Power distribution network pseudo measurement modeling generation method, state estimation method, device and equipment

Abstract

The invention discloses a power distribution network pseudo measurement modeling generation method, a state estimation method, devices and equipment, and the method comprises the steps: inputting a node feature matrix, an edge feature matrix and an adjacent matrix in a multi-section historical graph data set into a pseudo measurement model of a power distribution network, so as to determine a predicted pseudo measurement value; updating a target parameter of the pseudo-measurement model based on the predicted pseudo-measurement value and the real pseudo-measurement value; repeating the above steps until a trained pseudo measurement model is obtained; the process of determining the predicted pseudo measurement value through the pseudo measurement model comprises the following steps: determining a node model of each node and an edge model of each edge through each hidden layer; determining an initial graph embedding vector based on the node model and the edge model; vector embedding processing is carried out on the initial image through the image pooling layer and the full connection layer so as to determine a predicted pseudo measurement value. According to the method, the prediction pseudo measurement value is determined based on the node model and the edge model, and prediction of power injection pseudo measurement of nodes without power measurement equipment or nodes with measurement data missing is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of AC power systems, and particularly to a method for generating pseudo-measurement models for distribution networks, a state estimation method, a device, and equipment. Background Art

[0002] As the penetration rate of Distributed Renewable Energy (DRE) in power systems continues to increase, distribution networks are facing a series of new challenges. Traditional distribution system designs are based on a unidirectional power flow pattern, while the integration of DRE has transformed the power flow into a bidirectional one and triggered frequent changes in the network topology. To ensure the security, stability, and economic efficiency of distribution networks, it is crucial to accurately and real-time grasp the state information and topology of the distribution system.

[0003] However, current distribution network state estimation technologies are limited by the problem of low measurement redundancy, that is, the available measurement data is limited, which restricts the comprehensive perception of the system state. In addition, traditional methods highly rely on model-driven and are difficult to adapt to rapidly changing network topologies, thus affecting their accuracy and practicality. Especially when facing the uncertainties brought by DRE, traditional state estimation methods often fail to meet the actual needs. On the other hand, data-driven methods establish the mapping relationship between grid measurements and state variables through training with historical measurement data, providing a new approach that does not completely rely on detailed physical models. However, in practical applications, with the development of distribution networks, data-driven methods also need to be continuously updated to adapt to new operating conditions. The above methods all require sufficient data redundancy and high data quality.

[0004] In view of this, pseudo-measurement modeling has become an important direction to solve the problems of distribution network state estimation. Therefore, strengthening the research on pseudo-measurement modeling is of great significance for promoting the modernization of distribution networks. Summary of the Invention

[0005] The present invention provides a method for generating pseudo-measurement models for distribution networks, a state estimation method, a device, and equipment, which realizes the prediction of power injection pseudo-measurements for nodes without power measurement equipment or nodes with missing measurement data, and improves the redundancy of measurement data in distribution networks.

[0006] In a first aspect, a method for generating a pseudo-measurement model for a distribution network is provided, including:

[0007] Inputting the node feature matrix, edge feature matrix, and adjacency matrix in the multi-section historical graph dataset into the pseudo-measurement model of the distribution network to determine the predicted pseudo-measurement values through the pseudo-measurement model;

[0008] Updating the target parameters of the pseudo-measurement model based on the predicted pseudo-measurement values and the true pseudo-measurement values;

[0009] Repeat the above steps of determining the predicted pseudo-measurement values and updating the target parameters of the pseudo-measurement model until the training stop condition is met, and obtain a trained pseudo-measurement model;

[0010] Wherein, the pseudo-measurement model is a graph neural network, and the pseudo-measurement model includes a graph pooling layer, a fully connected layer, and multiple hidden layers; the process of determining the predicted pseudo-measurement values by the pseudo-measurement model includes:

[0011] Through each hidden layer, based on the node feature matrix, the edge feature matrix, and the adjacency matrix, determine the node models of each node and the edge models of each edge in the distribution network; and determine the initial graph embedding vector based on the node models and the edge models;

[0012] Process the initial graph embedding vector through the graph pooling layer and the fully connected layer to determine the predicted pseudo-measurement values; the predicted pseudo-measurement values include node active injection power measurements and / or node reactive injection power measurements.

[0013] In a second aspect, a state estimation method based on distribution network pseudo-measurement is provided, including:

[0014] For each target node in the distribution network, collect the current multi-section map data set corresponding to the target node;

[0015] Input the current multi-section map data set into at least one pre-trained target pseudo-measurement model to determine the pseudo-measurement values corresponding to the target node through the target pseudo-measurement model;

[0016] Determine the target error according to the pseudo-measurement values and the actual load values of the target nodes;

[0017] Model the target error based on the Gaussian mixture model GMM to determine the error analysis model;

[0018] Determine the weights corresponding to each pseudo-measurement value based on the error analysis model, and perform state estimation on the distribution network through the weights;

[0019] Wherein, the target pseudo-measurement model is trained by the distribution network pseudo-measurement modeling generation method described in the first aspect above.

[0020] In a third aspect, a distribution network pseudo-measurement modeling generation device is provided, including:

[0021] An input module, configured to input the node feature matrix, the edge feature matrix, and the adjacency matrix in the multi-section historical map data set into the pseudo-measurement model of the distribution network to determine the predicted pseudo-measurement values through the pseudo-measurement model;

[0022] An update module, configured to update target parameters of the pseudo-measurement model based on the predicted pseudo-measurement values and the true pseudo-measurement values;

[0023] A model determination module, configured to repeatedly execute the steps of determining the predicted pseudo-measurement values and updating the target parameters of the pseudo-measurement model until a training stop condition is met, so as to obtain a trained pseudo-measurement model;

[0024] Wherein, the pseudo-measurement model is a graph neural network, and the pseudo-measurement model includes a graph pooling layer, a fully connected layer, and multiple hidden layers; the process of determining the predicted pseudo-measurement values through the pseudo-measurement model includes:

[0025] Through each hidden layer, based on the node feature matrix, the edge feature matrix, and the adjacency matrix, determine the node models of each node and the edge models of each edge in the distribution network; and determine an initial graph embedding vector based on the node models and the edge models;

[0026] Process the initial graph embedding vector through the graph pooling layer and the fully connected layer to determine the predicted pseudo-measurement values; the predicted pseudo-measurement values include node active injection power measurements and / or node reactive injection power measurements.

[0027] In a fourth aspect, a state estimation device based on distribution network pseudo-measurement is provided, including:

[0028] A dataset acquisition module, configured to acquire the current multi-section graph dataset corresponding to each target node in the distribution network;

[0029] A pseudo-measurement value determination module, configured to input the current multi-section graph dataset into at least one pre-trained target pseudo-measurement model, so as to determine the pseudo-measurement values corresponding to the target nodes through the target pseudo-measurement model;

[0030] A target error determination module, configured to determine a target error according to the pseudo-measurement values and the actual load values of the target nodes;

[0031] An error analysis model determination module, configured to model the target error based on a Gaussian mixture model GMM to determine an error analysis model;

[0032] A state estimation module, configured to determine weights corresponding to each pseudo-measurement value based on the error analysis model, so as to perform state estimation on the distribution network through the weights;

[0033] Wherein, the target pseudo-measurement model is trained by the distribution network pseudo-measurement modeling generation method described in the first aspect above.

