Community partition-based power grid coarsening model construction

The coarse grid model is constructed through the community partitioning method, which solves the problems of unreasonable grid decomposition and insufficient electrical characteristics retention in the existing technology, and realizes efficient analysis and optimization of large power grid systems.

CN119004720APending Publication Date: 2024-11-22GUANGDONG POWER GRID CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411095740.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-12
Publication Date
2024-11-22

AI Technical Summary

Technical Problem

The existing grid roughened model construction technology cannot fully and reasonably decompose large power grids according to the grid structure, and cannot retain key electrical characteristics, resulting in low analysis efficiency and insufficient accuracy.

Method used

The community partitioning method is adopted, and the grid partition is constructed by building a power grid network model, and the target community partition is obtained. The admission matrix and retaining node are used for modeling and optimization, and a roughened grid model is generated to retain key electrical characteristics.

Benefits of technology

It realizes efficient, accurate and flexible analysis of large-scale power grid systems, and improves the optimization capabilities of power grid planning, operation and control.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119004720A_ABST
    Figure CN119004720A_ABST
Patent Text Reader

Abstract

The invention relates to a community partitioning-based power grid coarsening model construction method and device, and relates to the technical field of algorithm models, and the method comprises the steps: determining a to-be-partitioned power system, constructing a power grid network model based on power grid parameters corresponding to the to-be-partitioned power system, carrying out the power grid partitioning of the power grid network model through a preset algorithm, and obtaining a power grid coarsening model; the method comprises the steps of obtaining a target community partition, performing matrix construction based on at least two to-be-analyzed nodes included in the target community partition and connection information between the to-be-analyzed nodes to obtain an admittance matrix, reserved nodes and to-be-eliminated nodes, and performing modeling optimization processing based on the reserved nodes, the admittance matrix and the to-be-eliminated nodes included in the admittance matrix to obtain a to-be-eliminated node. Through reasonable decomposition of the power grid structure, modeling of the power grid coarsening model is realized, key electrical characteristics can be retained, and researchers can conveniently carry out efficient, accurate and flexible analysis on a large-scale power grid system.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the technical field of algorithm models, and in particular to the construction of a coarsened power grid model based on community partitioning. Background Art

[0002] At present, under the background of global energy transformation, my country is committed to achieving the goals of "carbon neutrality and carbon peak", and actively promoting the use of renewable energy such as wind power and solar energy to reduce greenhouse gas emissions. However, this transformation has also brought challenges to the power system, especially the instability of distributed intermittent energy and the diversification of user-side loads, which has led to the problem of spatial and temporal imbalance between power generation and load in the power system, which is specifically manifested in the increase of complexity and uncertainty in the power system. The maturity of energy storage technology has made it possible to solve these problems. However, the coordinated analysis of source, grid, load and storage on a large scale is time-consuming, labor-intensive and inefficient, so it is necessary to conduct coarse modeling of the large power grid, use computing resources on key nodes and sections that researchers are concerned about, and improve the efficiency of large power grid analysis.

[0003] The existing grid coarse-grained model construction technology mainly focuses on the equivalence of low voltage levels to high voltage levels or the zoning equivalence based on jurisdictional divisions. It relies on the experience of practitioners and cannot comprehensively and reasonably decompose large grids according to the grid structure, nor can it retain key electrical characteristics. Summary of the invention

[0004] The present application provides a method and device for constructing a coarsening model of a power grid based on community partitioning. By reasonably decomposing the structure, the coarsening model of the power grid is modeled, and key electrical characteristics can be retained, so that researchers can determine the retained nodes and line electrical characteristics according to research needs to achieve efficient, accurate and flexible analysis of large-scale power grid systems, which is helpful for large-scale power grid analysis such as optimization of power grid planning, operation and control, and solves the problem that the existing power grid coarsening model construction technology cannot comprehensively and reasonably decompose the large power grid according to the power grid structure and cannot retain key electrical characteristics.

[0005] In a first aspect, the present application provides a method for constructing a coarse-grained model of a power grid based on community partitioning, comprising:

[0006] Determine the power system to be partitioned, and construct a power grid network model based on the power grid parameters corresponding to the power system to be partitioned;

[0007] By using a preset algorithm, the power grid network model is partitioned to obtain a target community partition, wherein the target community partition includes at least two nodes to be analyzed and connection information between the nodes to be analyzed;

[0008] Performing matrix construction based on the connection information and the nodes to be analyzed to obtain an admittance matrix and retained nodes, wherein the admittance matrix includes the nodes to be eliminated;

[0009] Modeling optimization processing is performed based on the retained nodes, the admittance matrix and the nodes to be eliminated to obtain a coarsened model of the power grid.

[0010] Optionally, the step of constructing a power grid network model based on power grid parameters corresponding to the power system to be partitioned includes:

[0011] Constructing an initial unweighted network graph based on the power system to be analyzed, wherein the initial unweighted network graph includes at least two power grid nodes and connection relationships between the power grid nodes;

[0012] Performing an intimacy analysis based on the at least two power grid nodes and a connection relationship corresponding to the at least two power grid nodes to obtain weight information;

[0013] The initial unweighted network graph is updated by using the weight information to obtain a target unweighted network graph;

[0014] A power grid network model is constructed based on the target unweighted network graph.

[0015] Optionally, performing intimacy analysis according to the at least two power grid nodes in combination with a connection relationship corresponding to the at least two power grid nodes to obtain weight information includes:

[0016] Calculating a reactive voltage sensitivity matrix L between the power grid nodes;

[0017] According to the formula Calculate the weight information between power grid nodes;

[0018] Among them, ΔQ i is the reactive power unbalance vector injected into the grid node i, U i is the voltage amplitude of grid node i, U j is the voltage amplitude of grid node j, ΔU i is the amplitude change vector of grid node i, ΔU j is the amplitude change vector of grid node j, θ ij is the voltage phase difference between grid node i and grid node j, G ij is the real part of the corresponding element of the grid node admittance matrix, B ij is the imaginary part of the corresponding element of the grid node admittance matrix, L ij is the weight of the edge between grid node i and grid node j.

