Power distribution network state identification method and device, computer equipment and storage medium

Abstract

The invention provides a power distribution network state identification method and device, computer equipment and a storage medium, and belongs to the field of power system topological structure identification research, and the method comprises the steps: abstracting a power distribution network into a line graph model; constructing an adjacent matrix and a node feature matrix according to the line graph model; constructing a power distribution network state identification network; and inputting the adjacent matrix and the node characteristic matrix into a power distribution network state identification network to obtain the state of the circuit breaker. By abstracting the power distribution network into the line graph model, the complex conditions of various connection nodes and lines in the power distribution network are simplified, influence factors can be reduced, then the matrix reflecting the condition of the power distribution network is constructed through the line graph model, the matrix is recognized through the power distribution network state recognition network, the condition of the circuit breaker can be accurately obtained, and the accuracy of the circuit breaker is improved. Therefore, under the current large-scale distributed power supply access condition, according to the dynamic change of the circuit breaker in the power distribution network, the topological structure of the power distribution network can be conveniently monitored in real time, and topological data support is provided for operation of a power distribution network optimization graph.

Description

Technical Field

[0001] The present invention belongs to the research field of power system topology identification, and specifically relates to a distribution network state identification method, device, computer equipment and storage medium. Background Art

[0002] The large-scale integration of distributed energy has greatly increased the uncertainty of the operation of the distribution network itself, and has put forward higher requirements for the planning, scheduling, and operation and maintenance management of the distribution network. Accurate distribution network topology is the basis for distribution network flow calculation, fault location, state estimation and other analysis functions, and is also the prerequisite for carrying out power grid planning, operation and scheduling. Therefore, it is of great significance to study the identification of distribution network status through distribution network topology identification methods.

[0003] In the prior art, convolutional neural networks are usually used to identify the topology of power grids. However, when using convolutional neural networks (CNNs) to identify the connection relationship of power grid topology images, CNNs mainly focus on extracting local features in images and gradually expand the receptive field through convolution operations to capture a wider range of spatial information. However, when processing images with complex spatial relationships, CNNs find it difficult to fully capture these relationships, especially when the spatial relationships span a large range or present irregular patterns. This results in low accuracy in identifying the topology of power grids and large errors in identifying the state of distribution networks. Summary of the invention

[0004] In order to solve the problem of lack of efficient judgment method for the on-off status of a circuit breaker, the present invention provides a distribution network status recognition method, device, computer equipment and storage medium.

[0005] In order to achieve the above object, the present invention provides the following technical solutions:

[0006] First, a distribution network state identification method is provided, the method comprising:

[0007] Acquire multiple circuit breakers of the target distribution network, and abstract the distribution network into a line graph model according to the multiple circuit breakers;

[0008] Constructing an adjacency matrix reflecting the adjacent relationship between each circuit breaker and a node characteristic matrix reflecting the voltage distribution of each circuit breaker according to the line graph model;

[0009] Constructing a graph neural network algorithm, and training the graph neural network algorithm through pre-acquired training samples to obtain a distribution network state recognition network;

[0010] The adjacency matrix and the node feature matrix are input into the distribution network state identification network to obtain the state of the circuit breaker in the target distribution network connection node.

[0011] Optionally, abstracting the distribution network into a line diagram model according to the plurality of circuit breakers comprises:

[0012] The plurality of circuit breakers are abstracted into connection nodes, the lines between the circuit breakers are converted into edges, and the adjacent nodes and edges are connected to each other to form a line graph model.

[0013] Optionally, the graph neural network algorithm includes an input layer, a convolution layer, a pooling layer and an output layer; the input adjacency matrix and the node feature matrix are received through the input layer, and the convolution layer extracts the convolution feature matrix of the adjacency matrix and the node feature matrix. The specific expression is:

[0014] H l =σ(L sym H l-1 W l );

[0015] Among them, Hl is the convolution feature matrix output by the lth convolution layer, the 0th layer is the input layer, Wl is the parameter matrix, σ is the activation function; Lsym is the normalized Laplace matrix;

[0016] The feature matrix is ​​reduced in dimension by the pooling layer, and further feature extraction is performed to obtain a state feature vector. The output layer classifies the state feature vector by a Softmax classifier to determine the state corresponding to the state feature vector.

[0017] Optionally, the expression of the normalized Laplace matrix is:

[0018]

[0019] Among them, A is the adjacency matrix of the graph, IN is the identity matrix, D is the degree matrix of the graph, and D ii is the value of the i-th row and i-th column of the degree matrix D, is the diagonal weighted matrix of A, Represents the value of row i and column j.