[0034] In a fifth aspect, an electronic device is provided, including:

[0035] at least one processor; and,

[0036] a memory communicatively connected to the at least one processor; wherein,

[0037] the memory stores a computer program executable by the at least one processor, and when the computer program is executed by the at least one processor, the at least one processor is enabled to execute the method for generating a pseudo-measurement model of a distribution network as described in the first aspect above, or to implement the state estimation method for pseudo-measurement of a distribution network as described in the second aspect above.

[0038] In a sixth aspect, a computer-readable storage medium is provided, on which a computer program is stored, and when the program is executed by a processor, the method for generating a pseudo-measurement model of a distribution network as described in the first aspect above is implemented, or the state estimation method for pseudo-measurement of a distribution network as described in the second aspect above is implemented.

[0039] In a seventh aspect, a computer program product is provided, the computer program product includes a computer program, and when the computer program is executed by a processor, the method for generating a pseudo-measurement model of a distribution network as described in the first aspect above is implemented, or the state estimation method for pseudo-measurement of a distribution network as described in the second aspect above is implemented.

[0040] An embodiment of the present invention discloses a method, an apparatus and a device for generating a pseudo-measurement model of a distribution network, and a state estimation method. The method includes: inputting a node feature matrix, an edge feature matrix and an adjacency matrix in a multi-section historical graph dataset into a pseudo-measurement model of the distribution network to determine predicted pseudo-measurement values through the pseudo-measurement model; updating target parameters of the pseudo-measurement model based on the predicted pseudo-measurement values and true pseudo-measurement values; repeatedly executing the steps of determining the predicted pseudo-measurement values and updating the target parameters of the pseudo-measurement model until a training stop condition is satisfied, and obtaining a trained pseudo-measurement model; wherein, the pseudo-measurement model is a graph neural network, and the pseudo-measurement model includes a graph pooling layer, a fully connected layer and a plurality of hidden layers; the process of determining the predicted pseudo-measurement values through the pseudo-measurement model includes: determining a node model of each node and an edge model of each edge in the distribution network based on the node feature matrix, the edge feature matrix and the adjacency matrix through each hidden layer; and determining an initial graph embedding vector based on the node model and the edge model; processing the initial graph embedding vector through the graph pooling layer and the fully connected layer to determine the predicted pseudo-measurement values; the predicted pseudo-measurement values include node active power injection power measurement and / or node reactive power injection power measurement. The technical solution of the present invention transfers information based on the constructed node model and edge model to determine the predicted pseudo-measurement values, realizes the prediction of power injection pseudo-measurement for nodes without power measurement equipment or nodes with missing measurement data, improves the redundancy of the measurement data of the distribution network, makes up for the problem of insufficient deployment of physical sensors to a certain extent, reduces the demand for additional measurement equipment, saves costs and improves the observability of the system at the same time.

[0041] It should be understood that the content described in this part is not intended to identify the key or important features of the embodiments of the present invention, nor is it used to limit the scope of the embodiments of the present invention. Other features of the embodiments of the present invention will become easily understood through the following description. Brief Description of the Drawings

[0042] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0043] Figure 1 is a flowchart of a method for generating a pseudo-measurement model of a distribution network provided in Embodiment 1 of the present invention;

[0044] Figure 2 is a flowchart of a state estimation method based on pseudo-measurement of a distribution network provided in Embodiment 2 of the present invention;

[0045] Figure 3 It is a schematic diagram of the execution process of a state estimation method based on pseudo-measurement of a distribution network provided in the second embodiment of the present invention;

[0046] Figure 4 It is a schematic diagram of the structure of a power test system for a distribution network provided in the second embodiment of the present invention;

[0047] Figure 5 It is a distribution diagram of the active power pseudo-measurement error of nodes in the error distribution diagram after using GMM to model the pseudo-measurement provided in the second embodiment of the present invention;

[0048] Figure 6 It is a distribution diagram of the reactive power pseudo-measurement error of nodes in the error distribution diagram after using GMM to model the pseudo-measurement provided in the second embodiment of the present invention;

[0049] Figure 7 It is a schematic diagram of the structure of a device for generating a pseudo-measurement model for a distribution network provided in the third embodiment of the present invention;

[0050] Figure 8 It is a schematic diagram of the structure of a state estimation device based on pseudo-measurement of a distribution network provided in the fourth embodiment of the present invention;

[0051] Figure 9 It is a schematic diagram of the structure of an electronic device provided in the fifth embodiment of the present invention. Detailed implementation manners

[0052] In order to enable those skilled in the art to better understand the solutions of the embodiments of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments in the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the embodiments of the present invention.

[0053] It should be noted that the terms "first", "second", etc. in the specification and claims of the embodiments of the present invention are used to distinguish similar objects, and do not necessarily need to describe a specific order or sequence. It should be understood that such data can be interchanged under appropriate circumstances so that the embodiments of the present invention described here can be implemented in an order other than those illustrated or described here. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products, or devices.

[0054] Embodiment 1

[0055] Figure 1 This is a flow chart of a method for generating pseudo-measurement modeling of a distribution network provided in the first embodiment of the present invention. This embodiment is applicable to the case of pseudo-measurement modeling of a distribution network. The method can be executed by a pseudo-measurement modeling generating device for a distribution network. The pseudo-measurement modeling generating device for a distribution network can be implemented in the form of hardware and / or software. The pseudo-measurement modeling generating device for a distribution network can be configured in an electronic device, including but not limited to computers, computers, terminals, servers and other devices with data processing capabilities. Figure 1 As shown, the method includes:

[0056] S110, inputting the node feature matrix, edge feature matrix and adjacency matrix in the multi-section historical graph data set into the pseudo measurement model of the distribution network to determine the predicted pseudo measurement value through the pseudo measurement model.

[0057] In this embodiment, the multiple sections can be multiple different sections or nodes in the power system, and these nodes are used to monitor and analyze state quantities such as current and voltage in the distribution network. The node feature matrix includes the node state corresponding to each node in the distribution network; the edge feature matrix includes the edge features corresponding to each edge in the distribution network.

[0058] Following the above description, the node feature matrix, edge feature matrix and adjacency matrix in the multi-section historical graph data set can be input into the pseudo-measurement model of the distribution network to determine the predicted pseudo-measurement value. Among them, the pseudo-measurement model is a graph neural network, and graph neural networks (GNN) are a type of deep learning model for processing graph structure data. The network structure and model parameters of the graph neural network model GNN can be set, and the pseudo-measurement model includes a graph pooling layer, a fully connected layer and multiple hidden layers; illustratively, the pseudo-measurement model can have three hidden layers.