[0019] Optionally, performing grid partitioning on the grid network model to obtain target community partitions includes:

[0020] Performing an initial division on all grid nodes in the grid network model to obtain at least two initial grid communities, wherein the initial grid community includes at least one of the grid nodes;

[0021] Randomly sorting the power grid nodes to obtain a community traversal table, wherein the community traversal table includes nodes to be traversed, and the nodes to be traversed correspond to the power grid nodes one by one;

[0022] Performing gain analysis on each of the nodes to be traversed to obtain gain information, wherein the gain information is used to characterize the modularity gain of the node to be traversed from the current initial power grid community to the neighboring power grid community;

[0023] Perform secondary division on the power grid nodes corresponding to the to-be-traversed nodes according to the gain information to obtain secondary division communities;

[0024] The secondary partitioned communities are optimized by a preset simulated annealing algorithm to obtain target community partitions.

[0025] Optionally, performing gain analysis on each of the nodes to be traversed to obtain gain information includes:

[0026] according to Calculate the modularity Q, which is used to characterize the quality of the initial power grid community; visit each power grid node in turn, and calculate the modularity gain ΔQ of power grid node i leaving the current community d and joining the neighbor community c i_fd2c ; Where m is the sum of all value information in the network, 2m is the sum of all grid node degrees, W ij is the weight from grid node i to grid node j, d i is the degree of grid node i, δ(a,b) is the judgment symbol, which is used to characterize the expected value of the weight of the grid node and the edge connecting the grid node;

[0027] The secondary division of the grid nodes corresponding to the to-be-traversed nodes according to the gain information to obtain the secondary division community includes: visiting each grid node i in turn, determining the modularity gain ΔQ i_fd2c Is it greater than zero? If the modularity gain ΔQ i_fd2c If it is greater than zero, grid node i is added from community d to neighbor community c;

[0028] Optionally, the secondary partitioned communities are optimized by a preset simulated annealing algorithm to obtain a target community partition, including: taking the secondary partitioned communities as equivalent nodes and combining the connection relationships between the secondary partitioned communities to generate a community sequence; traversing all communities in the community sequence to determine the energy change between the current community and each neighboring community; determining the number of community mergers based on the energy change; determining a merger stop threshold; if the number of community mergers is less than the merger stop threshold, performing a community merger on the current community and the neighboring community to obtain the target community partition.

[0029] Optionally, constructing a matrix based on the connection information and the nodes to be analyzed to obtain an admittance matrix and reserved nodes includes:

[0030] Analyzing the target community partition according to the equivalent requirements to obtain information of retained nodes and to-be-eliminated nodes, wherein the to-be-eliminated node information includes connection relationships between the to-be-eliminated nodes and connection relationships between the to-be-eliminated nodes and the retained nodes, and the retained nodes include at least one of boundary nodes, scene setting nodes, and partition dominant nodes;

[0031] The mutual admittance is performed based on the information of the nodes to be eliminated to generate the admittance matrix.

[0032] Optionally, the connection relationship includes a connection relationship of a line model and a connection relationship of a transformer model, and the mutual admittance based on the information of the nodes to be eliminated to generate the admittance matrix includes:

[0033] According to Y i =Y i_load +Y i_shunt +Y i_line +Y i_tran Calculate the self-admittance according to Y ij Calculate the mutual admittance and get the admittance matrix, where Y i_load is the self-admittance of the node load to be eliminated, Y i_shunt To compensate for the self-admittance of the capacitor, Y i_line is the connection relationship self-admittance of the line model, Y i_tran is the connection relationship self-admittance of the transformer model.

[0034] Optionally, the performing modeling optimization processing based on the retained nodes, the admittance matrix and the nodes to be eliminated to obtain a power grid coarsening model includes:

[0035] Eliminating the to-be-eliminated nodes from the admittance matrix using Gaussian elimination to obtain an equivalent system admittance matrix;

[0036] The equivalent system admittance matrix is ​​concatenated with the reserved nodes to obtain the coarsened model of the power grid.

[0037] Optionally, the retained nodes include boundary nodes, and the Gaussian elimination is used to eliminate the nodes to be eliminated from the admittance matrix to obtain an equivalent system admittance matrix, including:

[0038] according to Calculate the equivalent system admittance matrix, where Y ee4 is the mutual admittance between the boundary node and the node to be eliminated, Y et is the mutual admittance between the boundary node and the internal node, is the mutual admittance between internal nodes, Y te is the mutual admittance between the internal node and the boundary node;

[0039] Optionally, the step of systematically splicing the equivalent system admittance matrix with the power grid parameters to obtain the power grid coarsening model includes: adding each row of the equivalent system admittance matrix, taking the added value as the admittance to ground of the retained node corresponding to the row, adding the original admittance to ground of the node and updating the admittance to ground of the retained node; and Generate additional ground admittance Y of boundary node i i ;according to Generate an equivalent power grid system, where P iload_new is the active power load of the boundary node i of the equivalent power grid system, Q iload_new is the reactive power load of node i at the boundary of the equivalent power grid system, P iload is the original active power load of boundary node i, Q iload is the original reactive power load of boundary node i.

[0040] In a second aspect, the present application provides a device for constructing a coarsened model of a power grid based on community partitioning, comprising:

[0041] A power grid network model building module, used to determine the power system to be partitioned, and build a power grid network model based on the power grid parameters corresponding to the power system to be partitioned;

[0042] A power grid partitioning module, used to partition the power grid network model by a preset algorithm to obtain a target community partition, wherein the target community partition includes at least two nodes to be analyzed and connection information between the nodes to be analyzed;

[0043] A matrix construction module, used to construct a matrix based on the connection information and the nodes to be analyzed to obtain an admittance matrix and retained nodes, wherein the admittance matrix includes the nodes to be eliminated;

[0044] The modeling optimization module is used to perform modeling optimization processing based on the retained nodes, the admittance matrix and the nodes to be eliminated to obtain a coarsened model of the power grid.