[0020] Optionally, the training samples include a historical adjacency matrix and a historical node feature matrix of the target distribution network and a real circuit breaker state corresponding to the adjacency matrix and the historical node feature matrix; the step of training the graph neural network algorithm using the pre-acquired training samples to obtain a distribution network state recognition network includes:

[0021] The historical adjacency matrix and the historical node feature matrix are input into a graph neural network algorithm to obtain a predicted circuit breaker state. The graph neural network algorithm is trained with the goal of minimizing the difference between the predicted circuit breaker state and the actual circuit breaker state to obtain a distribution network state recognition network.

[0022] Optionally, the method further comprises:

[0023] According to the state of the circuit breaker, the topology of the target distribution network is determined.

[0024] Secondly, a distribution network state identification device is provided, the device comprising:

[0025] An acquisition module, used for acquiring a plurality of circuit breakers of a target distribution network, and abstracting the distribution network into a line graph model according to the plurality of circuit breakers;

[0026] A construction module is configured to construct an adjacency matrix reflecting the adjacent relationship between each circuit breaker and a node feature matrix reflecting the voltage distribution of each circuit breaker according to the line graph model; a graph neural network algorithm is constructed, and the graph neural network algorithm is trained by pre-acquired training samples to obtain a distribution network state recognition network;

[0027] The identification module is used to input the adjacency matrix and the node feature matrix into the distribution network state identification network to obtain the state of the circuit breaker in the target distribution network connection node.

[0028] In addition, a computer-readable storage medium is provided, wherein the storage medium stores a computer program, and when the computer program is executed by a processor, the above-mentioned distribution network state identification method is implemented.

[0029] Finally, a computer device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above-mentioned distribution network status identification method when executing the program.

[0030] The distribution network state identification method provided by the present invention has the following beneficial effects:

[0031] First, by abstracting the distribution network into a line graph model, the complex situation of various connection nodes and lines in the distribution network is simplified, which is conducive to reducing influencing factors. Secondly, a matrix reflecting the distribution network situation is constructed through the line graph model, and the distribution network state recognition network obtained by graph neural network training is used to identify the matrix. It is effectively represented and learned in the form of topological structure, which can capture more global information in the power grid topology structure. Through the connection relationship of the topological structure, the local features are integrated into the global representation, and the circuit breaker situation is confirmed. In this way, under the current large-scale distributed power supply access situation, according to the dynamic changes of the circuit breakers in the distribution network, it is convenient to monitor the distribution network topology in real time, and provide topological data support for the operation of the distribution network optimization graph. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the embodiment of the present invention and its design scheme, the following briefly introduces the drawings required for this embodiment. The drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0033] Figure 1 The present invention is a flowchart of a method for identifying a distribution network state according to an exemplary embodiment of the present invention.

[0034] Figure 2 A schematic diagram of line graph model conversion provided according to an exemplary embodiment of the present invention.

[0035] Figure 3 The present invention is a block diagram of a device for identifying a distribution network state according to an exemplary embodiment of the present invention. DETAILED DESCRIPTION

[0036] In order to enable those skilled in the art to better understand the technical solution of the present invention and implement it, the present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and cannot be used to limit the scope of protection of the present invention.

[0037] The technical solutions provided by various embodiments of the present invention are described in detail below in conjunction with the accompanying drawings.

[0038] First, the present invention provides a method for identifying the state of a distribution network, specifically, Figure 1 As shown, the following steps are included:

[0039] S101. Acquire connection nodes of a target distribution network, and abstract the distribution network into a line graph model according to the connection nodes.

[0040] Among them, the connection node includes a circuit breaker.

[0041] Specifically, the circuit breaker is abstracted into a connection node, and the lines between the circuit breakers are converted into edges to obtain a line graph model.

[0042] For example, the distributed power and load units of the distribution network are connected through lines and circuit breakers. When the switch state of a circuit breaker changes, the network topology of the distribution network changes. The distributed power and load nodes (transformers in the distribution network) in the distribution network are abstracted into nodes, and the lines between the nodes are regarded as edges.

[0043] S102: constructing an adjacency matrix reflecting the neighbor relationship between each connection node and a node characteristic matrix reflecting the voltage distribution of each connection node according to the line graph model.

[0044] The voltage distribution can be obtained by converting the line diagram model. Specifically, Figure 2 As shown in the figure, the line graph model is converted, that is, the nodes and edges are interchanged. The converted node information includes circuit breaker information and voltage difference, that is, the circuit breaker switch of the original distribution network. The circuit breaker is closed and the circuit breaker is open and the voltage difference is 0. For example, the SCADA system and AMI system can be used to obtain the voltage data of the distribution network nodes. The voltages of node 1 and node 2 are V N1 and V N2 , then the voltage difference V diff =V N1 -V N2 , the circuits 1 and 2 are closed, then the circuit breaker input data Y s =1, then the node information is {V diff ,Y s}.