[0059] S120 . Update target parameters of the pseudo measurement model based on the predicted pseudo measurement value and the actual pseudo measurement value.

[0060] Specifically, after determining that the prediction is a measurement value, the target parameters of the pseudo-measurement model can be updated based on the predicted pseudo-measurement value and the real pseudo-measurement value. Among them, the real pseudo-measurement value can be the real measurement value corresponding to the node. The real pseudo-measurement value can obtain a theoretically simulated measurement value based on the power flow equation of the power system. The real pseudo-measurement value may include node active injection power measurement and node reactive injection power measurement. The target parameter may be a parameter in a fully connected layer, and exemplary target parameters may include parameters such as weights and biases.

[0061] Exemplarily, for an AC system, the true pseudo-measurement values may include node voltage magnitude measurements, node active power injection measurements, node reactive power injection measurements, branch active power measurements, and branch reactive power measurements. Among them, the node voltage magnitude measurement refers to the magnitude of the voltage measured at a certain node; the node active power injection measurement refers to the active power injected or consumed at a certain node; the node reactive power injection measurement refers to the reactive power injected or consumed at a certain node; the branch active power measurement refers to the active power transmitted on a certain branch (line); the branch reactive power measurement refers to the reactive power transmitted on a certain branch (line); the power flow equations for each power measurement can be expressed as:

[0062]

[0063] Where P ij and Q ij respectively represent the branch active power measurement and branch reactive power measurement of the branch connecting the i-th node and the j-th node, V i and V j respectively represent the voltage magnitudes of the i-th node and the j-th node, θ i and θ j represent the phase angles of the i-th node and the j-th node, P i and Q i respectively represent the node active power injection and reactive power injection of the i-th node, G ij and B ij respectively represent the real part and imaginary part of the admittance matrix of the branch connecting the i-th node and the j-th node, and N represents the set of nodes adjacent to node i. It should be noted that the true pseudo-measurement values are of great significance for the subsequent evaluation of the power system. At the same time, the true pseudo-measurement values can also be compared with the output of the graph neural network (predicted pseudo-measurement values). Usually, all the measurement values cannot be obtained in the distribution network. Here, the actual measurement values (nodes and branches equipped with measurement devices) are used for GNN training to obtain the pseudo-measurements of the nodes without equipment, and then compared with the theoretically simulated measurement values (true pseudo-measurement values) to verify the effect.

[0064] S130. Determine whether the pseudo-measurement model corresponding to the current iteration meets the requirements. If so, execute S140; if not, execute S110.

[0065] Specifically, after updating the target parameters of the pseudo-measurement model, it can be determined whether the pseudo-measurement model corresponding to the current iteration meets the requirements. When it meets the requirements, S150 is executed; when it does not meet the requirements, S120 is executed for multiple rounds of iterative training. Exemplarily, it can be determined whether the pseudo-measurement model corresponding to the current iteration meets the requirements by calculating the predicted pseudo-measurement value and the true pseudo-measurement value. If it does not meet the requirements, the above steps of determining the predicted pseudo-measurement value and updating the target parameters of the pseudo-measurement model are repeatedly executed until the training stop condition is satisfied, and a trained pseudo-measurement model is obtained.

[0066] S140. Obtain a trained pseudo-measurement model.

[0067] Specifically, if the pseudo-measurement model corresponding to the current iteration meets the requirements, a trained pseudo-measurement model can be obtained.

[0068] Among them, the process of determining the predicted pseudo-measurement value through the pseudo-measurement model includes:

[0069] 1) Through each hidden layer, based on the node feature matrix, edge feature matrix, and adjacency matrix, determine the node models of each node and the edge models of each edge in the distribution network; and determine the initial graph embedding vector based on the node models and edge models.

[0070] Specifically, the pseudo-measurement model includes multiple hidden layers. Exemplarily, the number of hidden layers can be 3. The activation function of each hidden layer can be a non-linear activation function. Exemplarily, the activation function of each hidden layer can be the ReLU activation function. Through each hidden layer, based on the node feature matrix, edge feature matrix, and adjacency matrix, the node models of each node and the edge models of each edge in the distribution network can be determined. Among them, the node feature matrix can include information such as the initial active injection power and initial reactive injection power; the edge feature matrix includes the active power of the head-end branch and the reactive power of the head-end branch; the adjacency matrix can be determined based on the distribution network topology structure information.

[0071] Following the above description, after determining the node model and edge model, information transfer between nodes and edges can be performed based on the node model and edge model to determine the topology structure information of the graph, and the topology structure information of the graph can be mined, and then the initial graph embedding vector can be determined.

[0072] Optionally, for any node in the distribution network, the node model of the node includes a node receiving information function, a node state function, and a node message passing function; the node receiving information function is determined by the node state at the previous moment of the node, the node states of the adjacent nodes of the node at the previous moment, and the first edge feature corresponding to the edge between the node and the adjacent node; the node state function is determined by the node state at the previous moment of the node and the node receiving information function; the node message passing function is determined by the node state at the previous moment of the node, the activation function of the fully connected layer at the previous moment, and the second edge feature corresponding to the edge between the node and the adjacent node. Exemplarily, for the i-th node, the t + 1 moment can be the current moment of the node, the t moment can be the previous moment of the node, and the expression of the node receiving information function can be expressed as:

[0073]

[0074] In the formula, can represent the information received by node i at the t + 1 moment; N(i) can represent all adjacent nodes of node i; can be the state of node i at moment t; can be the state of the adjacent node j of node i at moment t; e ij can be the first edge feature between node i and node j; can be the node message passing function at moment t, and this formula can represent that the information received by node i comes from and the state of adjacent node j at moment t and the edge feature e connecting them ij .

[0075] After the node collects the information, the node state is updated based on the node state function, and the node state function can be expressed as;

[0076]

[0077] Among them, can be the node update function, and this function can use and as inputs to generate the node state at the t + 1 moment Exemplarily, the state update function can adopt the gated recurrent unit GRU:

[0078]

[0079] The node message passing function can be expressed as:

[0080]

[0081] Among them, can be a learnable parameter matrix; can be the activation function of the fully connected layer of the node at time t; can be the state of node i at time t; can be the state of the adjacent node j of node i at time t; can be the second edge feature, and the second edge feature can be the edge feature of edge ij at time t. It should be noted that the edge feature can include the first edge feature and the second edge feature corresponding to the previous time.

[0082] For the edge between a node and its adjacent node, the edge model of the edge includes an edge information receiving function, an edge state function, and an edge message passing function; the edge information receiving function is determined by the second edge feature corresponding to the edge, the node state of the node at the previous time, and the node state of the adjacent node of the node at the previous time; the edge state function is determined by the second edge feature corresponding to the edge and the edge information receiving function; the edge message passing function is determined by the second edge feature corresponding to the edge, the node state of the node at the previous time, and the node state of the adjacent node of the node at the previous time.