[0045] In summary, the embodiment of the present application determines the power system to be partitioned, and constructs a power grid network model based on the power grid parameters corresponding to the power system to be partitioned. The power grid network model is partitioned by a preset algorithm to obtain a target community partition, and the connection information and nodes to be analyzed contained in the target community partition are used to construct a matrix to obtain an admittance matrix. Modeling optimization processing is performed based on the admittance matrix containing retained nodes and to-be-eliminated nodes to obtain a power grid coarsening model. By reasonably decomposing the power grid structure, the modeling of the power grid coarsening model is realized, which can retain key electrical characteristics and facilitate researchers to perform efficient, accurate and flexible analysis of large-scale power grid systems, solving the problem that the existing power grid coarsening model construction technology cannot comprehensively and reasonably decompose the large power grid according to the power grid structure and cannot retain key electrical characteristics. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0047] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0048] Figure 1 A schematic diagram of a flow chart of a method for constructing a coarsened power grid model based on community partitioning provided in an embodiment of the present application;

[0049] Figure 2 It is a flowchart of a method for constructing a coarsened model of a power grid based on community partitioning provided by an optional embodiment of the present application;

[0050] Figure 3 It is a schematic diagram of the steps of a method for constructing a coarsened model of a power grid based on community partitioning provided by an optional embodiment of the present application;

[0051] Figure 4 It is a community aggregation schematic diagram provided in an embodiment of the present application;

[0052] Figure 5 This is a schematic diagram of a community node aggregation and partitioning step process provided by an embodiment of the present application;

[0053] Figure 6 A schematic diagram of a community node aggregation process provided in an embodiment of the present application;

[0054] Figure 7 It is a schematic diagram of a power grid coarsening model device based on community partitioning provided in an embodiment of the present application. DETAILED DESCRIPTION

[0055] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, what ordinary technicians in this field obtain without making creative work belongs to the scope of protection of this application.

[0056] In the related technology, the existing scheme for constructing a coarsening model of the power grid can only be based on the equivalence of low voltage levels to high voltage levels or directly decomposing the power grid structure according to the jurisdiction area. The decomposition method is simple, resulting in the inability to fully consider the electrical coupling relationship between each site, and over-reliance on the experience of practitioners to decompose the power grid structure, resulting in the failure to comprehensively and reasonably decompose the large power grid according to the power grid structure, resulting in the generated coarsening model of the power grid being unable to retain key electrical characteristics, which brings great difficulties to subsequent analysis.

[0057] In order to solve the above problems, one of the technical concepts of the present application is to construct a power grid network model using the power grid parameters of the power system, and then partition the power grid network model through a preset algorithm to obtain a target community partition, and use the target community partition to construct a matrix using at least two nodes to be analyzed and the connection information between the nodes to be analyzed to obtain an admittance matrix and retained nodes, and then perform modeling optimization based on the retained nodes, the admittance matrix, and the nodes to be eliminated contained in the admittance matrix to obtain a power grid coarsening model, so that the power grid structure can be reasonably decomposed to achieve the modeling of the power grid coarsening model, and the key electrical characteristics are characterized by the power grid parameters, that is, the key electrical characteristics are retained, which facilitates researchers to conduct efficient, accurate and flexible analysis of the power grid system. It can be seen that the present application effectively solves various problems existing in the existing power grid coarsening model construction technology.

[0058] To facilitate the understanding of the embodiments of the present application, further explanation will be given below in conjunction with the drawings and specific embodiments. The embodiments do not constitute a limitation on the embodiments of the present application.

[0059] Figure 1 A schematic diagram of a process flow of a method for constructing a grid coarsening model based on community partitioning provided in an embodiment of the present application. Figure 1 As shown, the method for constructing a coarsened power grid model based on community partitioning provided in the embodiment of the present application may specifically include the following steps:

[0060] Step 110: determine the power system to be partitioned, and construct a power grid network model based on power grid parameters corresponding to the power system to be partitioned.

[0061] Specifically, the power system to be partitioned may be a power grid, which is composed of multiple devices, and the power grid may include a power substation, a power plant, and a load node, etc. The embodiment of the present application is not limited to this. The power grid parameters mainly include electrical parameters of each device in the power grid, which are used to characterize its electrical characteristics to ensure that the subsequently obtained power grid coarsening model can accurately characterize the key electrical characteristics and achieve the retention of the key electrical characteristics.

[0062] In a specific implementation, the embodiments of the present application will characterize the various devices / devices / sites that make up the power grid in the form of power grid nodes (referred to as: nodes), that is, the power grid substations, power plants, and load nodes, etc. are characterized in the form of nodes. For the possible connection relationships between nodes, the connection relationships between nodes are characterized in the form of edges, wherein the edges may include lines or transformers, and the embodiments of the present application do not impose any restrictions on this. The various devices of the power grid are characterized in the form of nodes, and the connection relationships are characterized in the form of edges. Subsequently, the nodes and edges can be used to establish an unweighted network graph (topology) according to the actual topology as a power grid network model, with reference to Figure 2 Schematic diagram of the steps shown.

[0063] It should be noted that, since different power grids have different compositions and actual structures, the power grid in this application is not limited to only power grid substations, power plants, and load nodes, and other equipment components can also be set according to the actual composition requirements of the power grid. In addition, since the distribution and actual functions of nodes in the power grid may not be exactly the same, the power grid nodes in this application are not limited to boundary nodes, scene setting nodes, and partition leading nodes.

[0064] Step 120, partitioning the power grid network model by a preset algorithm to obtain target community partitions.

[0065] The target community partition includes at least two nodes to be analyzed and connection information between the nodes to be analyzed.