[0045] S103: construct a graph neural network algorithm, and train the graph neural network algorithm using pre-acquired training samples to obtain a distribution network state recognition network.

[0046] The graph neural network algorithm includes an input layer, a convolution layer, a pooling layer, and an output layer. Specifically, the input layer receives the input matrix, wherein the input data of the input layer is an N×M feature matrix composed of an adjacency matrix and a node feature matrix, where N is the number of nodes and M is the dimension of the input data vector; the feature matrix is ​​extracted through the convolution layer, and the specific expression is:

[0047] H l =σ(L sym H l-1 W l ).

[0048] Among them, H l is the output matrix of the lth convolutional layer, the 0th layer is the input layer, W l is the parameter matrix to be trained, σ is the activation function; L sym is a normalized Laplace matrix; the feature matrix is ​​reduced in dimension through the pooling layer, and further features are extracted to obtain a state feature vector; the output layer classifies the state feature vector through a Softmax classifier to determine the state corresponding to the state feature vector.

[0049] Among them, the normalized Laplace matrix L sym The expression is:

[0050]

[0051] Among them, A is the adjacency matrix of the graph, I N is the identity matrix, D is the degree matrix of the graph, and D ii is the value of the i-th row and i-th column of the degree matrix D, is the diagonal weighted matrix of A, Represents the value of row i and column j.

[0052] After the graph neural network algorithm is constructed, the graph neural network algorithm is first initialized. For example, the structure and parameters of the graph neural network can be initialized. The specific initialization content is as shown in Table 1 below:

[0053] Table 1 Initialization content

[0054] parameter Numeric Number of graph convolutional layers 2 Activation Function ReLU Loss Function Cross Entropy Loss Function Epoch 300 Learning Rate 0.02 Number of hidden neurons 1 Number of fully connected layers 1

[0055] After initializing the graph neural network algorithm, it is also necessary to train the graph neural network algorithm with pre-acquired training samples to obtain a distribution network state recognition network. The training samples include the historical adjacency matrix and historical node feature matrix of the target distribution network and their corresponding real circuit breaker states; the specific training process can be as follows:

[0056] The historical adjacency matrix and the historical node feature matrix are input into the graph neural network algorithm to obtain the predicted circuit breaker state. The graph neural network algorithm is trained with the goal of minimizing the difference between the predicted circuit breaker state and the actual circuit breaker state to obtain the distribution network state recognition network.

[0057] In addition, the training samples can be randomly divided into training set, validation set and test set in the proportion of 50%, 20% and 30% respectively in advance. The distribution network state recognition network is obtained by applying the graph neural network algorithm to the training set, and then the distribution network state recognition network is verified and tested by the validation set and test set respectively.

[0058] S104, inputting the adjacency matrix and node feature matrix of the line graph model corresponding to the current distribution network into the distribution network state recognition network to obtain the state of the circuit breaker in the target distribution network connection node.

[0059] In this step, in order to further obtain an accurate circuit breaker state, the input data may be preprocessed.

[0060] For example, the input feature matrix is ​​also normalized. For example, the normalization process may be performed as follows:

[0061]

[0062] in, and x m are the characteristic components of the mth node before and after normalization, m∈[1,M]; x min and x max are the maximum and minimum values ​​in the feature data respectively.

[0063] After determining the state of the circuit breaker in the distribution network, the topology of the target distribution network may also be determined according to the state of the circuit breaker.

[0064] By adopting the above method, the distribution network is abstracted into a line graph model, which simplifies the complex situation of various connection nodes and lines in the distribution network, which is conducive to reducing influencing factors. Secondly, a matrix reflecting the distribution network situation is constructed through the line graph model, and the matrix is ​​identified through the distribution network state recognition network, so that the circuit breaker situation can be accurately obtained. In this way, under the current large-scale distributed power supply access situation, according to the dynamic changes of the circuit breakers in the distribution network, it is convenient to monitor the distribution network topology in real time, and provide topological data support for the operation of the distribution network optimization diagram.

[0065] Secondly, the present invention also provides a distribution network status identification device, such as Figure 3 As shown, including:

[0066] An acquisition module 301 is used to acquire a plurality of circuit breakers of a target distribution network, and abstract the distribution network into a line graph model according to the plurality of circuit breakers;

[0067] Constructing module 302, which is equivalent to constructing an adjacency matrix reflecting the adjacent relationship between each circuit breaker and a node feature matrix reflecting the voltage distribution of each circuit breaker according to the line graph model; constructing a graph neural network algorithm, and training the graph neural network algorithm through pre-acquired training samples to obtain a distribution network state recognition network;

[0068] The identification module 303 is used to input the adjacency matrix and the node feature matrix into the distribution network state identification network to obtain the state of the circuit breaker in the target distribution network connection node.