[0083] Exemplarily, based on the node model of the above node i, the edge model of the edge between node i and adjacent node j includes an edge information receiving function, an edge state function, and an edge message passing function. The edge information receiving function can be expressed as:

[0084]

[0085] where, represents the information received by edge ij at time t + 1, is the edge message passing function at time t; can be the state of node i at time t; can be the state of the adjacent node j of node i at time t; can be the second edge feature; the edge state function can be expressed as:

[0086]

[0087] where, represents the state of edge ij at time t + 1, can be the second edge feature; represents the information received by edge ij at time t + 1. The edge message passing function can be expressed as:

[0088]

[0089] where, is a learnable parameter matrix, is the activation function of the fully connected layer of the edge at time t; It can be a second-side feature, which can be the state of node i at time t; which can be the state of adjacent node j of node i at time t.

[0090] 2) Process the initial graph embedding vector through the graph pooling layer and the fully connected layer to determine a predicted pseudo-measurement value; the predicted pseudo-measurement value includes a node active power injection measurement and / or a node reactive power injection measurement.

[0091] Specifically, after receiving the initial graph embedding vector, the graph pooling layer can perform pooling compression on the initial graph embedding vector to obtain a final graph embedding vector, and input the final graph embedding vector into the fully connected layer. The fully connected layer uses the Sigmoid activation function, and finally outputs the power injection prediction pseudo-measurement value of the node without a power measurement device or the node with missing measurement data. The predicted pseudo-measurement value includes a node active power injection measurement and / or a node reactive power injection measurement.

[0092] Exemplarily, the final output of the model is:

[0093]

[0094] In the formula, y represents the final output vector of the model, T represents the last time step of the message passing stage, represents the output at the last moment of the i-th node, is the fully connected layer, and its parameters can be determined through network training. G can represent the distribution network modeled as a graph. It should be noted that node i in the distribution network modeled as a graph can also include the node information of node i, and the node information can include basic information of the node, electrical parameters, equipment information, and / or connection node information, etc.

[0095] This embodiment provides a method for generating a pseudo-measurement model of a distribution network, including: inputting a node feature matrix, an edge feature matrix, and an adjacency matrix in a multi-section historical graph dataset into a pseudo-measurement model of the distribution network to determine predicted pseudo-measurement values through the pseudo-measurement model; updating target parameters of the pseudo-measurement model based on the predicted pseudo-measurement values and true pseudo-measurement values; repeatedly executing the steps of determining the predicted pseudo-measurement values and updating the target parameters of the pseudo-measurement model until a training stop condition is met, to obtain a trained pseudo-measurement model; wherein, the pseudo-measurement model is a graph neural network, and the pseudo-measurement model includes a graph pooling layer, a fully connected layer, and multiple hidden layers; the process of determining the predicted pseudo-measurement values through the pseudo-measurement model includes: through each hidden layer, determining a node model for each node and an edge model for each edge in the distribution network based on the node feature matrix, the edge feature matrix, and the adjacency matrix; and determining an initial graph embedding vector based on the node model and the edge model; processing the initial graph embedding vector through the graph pooling layer and the fully connected layer to determine the predicted pseudo-measurement values; the predicted pseudo-measurement values include node active injection power measurements and / or node reactive injection power measurements, realizing the prediction of power injection pseudo-measurements for nodes without power measurement equipment or nodes with missing measurement data, and improving the redundancy of distribution network measurement data.

[0096] As an optional implementation manner of this embodiment, before inputting the node feature matrix, the edge feature matrix, and the adjacency matrix in the multi-section historical graph dataset into the pseudo-measurement model of the distribution network, the method further includes:

[0097] 1) Determining topological structure information based on the distribution network.

[0098] Specifically, the topological structure information of the distribution network can be determined based on the distribution network, and the topological structure information can include each node (such as a substation, a switch, a load point, etc.) in the distribution network and the connection relationship between each node.

[0099] 2) Determining multi-section power flow data based on multi-time-section historical state quantities.

[0100] Specifically, multi-time-section historical state quantities are obtained from a historical database. The multi-time-section historical state quantities can be actual operation data recorded by a node in a past time period, such as current, voltage, power, etc. Multi-section power flow data can be obtained through computer simulation software. Among them, the power flow data can be the flow situation of current, voltage, power, etc. between different nodes in the power system.

[0101] 3) Adding a preset noise signal to the multi-section power flow data to determine initial historical measurement information.

[0102] It can be known that after determining the multi-time-section power flow data, a preset noise signal can be added to the multi-time-section power flow data, and the multi-time-section power flow data with the preset noise signal added is determined as the initial historical measurement information. The preset noise signal can be a pre-set noise signal. Exemplarily, the preset noise signal can be Gaussian white noise.

[0103] 4) Normalize the initial historical measurement information to determine the target historical measurement information.

[0104] It can be known that by performing normalization processing on the initial historical measurement information, the target historical measurement information can be obtained. Exemplarily, the formula for the normalization processing can be expressed as:

[0105]

[0106] where u i ' is the i-th normalized initial historical measurement information, u i is the i-th initial historical measurement information, is the minimum value in a set of data, is the maximum value in a set of data.

[0107] 5) Determine the multi-section historical graph dataset based on the topological structure information and the target historical measurement information.

[0108] Specifically, construct a multi-time-section historical graph dataset that describes node features and edge features from the topological structure information and the target historical measurement information.

[0109] As an optional implementation manner of this embodiment, after determining the multi-section historical graph dataset based on the topological structure information and the target historical measurement information, the method further includes:

[0110] 1) Divide the multi-section historical graph dataset into a training set and a test set according to a preset ratio; the training set is used to train the pseudo-measurement model; the test set is used to evaluate the performance of the trained pseudo-measurement model to determine the evaluation result.

[0111] Specifically, after determining the multi-time-section historical graph dataset, the multi-section historical graph dataset can be divided into a training set and a test set according to a preset ratio; the training set is used to train the pseudo-measurement model; the test set is used to evaluate the performance of the trained pseudo-measurement model, and thus the evaluation result can be obtained. Exemplarily, the preset ratio can be 9:1.

[0112] 2) Update the target parameters of the pseudo-measurement model based on the evaluation result.

[0113] It can be known that the target parameters of the pseudo-measurement model can be updated according to the performance evaluation results of the test set.

[0114] The technical solution of the present invention has the following beneficial technical effects:

[0115] Improve the model expression ability: The graph neural network can directly operate on graph-structured data, so it can better capture the complex relationships between nodes and edges in the distribution network. This makes the pseudo-measurement modeling more accurate and can more realistically reflect the actual operating state of the distribution network; Reduce hardware dependence: Through efficient pseudo-measurement modeling, the problem of insufficient deployment of physical sensors can be compensated to a certain extent, reducing the demand for additional measurement equipment, saving costs and improving the observability of the system.