[0066] Specifically, the preset algorithm may be a simulated annealing algorithm. Figure 2In this embodiment, the power grid network model can be partitioned for multiple times to achieve the decomposition of the power grid structure. Specifically, the community discovery algorithm (louvain) can be used to perform the first partition on each node in the power grid network model, and each node is divided into an independent community to form a community partitioning scheme. For each node, the node that has a connection relationship with the node can be called a neighbor node, and the community where the neighbor node is located is called a neighbor community. Then the second partition is performed, the nodes are sorted into a sequence, and each node is traversed to determine whether to move the node from the community where it is located to the neighbor community where the neighbor node is located. If so, the node movement action is performed until there is no node movement action traversed, the traversal is terminated, the node-free community is cleared, and the community after the second partition is generated. Finally, the simulated annealing algorithm is introduced to perform the third partition on the community after the second partition, and all the communities after the second partition are traversed to calculate the energy change of merging two communities. The community merger is performed according to the energy change to obtain the final community partitioning scheme as the target community partition, so that the power grid structure can be reasonably decomposed. The formulas for performing grid partitioning and implementing community partitioning for a specific grid network model will not be described in detail here.

[0067] Step 130 , constructing a matrix based on the connection information and the nodes to be analyzed to obtain an admittance matrix and retained nodes, wherein the admittance matrix includes retained nodes and nodes to be eliminated.

[0068] Specifically, retained nodes refer to nodes that can be retained in the coarsening model of the power grid, including but not limited to boundary nodes, scene setting nodes, partition dominant nodes, etc., and nodes to be eliminated refer to nodes that need to be eliminated in the coarsening model of the power grid.

[0069] Reference Figure 2 In this embodiment, node analysis is performed based on the connection information and the nodes to be analyzed to obtain the retained nodes and the nodes to be eliminated, the retained nodes are aggregated into a retained node set, the nodes to be eliminated are aggregated into a node set to be eliminated, and the mutual admittance of the nodes to be eliminated and the retained nodes is performed to generate an admittance matrix model as an admittance matrix. By keeping the other electrical parameters of the boundary nodes consistent with the original power system, it is ensured that the devices corresponding to the boundary nodes in the original power system can represent the same electrical characteristics as the original power system. The specific process of determining the retained nodes and the nodes to be eliminated, and calculating the admittance matrix for the nodes to be eliminated and the retained nodes will not be described in detail here.

[0070] Step 140: Perform modeling optimization processing based on the retained nodes and the nodes to be eliminated to obtain a coarsened model of the power grid.

[0071] In this embodiment, Gaussian elimination is performed on the admittance matrix, as shown in FIG. Figure 2, the nodes to be eliminated in the admittance matrix can be eliminated to obtain the simplified equivalent system admittance matrix, and then the simplified equivalent system admittance matrix is ​​used to splice with the retained nodes to obtain a coarsened model of the power grid. This achieves a reasonable decomposition of the power grid structure and can effectively retain key electrical characteristics, making it easier for researchers to determine the retained nodes and line electrical characteristics according to research needs, and further realize efficient, accurate and flexible analysis of large-scale power grid systems, which is helpful for large-scale power grid analysis work such as optimization of power grid planning, operation and control.

[0072] It can be seen that the embodiment of the present application determines the power system to be partitioned, and constructs a power grid network model based on the power grid parameters corresponding to the power system to be partitioned. The power grid network model is partitioned by a preset algorithm to obtain a target community partition, and the connection information and nodes to be analyzed contained in the target community partition are used to construct a matrix to obtain an admittance matrix. Modeling optimization processing is performed based on the admittance matrix containing retained nodes and to-be-eliminated nodes to obtain a power grid coarsening model. By reasonably decomposing the power grid structure, the modeling of the power grid coarsening model is realized, which can retain key electrical characteristics and facilitate researchers to conduct efficient, accurate and flexible analysis of large-scale power grid systems.

[0073] Reference Figure 3 , shows a schematic flow chart of the steps of a method for constructing a coarsened model of a power grid based on community partitioning provided by an optional embodiment of the present application. The method may specifically include the following steps:

[0074] Step 310: determine the power system to be partitioned, and construct a power grid network model based on the power grid parameters corresponding to the power system to be partitioned.

[0075] For the relevant description of step 310, please refer to the detailed description of the above step 110, which will not be repeated here.

[0076] In an optional embodiment, the above-mentioned construction of the power grid network model based on the power grid parameters corresponding to the power system to be partitioned may specifically include the following sub-steps:

[0077] Sub-step 3101: constructing an initial unweighted network graph based on the power system to be analyzed, wherein the initial unweighted network graph includes at least two power grid nodes and connection relationships between the power grid nodes.

[0078] In the specific implementation, each component equipment / device / site in the power system to be analyzed is mapped into a grid node, and the grid nodes are used to construct an unweighted network graph. Since the unweighted network graph needs to be updated later, the unweighted network graph at this time can be called the initial unweighted network graph, which is used to characterize the actual connection relationship between each site in the power grid.

[0079] Sub-step 3102: performing an intimacy analysis based on the at least two power grid nodes and the connection relationship corresponding to the at least two power grid nodes to obtain weight information.

[0080] In actual processing, each grid node in the initial unweighted network graph may have a direct connection relationship, an indirect connection relationship, or no connection relationship with other grid nodes, so other grid nodes connected to the grid node can be called neighbor nodes of the grid node.

[0081] In an optional implementation, the above-mentioned performing intimacy analysis based on the at least two grid nodes and the connection relationship corresponding to the at least two grid nodes to obtain weight information may specifically include: calculating a reactive voltage sensitivity matrix L between the grid nodes;

[0082] According to the formula Calculate the weight information between power grid nodes;

[0083] Among them, ΔQ i is the reactive power unbalance vector injected into the grid node i, U i is the voltage amplitude of grid node i, U j is the voltage amplitude of grid node j, ΔU i is the amplitude change vector of grid node i, ΔU j is the amplitude change vector of grid node j, θ ij is the voltage phase difference between grid node i and grid node j, G ij is the real part of the corresponding element of the grid node admittance matrix, B ij is the imaginary part of the corresponding element of the grid node admittance matrix, L ij is the weight of the edge between grid node i and grid node j.

[0084] Sub-step 3103, updating the initial unweighted network graph through the weight information to obtain a target unweighted network graph.

