment | |
| CN114943384A (en) | Transformer substation load prediction method, device, equipment and storage medium | |
| CN112667957A (en) | Intelligent electric energy meter failure rate prediction method based on deep neural network | |
| CN113963433B (en) | Motion search method, motion search device, electronic equipment and storage medium | |
| CN116488256A (en) | Power system unit combination method, device, equipment and storage medium | |
| CN117117849A (en) | Photovoltaic power prediction method, device, equipment and storage medium | |
| CN117950306A (en) | PID control parameter determining method, device, equipment and medium | |
| CN117009902A (en) | Data detection method, device, equipment and storage medium | |
| CN118012218A (en) | Maximum power tracking method and device of photovoltaic inverter and electronic equipment | |
| CN116258068A (en) | Transient stability evaluation method and device for power system and computer equipment | |
| CN117707899A (en) | Micro-service abnormality detection method, device, equipment and storage medium | |
| CN116484719A (en) | Method, device, equipment and storage medium for determining internal temperature rise value | |
| CN116087659A (en) | Data processing method, device, electronic equipment and storage medium | |
| CN117687993A (en) | Data migration method, device, equipment and storage medium | |
| CN117934137A (en) | Bad asset recovery prediction method, device and equipment based on model fusion | |
| CN116840588A (en) | Aging detection method, device and equipment for lightning arrester and storage medium |
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 |