[0116] Embodiment 2

[0117] Figure 2 FIG. 13 is a flowchart of a state estimation method based on distribution network pseudo-measurement provided in Embodiment 2 of the present invention. This embodiment is applicable to the situation of estimating the state of the distribution network. The method can be executed by a state estimation device based on distribution network pseudo-measurement. The state estimation device based on distribution network pseudo-measurement can be implemented in the form of hardware and / or software. The state estimation device based on distribution network pseudo-measurement can be configured in an electronic device, and the electronic device includes but is not limited to devices with data processing capabilities such as computers, laptops, terminals, and servers. As Figure 2 shown, the method includes:

[0118] S210. For each target node in the distribution network, collect the current multi-section map data set corresponding to the target node.

[0119] In this embodiment, the target node may be a node in the distribution network that is not equipped with a power measurement device or a node with missing measurement data. For each target node in the distribution network, the multi-section map data set corresponding to the target node can be collected, and the multi-section data set can be normalized to obtain the current multi-section map data set.

[0120] S220. Input the current multi-section map data set into at least one pre-trained target pseudo-measurement model to determine the pseudo-measurement value corresponding to the target node through the target pseudo-measurement model; wherein, the target pseudo-measurement model is trained by the distribution network pseudo-measurement modeling generation method of any embodiment of the present invention.

[0121] Specifically, input the current multi-section diagram dataset into at least one pre-trained target pseudo-measurement model to determine the pseudo-measurement value corresponding to the target node through the target pseudo-measurement model. The pseudo-measurement value is the power injection pseudo-measurement value of the node without a power measurement device or the node with missing measurement data in the current section of the system. Among them, the target pseudo-measurement model is obtained through offline training by the power distribution network pseudo-measurement modeling generation method of any embodiment of the present invention.

[0122] S230. Determine the target error according to the pseudo-measurement value and the actual load value of the target node.

[0123] It can be known that after determining the pseudo-measurement value, the target error can be determined according to the pseudo-measurement value and the actual load value of the target node. Exemplarily, the target error can be the error between the pseudo-measurement value and the actual load value.

[0124] S240. Model the target error based on the Gaussian mixture model GMM to determine the error analysis model.

[0125] In this embodiment, the Gaussian mixture model (Gaussian Mixture Model, GMM) is a probability-based model used to describe the situation where data is composed of multiple Gaussian distributions (normal distributions). Modeling the target error based on the Gaussian mixture model GMM generally includes: collecting data of the target error, initializing the parameters of the GMM, including the mean, covariance matrix, and mixing weight of each Gaussian component. These parameters can be obtained through random initialization or preliminary estimation based on the data. Use the Expectation-Maximization (EM) algorithm to iteratively update the parameters of the GMM to maximize the likelihood of the data until the model parameters converge (for example, the parameter change is less than a certain threshold) or reach the maximum number of iterations. The GMM is composed of multiple Gaussian distribution functions weighted and can smoothly approximate probability distributions of any shape. Then there is:

[0126]

[0127] In the formula: f(zγ) can represent the marginal density function; z is the pseudo-measurement; M is the order of the model; w i is the weight of the i-th Gaussian component and satisfies w i >0, u i can be the mean of the i-th Gaussian component; can be the variance of the i-th Gaussian component; γ can be the parameter of the Gaussian mixture model, and the parameter is generally solved by the maximum expectation algorithm.

[0128] Model the pseudo-measurement error output by the graph neural network with M Gaussian components. The correlation between this error and the j-th Gaussian component is represented by the marginal density function f(j|e i ,γ):

[0129]

[0130] Wherein: can represent a normal distribution, u j and u k are the means of the j-th and k-th Gaussian components; and are the variances of the j-th and k-th Gaussian components, w j and w k are the weights of the j-th and k-th Gaussian components. According to the above description, after modeling by GMM, the means, covariance matrices, and mixing weights of the Gaussian components obtained can be used to construct an error analysis model. This model can be used for: Quantifying error characteristics: By analyzing the means and variances of each Gaussian component, understand the central position and dispersion degree of the error; Identifying abnormal errors: By calculating the probability density values of data points in the GMM, identify low-probability abnormal errors; Uncertainty estimation: By analyzing the covariance matrix of each Gaussian component, provide an uncertainty estimate of the error.

[0131] S250. Determine the weights corresponding to each pseudo-measurement value based on the error analysis model to perform state estimation on the distribution network through the weights.

[0132] It can be known that the weights corresponding to each pseudo-measurement value can be determined based on the error analysis model. The weights can reflect the reliability and importance of each pseudo-measurement value. In the state estimation of the distribution network, using these weights can improve the accuracy and reliability of the state estimation.

[0133] This embodiment provides a state estimation method based on pseudo-measurement of a distribution network, including: for each target node in the distribution network, collecting the current multi-section map data set corresponding to the target node; inputting the current multi-section map data set into at least one pre-trained target pseudo-measurement model to determine the pseudo-measurement value corresponding to the target node through the target pseudo-measurement model; determining the target error according to the pseudo-measurement value and the actual load value of the target node; modeling the target error based on the Gaussian mixture model GMM to determine the error analysis model; determining the weights corresponding to each pseudo-measurement value based on the error analysis model to perform state estimation on the distribution network through the weights; wherein, the target pseudo-measurement model is trained by the distribution network pseudo-measurement modeling generation method described in any embodiment of the present invention. The technical solution in this embodiment can predict the power injection pseudo-measurement of nodes without power measurement equipment or nodes with missing measurement data, improve the redundancy of the measurement data of the distribution network, and improve the state estimation accuracy. At the same time, using GNN can provide relatively accurate power injection pseudo-measurement values for nodes without power measurement equipment or nodes with missing measurement data in the distribution network, provide data support for subsequent state estimation, improve the accuracy of state estimation, and provide technical support for subsequent safety assessment and optimal operation.

[0134] Figure 3 FIG. is a schematic diagram of the execution process of a state estimation method based on pseudo-measurement of a distribution network provided in this embodiment, as Figure 3 shown. First, during offline training, a multi-section historical map data set (map data set) can be constructed based on historical measurement data (target historical measurement information) and topological information (topological structure information). Train the graph neural network pseudo-measurement model GNN based on the map data set. During the online test process, real-time measurement data and topological information can be used to construct a map data set (current multi-section map data set), and input it into the trained GNN model. Use the GNN model to predict the pseudo-measurement value. After obtaining the pseudo-measurement result, the target error can be determined based on the pseudo-measurement value and the actual load value of the node, and GMM can be used for error modeling. Furthermore, the weights corresponding to the pseudo-measurements can be obtained through the error analysis model, and then state estimation can be performed on the distribution network.