[0085] Sub-step 3104, constructing a power grid network model based on the target unweighted network graph.

[0086] The following is a unified description of sub-steps 3101 to 3104:

[0087] Specifically, this embodiment can use the reactive voltage sensitivity matrix between the power grid nodes to construct the intimacy between the power grid nodes, that is, to characterize the intimacy between two power grid nodes. Then, the reactive voltage sensitivity is used as the weight of the edge between the power grid nodes, and the weight is used to update the initial unweighted network graph to obtain the target unweighted network graph G, and then the target unweighted network graph is used to construct the power grid network model. Among them, the weight information can represent the weight of the edge between the two power grid nodes.

[0088] For example, for the grid nodes in the initial unweighted network graph, grid node i and grid node j represent two connected nodes, and the weight of the edge between grid node i and grid node j is represented as L ij , the weight L of the edge between the power grid nodes can be calculated by the following formula (1): ij :

[0089]

[0090] Among them, U i is the voltage amplitude of grid node i, U j is the voltage amplitude of the grid node j. According to the voltage amplitude of the grid node, the amplitude change of the grid node can be further calculated, that is, the amplitude change vector ΔU of the grid node i i , the amplitude change vector ΔU of grid node j j , ΔQ i is the reactive power unbalance vector injected into the grid node i, θ ij is the voltage phase difference between grid node i and grid node j, G ij is the real part of the corresponding element of the grid node admittance matrix, B ij is the imaginary part of the corresponding element of the grid node admittance matrix.

[0091] Step 320: partition the power grid network model into target community partitions using a preset algorithm.

[0092] The target community partition includes at least two nodes to be analyzed and connection information between the nodes to be analyzed.

[0093] For the relevant description of step 320, please refer to the detailed description of the above step 120, which will not be repeated here.

[0094] In an optional embodiment, the above-mentioned grid partitioning of the grid network model to obtain the target community partition may specifically include the following sub-steps:

[0095] Sub-step 3201, performing initial division on all grid nodes in the grid network model to obtain at least two initial grid communities, wherein the initial grid community includes at least one grid node.

[0096] Sub-step 3202, randomly sorting the power grid nodes to obtain a community traversal table, wherein the community traversal table includes nodes to be traversed, and the nodes to be traversed correspond to the power grid nodes one by one.

[0097] Sub-step 3203, performing gain analysis on each of the nodes to be traversed to obtain gain information.

[0098] The gain information is used to represent the modularity gain of the node to be traversed from the current initial power grid community to the neighboring power grid community.

[0099] The following is a unified description of sub-steps 3201 to 3203:

[0100] In a specific implementation, for all nodes contained in the power grid network model, each node is regarded as an independent community, or multiple adjacent nodes are divided into a community according to certain rules (such as a certain number or connection relationship). Here, this embodiment takes each node as an independent community as an example to describe the community division process in detail. Specifically, each power grid node represents an independent community, and multiple independent communities containing at least one power grid node are obtained, also called initial power grid communities. At this time, the power grid node corresponds to the initial power grid community one by one, that is, the power grid node and the initial power grid community can represent each other. Then, each power grid node is sorted according to random sorting or any existing sorting algorithm, and the power grid nodes are numbered according to the sorting order (wherein the numbering method can be digital numbering, letter numbering or other numbering, which is not limited in this example). Taking digital numbering as an example, the digital numbering order corresponding to the power grid node i is: i=1, 2, 3, 4, 5...n. Since the power grid node i corresponds to the initial power grid community one by one, the power grid node i and its current community are also one-to-one corresponding. After sorting the grid nodes according to the preset algorithm, the community traversal table can be obtained. The community traversal table is used to indicate the traversal order of the grid nodes, which includes the grid nodes / initial division communities. Then, for each grid node in the community traversal table, each grid node is traversed in order, and the modularity gain of grid node i leaving the current community and joining the communities of all other neighboring nodes is calculated. That is, for grid node i, all its neighboring nodes are determined respectively, and the modularity gain of grid node i joining the community of the neighboring node is calculated.

[0101] It should be noted that since the grid node i may have multiple neighbor nodes, that is, the grid node i can choose to join multiple communities after leaving the current community, the modularity gain corresponding to the grid node i can also be multiple, that is, the number of neighbor nodes of the grid node i is the same as the number of modularity gains, where the modularity gain corresponds one-to-one to the neighbor nodes of the grid node i.

[0102] In an optional embodiment, the embodiment of the present application performs gain analysis on each of the nodes to be traversed to obtain gain information, which may specifically include: Calculate the modularity Q, which is used to characterize the quality of the initial power grid community; visit each power grid node in turn, and calculate the modularity gain ΔQ of power grid node i leaving the current community d and joining the neighbor community c i_fd2c ; Where m is the sum of all value information in the network, 2m is the sum of all grid node degrees, W ij is the weight from grid node i to grid node j, d i is the degree of grid node i, δ(a, b) is the judgement, which is used to characterize the expected value of the edge weight between grid nodes and grid nodes.

[0103] As an example, see Figure 4 , Figure 4 The flowchart of the embodiment of the present application using the Louvain algorithm for node aggregation, first input the weight network (i.e., the power grid network model) of the power grid to be divided, and set the node access order and the impairment factor according to the Louvain algorithm. Then divide the initial community. When performing the initial division, an initialization network division scheme can be constructed, with each node representing an independent community to form a community division scheme (partition), calculate the total number of nodes n, and number the nodes in order according to formula (2):

[0104] partition=[{node1},{node2},{node3}......{node n}] Formula (2)

[0105] Then, the order of nodes node is randomly disrupted to form a node sequence node_list as the community traversal table. This sequence is used as the node access sequence to calculate the modularity Q used to evaluate the quality of the current community division, as shown in formula (3):

[0106]

[0107] Where m is the sum of all value information in the network, 2m is the sum of all grid node degrees, and W ij is the weight from grid node i to grid node j, d i is the degree of grid node i, δ(a, b) is the judgement, which is used to characterize the expected value of the edge weight between grid nodes and grid nodes.