[0069] The present invention also provides a computer-readable storage medium, which stores a computer program, which can be used to execute the above Figure 1 The steps of the distribution network status identification method are provided.

[0070] The present invention also provides a computer device. At the hardware level, the computer device includes a processor, an internal bus, a network interface, a memory, and a non-volatile memory. Of course, it may also include hardware required for other services. The processor reads the corresponding computer program from the non-volatile memory into the memory and then runs it to achieve the above Figure 1 The steps of the distribution network status identification method are provided.

[0071] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0072] The present invention is described with reference to flowcharts and / or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as a combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0073] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0074] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0075] It should be noted that the above specific implementation method can enable those skilled in the art to understand the invention more comprehensively, but does not limit the invention in any way. Therefore, although the present invention has been described in detail in this specification, those skilled in the art should understand that the invention can still be modified or replaced by equivalents; and all technical solutions and improvements that do not deviate from the spirit and scope of the invention are included in the protection scope of the patent for the invention. Any figure mark in the claims should not be regarded as limiting the claims involved.

Claims

1. A distribution network status identification method, characterized in that: The method comprises: Acquire multiple circuit breakers of the target distribution network, and abstract the distribution network into a line graph model according to the multiple circuit breakers; Constructing an adjacency matrix reflecting the adjacent relationship between each circuit breaker and a node characteristic matrix reflecting the voltage distribution of each circuit breaker according to the line graph model; Constructing a graph neural network algorithm, and training the graph neural network algorithm through pre-acquired training samples to obtain a distribution network state recognition network; The adjacency matrix and the node feature matrix are input into the distribution network state identification network to obtain the state of the circuit breaker in the target distribution network connection node.

2. A distribution network status identification method according to claim 1, characterized in that: Abstracting the distribution network into a line diagram model according to the plurality of circuit breakers comprises: The plurality of circuit breakers are abstracted into connection nodes, the lines between the circuit breakers are converted into edges, and the adjacent nodes and edges are connected to each other to form a line graph model.

3. A distribution network status identification method according to claim 1, characterized in that: The graph neural network algorithm includes an input layer, a convolution layer, a pooling layer and an output layer; the input adjacency matrix and the node feature matrix are received through the input layer, and the convolution layer extracts the convolution feature matrix of the adjacency matrix and the node feature matrix. The specific expression is: H l =σ(L sym H l-1 W l ); Among them, H l is the convolution feature matrix output by the lth convolutional layer, the 0th layer is the input layer, W l is the parameter matrix, σ is the activation function; L sym is the normalized Laplace matrix; The feature matrix is ​​reduced in dimension by the pooling layer, and further feature extraction is performed to obtain a state feature vector. The output layer classifies the state feature vector by a Softmax classifier to determine the state corresponding to the state feature vector.

4. A method for identifying a distribution network state according to claim 3, characterized in that: The expression of the normalized Laplace matrix is: Among them, A is the adjacency matrix of the graph, I N is the identity matrix, D is the degree matrix of the graph, and D ii is the value of the i-th row and i-th column of the degree matrix D, is the diagonal weighted matrix of A, Represents the value of row i and column j.

5. A distribution network status identification method according to claim 1, characterized in that: The training samples include a historical adjacency matrix and a historical node feature matrix of the target distribution network and real circuit breaker states corresponding to the adjacency matrix and the historical node feature matrix; The step of training the graph neural network algorithm by using the pre-acquired training samples to obtain the distribution network state recognition network includes: The historical adjacency matrix and the historical node feature matrix are input into a graph neural network algorithm to obtain a predicted circuit breaker state. The graph neural network algorithm is trained with the goal of minimizing the difference between the predicted circuit breaker state and the actual circuit breaker state to obtain a distribution network state recognition network.

6. A distribution network status identification method according to claim 1, characterized in that: The method further comprises: According to the state of the circuit breaker, the topology of the target distribution network is determined.

7. A distribution network status identification device, characterized in that: The device comprises: An acquisition module, used for acquiring a plurality of circuit breakers of a target distribution network, and abstracting the distribution network into a line graph model according to the plurality of circuit breakers; A construction module is configured to construct an adjacency matrix reflecting the adjacent relationship between each circuit breaker and a node feature matrix reflecting the voltage distribution of each circuit breaker according to the line graph model; a graph neural network algorithm is constructed, and the graph neural network algorithm is trained by pre-acquired training samples to obtain a distribution network state recognition network; The identification module is used to input the adjacency matrix and the node feature matrix into the distribution network state identification network to obtain the state of the circuit breaker in the target distribution network connection node.

8. A computer-readable storage medium, characterized in that: The storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

9. A computer device, characterized in that: The method comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method according to any one of claims 1 to 6 is implemented.

Images