[0135] This embodiment also provides an application example of a state estimation method based on pseudo-measurement of a distribution network. Figure 4 FIG. is a schematic diagram of the structure of a power test system for a distribution network provided in this embodiment, as Figure 4 shown. The power test system of the distribution network is a 33-node three-phase balance. The system measurement configuration is as follows:

[0136] Table 1 Measurement Configuration of 33-Node Test System

[0137]

[0138] Train a graph neural network pseudo-measurement model based on a multi-section historical graph dataset. Obtain the power flow results of 2000 time sections through experimental simulation. Then introduce corresponding Gaussian noise into the real power flow results to simulate measurement data, and divide them into a training set and a test set according to a ratio of 9:1. Finally, input real-time measurement data into the trained model to obtain pseudo-measurement values under time sections, and then use the Gaussian mixture model for the pseudo-measurement error distribution. The present invention uses pseudo-measurement data under 100 time sections for testing, normalizes the real power injection and the pseudo-measurement obtained by the GNN, and then processes the results of the pseudo-measurement error distribution after normalization by the GMM. Figure 5 It is the active power pseudo-measurement error distribution diagram of the nodes in the error distribution diagram of the pseudo-measurement after modeling with the GMM, as Figure 5 shown. The dotted lines (Gaussian Component1, Gaussian Component2, Gaussian Component3, Gaussian Component4) represent the differences and the frequencies of the differences in the active power pseudo-measurement errors of the nodes under different Gaussian combinations; the red solid line (Total GMM Fit) represents the frequency of the differences in the active power pseudo-measurement errors of the nodes under different Gaussian combinations after summarization; the bar (error data) represents the frequency of the errors after summarization of each error. Figure 6 It is the reactive power pseudo-measurement error distribution diagram of the nodes in the error distribution diagram of the pseudo-measurement after modeling with the GMM. As Figure 6 shown. The dotted lines (Gaussian Component1, Gaussian Component2, Gaussian Component3, Gaussian Component4) represent the differences and the frequencies of the differences in the reactive power pseudo-measurement errors of the nodes under different Gaussian combinations; the red solid line (Total GMM Fit) represents the frequency of the differences in the reactive power pseudo-measurement errors of the nodes under different Gaussian combinations after summarization; the bar (error data) represents the frequency of the errors after summarization of each error. It can be seen that whether it is the active power pseudo-measurement error of the nodes or the reactive power pseudo-measurement error of the nodes, the normalized error values (abscissa) of both are basically in a very small range, indicating that there is a very small difference between the output of the GNN and the real value, and the pseudo-measurement data has high accuracy.

[0139] Embodiment 3

[0140] Figure 7 It is a schematic structural diagram of a distribution network pseudo-measurement modeling generation device provided in Embodiment 3 of the present invention; as Figure 7As shown in the figure, the device includes: a predicted pseudo-measurement value determination module 310, an update module 320, and a model determination module 330.

[0141] Among them, the predicted pseudo-measurement value determination module 310 is configured to input the node feature matrix, edge feature matrix, and adjacency matrix in the multi-section historical map dataset into the pseudo-measurement model of the distribution network, so as to determine the predicted pseudo-measurement value through the pseudo-measurement model;

[0142] The update module 320 is configured to update the target parameters of the pseudo-measurement model based on the predicted pseudo-measurement value and the true pseudo-measurement value;

[0143] The model determination module 330 is configured to repeatedly execute the steps of determining the predicted pseudo-measurement value and updating the target parameters of the pseudo-measurement model until the training stop condition is met, and obtain the trained pseudo-measurement model;

[0144] Among them, the pseudo-measurement model is a graph neural network, and the pseudo-measurement model includes a graph pooling layer, a fully connected layer, and multiple hidden layers; the process of determining the predicted pseudo-measurement value through the pseudo-measurement model includes:

[0145] Through each hidden layer, based on the node feature matrix, the edge feature matrix, and the adjacency matrix, determine the node model of each node and the edge model of each edge in the distribution network; and determine the initial graph embedding vector based on the node model and the edge model;

[0146] Process the initial graph embedding vector through the graph pooling layer and the fully connected layer to determine the predicted pseudo-measurement value; the predicted pseudo-measurement value includes node active power injection measurement and / or node reactive power injection measurement.

[0147] Embodiment 3 of the present invention provides a device for generating a pseudo-measurement model of a distribution network, which realizes the prediction of power injection pseudo-measurement for nodes without power measurement equipment or nodes with missing measurement data, and improves the redundancy of the measurement data of the distribution network.

[0148] Furthermore, the device further includes:

[0149] The node feature matrix includes the node states corresponding to each node in the distribution network; the edge feature matrix includes the edge features corresponding to each edge in the distribution network; the edge features may include first edge features and second edge features corresponding to the previous moment;

[0150] For any node in the distribution network, the node model of the node includes a node receiving information function, a node state function, and a node message passing function; the node receiving information function is determined by the node state of the node at the previous moment, the node states of the adjacent nodes of the node at the previous moment, and the first edge feature corresponding to the edge between the node and the adjacent node; the node state function is determined by the node state of the node at the previous moment and the node receiving information function; the node message passing function is determined by the node state of the node at the previous moment, the activation function of the fully connected layer at the previous moment, and the second edge feature corresponding to the edge between the node and the adjacent node;

[0151] For the edge between the node and the adjacent node, the edge model of the edge includes an edge receiving information function, an edge state function, and an edge message passing function;

[0152] The edge receiving information function is determined by the second edge feature corresponding to the edge, the node state of the node at the previous moment, and the node states of the adjacent nodes of the node at the previous moment; the edge state function is determined by the second edge feature corresponding to the edge and the edge receiving information function; the edge message passing function is determined by the second edge feature corresponding to the edge, the node state of the node at the previous moment, and the node states of the adjacent nodes of the node at the previous moment.

[0153] Furthermore, the device further includes:

[0154] A topological structure information determination module, configured to determine topological structure information based on the distribution network;

[0155] A multi-time-section power flow data determination module, configured to determine multi-time-section power flow data based on multi-time-section historical state quantities;

[0156] An initial historical measurement information determination module, configured to add a preset noise signal based on the multi-time-section power flow data to determine initial historical measurement information;

[0157] A target historical measurement information determination module, configured to normalize the initial historical measurement information to determine the target historical measurement information;

[0158] A multi-section historical graph data set determination module, configured to determine the multi-section historical graph data set based on the topological structure information and the target historical measurement information.

[0159] Furthermore, the device further includes:

[0160] A splitting module, configured to divide the multi-section historical graph data set into a training set and a test set according to a preset ratio; the training set is used to train a pseudo-measurement model; the test set is used to evaluate the performance of the trained pseudo-measurement model;

[0161] A target parameter update module, configured to update the target parameters of the pseudo-measurement model based on the evaluation result.

[0162] The power distribution network pseudo-measurement modeling generation device provided by the embodiments of the present invention can execute the power distribution network pseudo-measurement modeling generation method provided by any embodiment of the embodiments of the present invention, and has the corresponding functional modules and beneficial effects for executing the method.