[0108] Sub-step 3204, performing secondary division on the grid nodes corresponding to the nodes to be traversed according to the gain information to obtain secondary division communities.

[0109] In this embodiment, for each grid node i and at least one modularity gain corresponding to the grid node i, it can be determined whether to move the grid node i to the community where the neighboring node is located according to the modularity gain. Specifically, it can be determined whether the modularity gain is greater than a preset value (such as the preset value can be 0, that is, it is determined whether the modularity gain is greater than 0). Since there may be multiple neighboring nodes in the grid node i, that is, there may be multiple values ​​corresponding to the modularity gain of the grid node i, that is, for the grid node i, there may be multiple modularity gains greater than the preset value (that is, there may be multiple modularity gains greater than 0). Therefore, for all modularity gains greater than the preset value of the grid node i, the modularity gains greater than the preset value can be compared twice to determine the maximum modularity gain, and the corresponding neighbor community is determined according to the maximum modularity gain, and the action of moving the grid node i to the neighbor community is executed. All grid nodes are traversed in the above manner, and the grid node is moved from the current community to the neighbor community corresponding to the maximum modularity gain according to the modularity gain. If all modularity gains of the grid node i are not greater than the preset value, the moving operation is not performed. After traversing all grid nodes once and moving them in the above manner, the node sequence can be further traversed multiple times until the modularity gain corresponding to all grid nodes is no greater than a preset value, and the traversal is terminated to obtain a secondary partition community.

[0110] In actual processing, after traversing the node sequence multiple times, due to the movement of power grid nodes, some communities become empty communities without nodes due to node movement. These empty communities can be directly deleted, and the remaining communities containing nodes are output to obtain the node set aggregation plan as the secondary division community.

[0111] For example, refer to Figure 4 , assuming that there are nine nodes from node n1 to node n9, by aggregating nodes n1 to n9, we can get four secondary partition communities, c1, c2, c3 and c4.

[0112] In an optional embodiment, the embodiment of the present application performs secondary division of the grid nodes corresponding to the nodes to be traversed according to the gain information to obtain secondary division communities, which may specifically include: visiting each grid node i in turn, determining the modularity gain ΔQ i_fd2c Is it greater than zero? If the modularity gain ΔQ i_fd2c If it is greater than zero, grid node i is added from community d to neighbor community c.

[0113] For example, assuming that the community where node i belongs is community d, and the community where the neighboring node belongs is community c, the modularity gain corresponding to node i is ΔQ i_fd2c Indicates that for all modularity gains corresponding to node i, ΔQ is determined respectively i_fd2c Is it greater than 0? If ΔQi_fd2c >0, then all ΔQ i_fd2c >0, and the maximum modularity gain is obtained. Node i is separated from the current community d and added to the community c corresponding to the maximum modularity gain. After traversing all nodes, community c aggregates all the nodes currently contained to form C max , as expressed in the following formula (4):

[0114]

[0115] Among them, C i_nbr is the set of communities where all neighbor nodes of node i are located, and partition is the current community partitioning scheme.

[0116] Reference Figure 5 , traverse the node sequence nodes_list multiple times, and judge whether there is a node movement action according to the above method, that is, judge whether move>0, until there is no node movement action after traversing, end the traversal, clear the communities without nodes in partition, and output the node-level aggregation partitioning scheme communities, refer to Figure 4 The node-level aggregation results shown include com1 and com2.

[0117] The node-level aggregation partitioning scheme communities is as shown in the following formula (5):

[0118] communities=[com1,com2...com m ]

[0119] com l ={node a ,node b ...node x} Formula (5)

[0120] Among them, m is the number of communities.

[0121] Sub-step 3205, optimizing the secondary partitioned communities by a preset simulated annealing algorithm to obtain target community partitions.

[0122] In the specific implementation, the secondary partition community obtained by node-level aggregation can be equivalent to a node, which can be referred to as a community node for short. That is, the secondary partition community is used as a community node for subsequent processing. Since there are power grid nodes connecting two communities through edges between different communities, the connecting edges between different communities can be used as mutual edges between community nodes. In the same community, the edge connecting two power grid nodes is used as a self-edge to generate a new community sequence. Subsequently, the simulated annealing algorithm is introduced to traverse each community node in the community sequence in turn, calculate the minimum system energy after merging it with each neighboring community node, and then calculate the change in the merged energy. According to the change in the merged energy, the probability of accepting the action of merging the community node with the neighboring community node is calculated, and whether to execute the community merging action is determined according to the probability. If it is determined to execute the community merging action, the community node is merged with the neighboring community node, and the community partition optimization result is output. The final community partition scheme is returned to obtain the target community partition.

[0123] In an optional embodiment, the embodiment of the present application optimizes the secondary partition community through a preset simulated annealing algorithm to obtain a target community partition, which may specifically include: taking the secondary partition community as an equivalent node, and combining the connection relationship between the secondary partition communities to generate a community sequence; traversing all communities in the community sequence to determine the energy change between the current community and each neighbor community; determining the number of community mergers based on the energy change; determining a merger stop threshold; if the number of community mergers is less than the merger stop threshold, merging the current community and the neighbor community to obtain the target community partition.

[0124] For example, the secondary partitioned communities are equivalent to nodes, called community nodes or equivalent nodes, and are composed as follows Figure 4 The new community node link graph shown in the figure forms the community node sequence coms_list. Then the simulated annealing algorithm is introduced, and the natural exponent of the total number of community nodes m is used as the initial temperature T of the simulated annealing algorithm, and the following formula (6) is obtained:

[0125]

[0126] Where m0 is the number of initial community nodes, and the parameter α is used to control the number of final community mergers, and its value can be between [0,1]. The inverse of the modularity is taken as the system energy E(t), and the following formula (7) is obtained:

[0127]

[0128] Set the community merging stop threshold ε and calculate the threshold ε according to formula (8):

[0129]

[0130] like Figure 6 As shown, the communities in coms_list are visited / traversed in turn, and the minimum system energy E(t+1) after merging with each neighbor community is calculated. Then, the minimum combined energy change ΔE of the current community node joining the neighbor community is calculated based on the minimum system energy E(t+1) and the system energy E(t). The calculation formula of the combined energy change ΔE is shown in formula (9):

[0131] ΔE=E(t+1)-E(t) Formula (9)

[0132] Reference Figure 6 , determine whether the merge energy change ΔE is greater than 0. If ΔE<0, directly accept the merge plan and execute the community merge action. If ΔE>0, calculate the probability P of accepting the merge action, and then determine whether to execute the community merge action according to the probability, and record the number of community merges n after visiting all nodes b_move . The calculation probability P is shown in the following formula (10):

[0133]

[0134] Determine the number of community mergers n b_move Is it greater than the merge stop threshold ε? If n b_move >ε, then recalculate the temperature T with the current number of communities and continue the above process; if n b_move <ε, then stop merging, output the community partition optimization result, and return the final community partition solution, such as Figure 4 The final community division plan is shown in the lower part.

[0135] Step 330: Analyze the target community partition according to the equal value requirements to obtain information about the retained nodes and the nodes to be eliminated.

[0136] The to-be-eliminated node information includes the connection relationship between the to-be-eliminated nodes and the connection relationship between the to-be-eliminated nodes and the retained nodes, and the retained nodes include at least one of boundary nodes, scene setting nodes and partition dominant nodes.

[0137] In a specific implementation, this embodiment determines the retained nodes of each partition according to the equal value requirements, constructs a retained node set together with the boundary nodes, sets the remaining nodes as nodes to be eliminated, and constructs a set of nodes to be eliminated based on the nodes to be eliminated, so that the nodes to be eliminated can be eliminated later.

[0138] Step 340: Perform mutual admittance based on the information of the nodes to be eliminated to generate the admittance matrix.

[0139] In actual processing, the information of the nodes to be eliminated can be used to convert the load of the nodes to be eliminated and the conductance of the compensation capacitor into self-admittance, and then the connecting line node transformer between the nodes to be eliminated and the line or transformer impedance between the nodes to be eliminated and the retained nodes are converted into inter-node self-admittance and mutual admittance, and the line-to-ground capacitance is converted into the self-admittance of each node, thereby generating an admittance matrix.

[0140] In an optional embodiment, in the embodiment of the present application, when the connection relationship includes the connection relationship of the line model and the connection relationship of the transformer model, the mutual admittance based on the to-be-eliminated node information is performed to generate the admittance matrix, which may specifically include: i =Y i_load +Y i_shunt +Y i_line +Y i_tran Calculate the self-admittance according to Y ij Calculate the mutual admittance and get the admittance matrix, where Y i_load is the self-admittance of the node load to be eliminated, Y i_shunt To compensate for the self-admittance of the capacitor, Y i_line is the connection relationship self-admittance of the line model, Y i_tran is the connection relationship self-admittance of the transformer model.

[0141] Exemplarily, the reserved nodes of each partition are determined according to the equal value requirements, and these reserved nodes and the boundary nodes are classified into the reserved node set edge_buses. The remaining nodes are classified into the node set to be eliminated intro_buses. Then, the load self-admittance of the node to be eliminated and the compensation capacitor conductance self-admittance are sequentially processed, wherein the conversion formula of the load self-admittance of the node to be eliminated is shown in the following formula (11):

[0142]

[0143] The conversion formula of the compensation capacitor conductance to self-admittance is shown in the following formula (12):

[0144] Y i_shunt =(g_shunt+j*b_shunt) / S b Formula (12)

[0145] Subsequently, the node transformers of the connecting lines between the nodes to be eliminated, the lines or transformers between the nodes to be eliminated and the retained nodes are converted into the mutual admittance between nodes, and the line-to-ground capacitance is converted into the mutual admittance of each node, that is, an admittance matrix model is generated.

[0146] For the line model, the line mutual admittance between the node to be eliminated and the retained node and the line-to-ground capacitance are converted into the mutual admittance of each node, which is converted into the admittance calculation formula (13) as follows:

[0147]

[0148] For the above formula (13), r ij is the line resistance, x ij is the line reactance, y1 is the admittance of line end i to ground, and y2 is the admittance of line end j to ground.

[0149] For the transformer model, the connection line node transformer between the nodes to be eliminated and the transformer between the nodes to be eliminated and the retained nodes are converted into inter-node mutual admittance. Taking BPA as an example, it records the reference voltage across the transformer. The reference voltage across the transformer U b1 , U b2 , tap voltage U k1 and U k2 , base capacity S b And the transformer leakage reactance X calculated by tapping on both sides σ , thus we can deduce that the conversion formula into the admittance matrix model is as shown in the following formula (14):

[0150]

[0151] Finally, the self-admittance Y of the admittance matrix is ​​calculated by formula (15): i , which is expressed as the mutual admittance between nodes i and j by formula (13) and formula (14).

[0152] Y i =Y i_load +Y i_shunt +T i_line +Y i_tran Formula (15)

[0153] Among them, Y i_load is the self-admittance of the node load to be eliminated, Y i_shunt To compensate for the self-admittance of the capacitor, Y i_line is the connection relationship self-admittance of the line model, Y i_tran is the connection relationship self-admittance of the transformer model.

[0154] Step 350: Eliminate the nodes to be eliminated from the admittance matrix by using Gaussian elimination to obtain an equivalent system admittance matrix.

[0155] In a specific implementation, the admittance matrix may include multiple separate admittance matrices, and Gaussian elimination is performed on the block admittance matrix to generate an equivalent system admittance matrix.

[0156] In actual processing, for a power grid with no power supply at the node to be eliminated, it usually includes internal nodes, boundary nodes, external nodes, etc. For the self-admittance item of the boundary node, it mainly includes: the admittance to the ground converted from the boundary node's own load, line high pitch, reactive / active compensation, etc.; the mutual admittance between boundary nodes; the mutual admittance between the boundary node and the boundary node of the neighboring community; the mutual admittance between the boundary node and the node to be eliminated within the community, etc. In order to reduce the modeling process, only the required mutual admittance matrix can be modeled, and the rest retain the original system equipment parameters to simplify the admittance matrix, and then perform Gaussian elimination to generate an equivalent system admittance matrix.