[0163] Embodiment 4

[0164] Figure 8 is a structural schematic diagram of a state estimation device based on power distribution network pseudo-measurement provided by Embodiment 4 of the present invention; as Figure 8 shown, the device includes: a data set acquisition module 410, a pseudo-measurement value determination module 420, a target error determination module 430, an error analysis model determination module 440, and a state estimation module 450.

[0165] Among them, the data set acquisition module 410 is configured to collect the current multi-section diagram data set corresponding to each target node in the power distribution network;

[0166] The pseudo-measurement value determination module 420 is configured to input the current multi-section diagram data set into at least one pre-trained target pseudo-measurement model, so as to determine the pseudo-measurement value corresponding to the target node through the target pseudo-measurement model;

[0167] The target error determination module 430 is configured to determine a target error according to the pseudo-measurement value and the actual load value of the target node;

[0168] The error analysis model determination module 440 is configured to model the target error based on the Gaussian mixture model GMM to determine an error analysis model;

[0169] The state estimation module 450 is configured to determine the weights corresponding to each pseudo-measurement value based on the error analysis model, so as to perform state estimation on the power distribution network through the weights;

[0170] Among them, the target pseudo-measurement model is trained by the power distribution network pseudo-measurement modeling generation method described in any embodiment of the embodiments of the present invention.

[0171] Embodiment 4 of the present invention provides a state estimation device based on power distribution network pseudo-measurement, which realizes accurate estimation of the power distribution network state.

[0172] The state estimation device based on power distribution network pseudo-measurement provided by the embodiments of the present invention can execute the state estimation method based on power distribution network pseudo-measurement provided by any embodiment of the embodiments of the present invention, and has the corresponding functional modules and beneficial effects for executing the method.

[0173] Example 5

[0174] Figure 9 FIG. shows a schematic structural diagram of an electronic device 10 that can be used to implement the present invention. The electronic device is intended to represent various forms of digital computers, such as, for example, laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The components shown in the present invention, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the embodiments of the present invention described and / or claimed herein.

[0175] As Figure 9 shown, the electronic device 10 includes at least one processor 11, and a memory communicatively connected to the at least one processor 11, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc. The memory stores a computer program executable by the at least one processor. The processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 into the random access memory (RAM) 13. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0176] A plurality of components in the electronic device 10 are connected to the I / O interface 15, including: an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a magnetic disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0177] The processor 11 can be various general-purpose and / or special-purpose processing components having processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microprocessor, etc. The processor 11 executes the various methods and processes described in the present invention, such as a method for generating a pseudo-measurement model of a distribution network and a state estimation method based on pseudo-measurements of a distribution network.

[0178] In some embodiments, the method for generating pseudo-measurement modeling of a distribution network and the state estimation method based on pseudo-measurements of a distribution network can be implemented as a computer program, which is tangibly embodied in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed onto the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the above-described method for generating pseudo-measurement modeling of a distribution network and the state estimation method based on pseudo-measurements of a distribution network can be performed. Alternatively, in other embodiments, the processor 11 can be configured to execute the method for generating pseudo-measurement modeling of a distribution network and the state estimation method based on pseudo-measurements of a distribution network by any other suitable means (e.g., by means of firmware).

[0179] The various embodiments of the systems and techniques described above in the present invention can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on a chip (SOCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include: being implemented in one or more computer programs, which can be executed and / or interpreted on a programmable system including at least one programmable processor, the programmable processor can be a dedicated or general-purpose programmable processor, and can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit the data and instructions to the storage system, the at least one input device, and the at least one output device.

[0180] The computer program for implementing the method of the embodiments of the present invention can be written in any combination of one or more programming languages. These computer programs can be provided to the processor of a general-purpose computer, a dedicated computer, or other programmable data processing devices, such that when the computer programs are executed by the processor, the functions / operations specified in the flowcharts and / or block diagrams are implemented. The computer programs can be executed entirely on the machine, partially on the machine, as a stand-alone software package partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0181] In the context of the embodiments 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, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. Alternatively, the computer-readable storage medium may be a machine-readable signal medium. More specific examples of the 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.

[0182] To provide for interaction with a user, the systems and techniques described herein 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 the user; and a keyboard and a pointing device (e.g., a mouse or a trackball) by which the user can provide input to the electronic device. Other kinds of devices can also be used to provide for interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0183] The systems and techniques described herein can be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer having a graphical user interface or a web browser through which the user can interact with an implementation of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend 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 network (LAN), wide area network (WAN), blockchain network, and the Internet.

[0184] A computing system may include a client and a server. The client and the server are generally far from each other and usually interact via a communication network. The relationship between the client and the server is created by computer programs running on respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or a cloud host, which is a host product in the cloud computing service system, solving the defects of difficult management and weak business scalability existing in traditional physical hosts and VPS services.

[0185] It should be understood that various forms of the processes shown above can be used, with steps reordered, added, or deleted. For example, the steps recited in the embodiments of the present invention can be executed in parallel, sequentially, or in a different order, as long as the desired results of the technical solutions of the embodiments of the present invention can be achieved, and no limitations are imposed herein.

[0186] The above specific embodiments do not constitute a limitation on the protection scope of the embodiments of the present invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent replacements, and improvements made within the spirit and principles of the embodiments of the present invention shall be included within the protection scope of the embodiments of the present invention.

[0187] The embodiments of the present invention also provide a computer program product, including a computer program and / or instructions, which, when executed by a processor, implement the distribution network pseudo-measurement modeling generation method and the state estimation method based on the distribution network pseudo-measurement provided in any embodiment of the present application.

[0188] In the process of implementing the computer program product, computer program code for performing the operations of the embodiments of the present invention can be written in one or more programming languages or combinations thereof. The programming languages include object-oriented programming languages such as Java, Smalltalk, C++, and also include conventional procedural programming languages such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, executed as an independent software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer through any type of network - including a local area network (LAN) or a wide area network (WAN) - or can be connected to an external computer (for example, by using an Internet service provider to connect through the Internet).

[0189] Note that the above is only the preferred embodiment of the embodiments of the present invention and the applied technical principles. Those skilled in the art will understand that the embodiments of the present invention are not limited to the specific embodiments here, and various obvious changes, re-adjustments and substitutions can be made by those skilled in the art without departing from the protection scope of the embodiments of the present invention. Therefore, although the embodiments of the present invention have been described in more detail through the above embodiments, the embodiments of the present invention are not limited to the above embodiments only. Without departing from the concept of the embodiments of the present invention, more other equivalent embodiments can be included, and the scope of the embodiments of the present invention is determined by the scope of the appended claims.

Claims

1. A method for generating a pseudo-measurement model of a distribution network, characterized in that Including: Inputting the node feature matrix, edge feature matrix, and adjacency matrix in the multi-section historical graph dataset into the pseudo-measurement model of the distribution network to determine the predicted pseudo-measurement values through the pseudo-measurement model; Updating the target parameters of the pseudo-measurement model based on the predicted pseudo-measurement values and the true pseudo-measurement values; Repeatedly executing the steps of determining the predicted pseudo-measurement values and updating the target parameters of the pseudo-measurement model until the training stop condition is met, and obtaining the trained pseudo-measurement model; Wherein, the pseudo-measurement model is a graph neural network, and the pseudo-measurement model includes a graph pooling layer, a fully connected layer, and multiple hidden layers; the process of determining the predicted pseudo-measurement values through the pseudo-measurement model includes: Through each hidden layer, determining the node models of each node and the edge models of each edge in the distribution network based on the node feature matrix, the edge feature matrix, and the adjacency matrix; and determining the initial graph embedding vector based on the node model and the edge model; Processing the initial graph embedding vector through the graph pooling layer and the fully connected layer to determine the predicted pseudo-measurement values; the predicted pseudo-measurement values include node active injection power measurement and / or node reactive injection power measurement.

2. The method according to claim 1, wherein The node feature matrix includes the node states corresponding to each node in the distribution network; the edge feature matrix includes the edge features corresponding to each edge in the distribution network; the edge features may include first edge features and second edge features corresponding to the previous moment; For any node in the distribution network, the node model of the node includes a node receiving information function, a node state function, and a node message passing function; The node receiving information function is determined by the node state of the node at the previous moment, the node states of the adjacent nodes of the node at the previous moment, and the first edge features corresponding to the edges between the node and the adjacent nodes; The node state function is determined by the node state of the node at the previous moment and the node receiving information function; The node message passing function is determined by the node state of the node at the previous moment, the activation function of the fully connected layer at the previous moment, and the second edge features corresponding to the edges between the node and the adjacent nodes; For the edge between the node and the adjacent node, the edge model of the edge includes an edge receiving information function, an edge state function, and an edge message passing function; The edge receiving information function is determined by the second edge features corresponding to the edge, the node state of the node at the previous moment, and the node states of the adjacent nodes of the node at the previous moment; The edge state function is determined by the second edge features corresponding to the edge and the edge receiving information function; The edge message passing function is determined by the second edge features corresponding to the edge, the node state of the node at the previous moment, and the node states of the adjacent nodes of the node at the previous moment.

3. The method according to claim 1, characterized in that Before inputting the node feature matrix, edge feature matrix, and adjacency matrix in the multi-section historical graph dataset into the pseudo-measurement model of the distribution network, the method further includes: Determining the topological structure information based on the distribution network; Determining the multi-time-section power flow data based on the multi-time-section historical state quantities; Add a preset noise signal to the multi-time-section power flow data to determine the initial historical measurement information; Normalize the initial historical measurement information to determine the target historical measurement information; Determine the multi-section historical graph dataset based on the topological structure information and the target historical measurement information.

4. The method according to claim 3, characterized in that, After determining the multi-section historical graph dataset based on the topological structure information and the target historical measurement information, the method further includes: Divide the multi-section historical graph dataset into a training set and a test set according to a preset ratio; the training set is used to train the pseudo-measurement model; the test set is used to evaluate the performance of the trained pseudo-measurement model to determine the evaluation result; Update the target parameters of the pseudo-measurement model based on the evaluation result.

5. A state estimation method based on pseudo measurements of a distribution network, characterized in that, Includes: For each target node in the distribution network, collect the corresponding current multi-section graph dataset of the target node; Input the current multi-section graph dataset into at least one pre-trained target pseudo-measurement model to determine the pseudo-measurement value corresponding to the target node through the target pseudo-measurement model; Determine the target error according to the pseudo-measurement value and the actual load value of the target node; Model the target error based on the Gaussian mixture model GMM to determine the error analysis model; Determine the weights corresponding to each pseudo-measurement value based on the error analysis model to perform state estimation on the distribution network through the weights; Wherein, the target pseudo-measurement model is trained by the distribution network pseudo-measurement modeling generation method according to any one of claims 1-4.

6. A pseudo-measurement modeling generation device for a distribution network, characterized in that, Includes: A predicted pseudo-measurement value determination module, configured to input the node feature matrix, edge feature matrix, and adjacency matrix in the multi-section historical graph dataset into the pseudo-measurement model of the distribution network to determine the predicted pseudo-measurement value through the pseudo-measurement model; An update module, configured to update the target parameters of the pseudo-measurement model based on the predicted pseudo-measurement value and the true pseudo-measurement value; A model determination module, configured to repeatedly execute the steps of determining the predicted pseudo-measurement value and updating the target parameters of the pseudo-measurement model until the training stop condition is met, and obtain the trained pseudo-measurement model; Wherein, the pseudo-measurement model is a graph neural network, and the pseudo-measurement model includes a graph pooling layer, a fully connected layer, and multiple hidden layers; the process of determining the predicted pseudo-measurement value through the pseudo-measurement model includes: Through each hidden layer, determine the node model of each node and the edge model of each edge in the distribution network based on the node feature matrix, the edge feature matrix, and the adjacency matrix; and determine the initial graph embedding vector based on the node model and the edge model; Process the initial graph embedding vector through the graph pooling layer and the fully connected layer to determine the predicted pseudo-measurement value; the predicted pseudo-measurement value includes node active injection power measurement and / or node reactive injection power measurement.

7. A state estimation device based on pseudo-measurements of a distribution network, characterized in that, Includes: A dataset collection module, configured to, for each target node in the distribution network, collect the corresponding current multi-section graph dataset of the target node; A pseudo-measurement value determination module, configured to input the current multi-section diagram data set into at least one pre-trained target pseudo-measurement model, so as to determine the pseudo-measurement value corresponding to the target node through the target pseudo-measurement model; A target error determination module, configured to determine a target error according to the pseudo-measurement value and the actual load value of the target node; An error analysis model determination module, configured to model the target error based on the Gaussian mixture model GMM to determine an error analysis model; A state estimation module, configured to determine the weights corresponding to each pseudo-measurement value based on the error analysis model, so as to perform state estimation on the distribution network through the weights; Wherein, the target pseudo-measurement model is trained by the distribution network pseudo-measurement modeling generation method according to any one of claims 1-4.

8. An electronic device, characterized in that, Comprising: At least one processor; And, A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor, so that the at least one processor can execute the distribution network pseudo-measurement modeling generation method according to any one of claims 1-4, or implement the state estimation method of the distribution network pseudo-measurement according to claim 5.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the distribution network pseudo-measurement modeling generation method according to any one of claims 1-4, or implements the state estimation method of the distribution network pseudo-measurement according to claim 5.

10. A computer program product, characterized in that, The computer program product includes a computer program, and when the computer program is executed by the processor, it implements the distribution network pseudo-measurement modeling generation method according to any one of claims 1-4, or implements the state estimation method of the distribution network pseudo-measurement according to claim 5.

Images