[0157] In an optional implementation manner, in this embodiment, when the retained nodes include boundary nodes, the Gaussian elimination is used to eliminate the nodes to be eliminated from the admittance matrix to obtain the equivalent system admittance matrix, which may specifically include: Calculate the equivalent system admittance matrix, where Y ee4 is the mutual admittance between the boundary node and the node to be eliminated, Y et is the mutual admittance between the boundary node and the internal node, is the mutual admittance between internal nodes, Y te is the mutual admittance between the internal node and the boundary node.

[0158] For example, for a power grid with no power supply at a node to be eliminated, the subscript t represents an internal node, the subscript e represents a boundary node, and the subscript o represents an external node, and the admittance matrix of the entire system is formed as shown in formula (16):

[0159]

[0160] Among them, Y ee Represents the self-admittance item of the boundary node, which mainly includes: the ground admittance Y converted from the boundary node's own load, line high voltage, reactive / active power compensation, etc. ee1 ; Mutual admittance Y between boundary nodes ee2 ; Mutual admittance Y between the boundary node and the neighboring community boundary node ee3 ; Mutual admittance Y between the boundary node and the node to be eliminated within the community ee4 , satisfying formula (17):

[0161] Y ee =Y ee1 +Y ee2 +Y ee3 +Y ee4 Formula (17)

[0162] In particular, to reduce the modeling process, only Y ee4 , Y et , Yt , Y te Admittance modeling is performed, and the rest of the original system equipment parameters are retained. The following system-related parameters are retained only for the convenience of representation.

[0163] Then, Gaussian elimination is performed on the admittance matrix, and the simplified network equation is shown in equation (18):

[0164]

[0165] Among them, Y eq The expression is shown in formula (19):

[0166]

[0167] Step 360: concatenate the equivalent system admittance matrix with the reserved nodes to obtain the coarsened model of the power grid.

[0168] Step S42: concatenate the equivalent system admittance matrix with the reserved nodes to obtain the coarsened model of the power grid.

[0169] In actual processing, when a single community is simplified, the scope of change is only on the boundary nodes within the community and the lines between the boundary nodes. The parameters of the connecting lines connected to other communities and other systems within the community will not be changed. Therefore, in this embodiment, multiple communities can be simplified as needed, and the simplification calculations can be performed in parallel. Finally, the simplified equivalent system admittance matrices of each community are spliced. For communities that do not need to be simplified, such as retained nodes, they can be directly combined with other simplified equivalent system admittance matrices to construct an equivalent system for the entire network and generate an equivalent power grid coarse model, so as to effectively keep other electrical parameters of the boundary nodes consistent with the original system to ensure that these devices represent the same electrical characteristics as the original system.

[0170] In an optional embodiment, the above-mentioned splicing of the simplified equivalent system with the original system to generate an equivalent power grid system may specifically include: the system splicing of the equivalent system admittance matrix with the power grid parameters to obtain the power grid coarse model, including: adding each row of the equivalent system admittance matrix, taking the added value as the admittance to the ground of the retained node corresponding to the row, and adding the original admittance to the ground of the node to update the admittance to the ground of the retained node; according to Generate additional ground admittance Y of boundary node i i ;according to Generate an equivalent power grid system, where P iload_new is the active power load of the boundary node i of the equivalent power grid system, Q iload_new is the reactive power load of node i at the boundary of the equivalent power grid system, P iload is the original active power load of boundary node i, Q iloadis the original reactive power load of boundary node i.

[0171] As an example, the simplified admittance matrix is ​​summed for each row, and the sum is used as the ground admittance of the retained node corresponding to the row, and the sum is added to the original ground admittance of the node to update the ground admittance of the retained node. The additional ground admittance Y of the boundary node i is generated using formula (20): i , the additional ground admittance is used to generate new boundary node reactive and active power loads through formula (21).

[0172]

[0173] Among them, P iload_new is the active power load of the boundary node i of the equivalent power grid system, Q iload_new is the reactive power load of node i at the boundary of the equivalent power grid system, P iload is the original active power load of boundary node i, Q iload is the original reactive power load of boundary node i, S B is the base capacity, V i is the voltage amplitude of node i. eq The off-diagonal elements of the matrix (Y eq ) ij It is converted into the line or transformer impedance between boundary nodes i and j. If there is a line / transformer connection between the two nodes, the new line is connected in parallel next to the original transformer / line, so that other electrical parameters of the boundary nodes can be effectively kept consistent with the original system to ensure that these devices represent the same electrical characteristics as the original system.

[0174] In an optional implementation manner, this embodiment systematically splices the equivalent system admittance matrix with the power grid parameters to obtain the power grid coarsening model, which may specifically include: adding each row of the equivalent system admittance matrix, taking the added value as the admittance to the ground of the retained node corresponding to the row, adding the original admittance to the ground of the node and updating the admittance to the ground of the retained node; according to Generate additional ground admittance Y of boundary node i i ;according to Generate an equivalent power grid system, where P iload_new is the active power load of the boundary node i of the equivalent power grid system, Q iload_new is the reactive power load of node i at the boundary of the equivalent power grid system, P iload is the original active power load of boundary node i, Q iload is the original reactive power load of boundary node i

[0175] In summary, the embodiment of the present application determines the power system to be partitioned, and constructs a power grid network model based on the power grid parameters corresponding to the power system to be partitioned. Then, the power grid network model is partitioned by a preset algorithm to obtain the target community partition, and then the target community partition is analyzed according to the equivalent demand to obtain the information of the retained nodes and the nodes to be eliminated, and then the mutual admittance is performed based on the information of the nodes to be eliminated to generate an admittance matrix, and the nodes to be eliminated are eliminated from the admittance matrix by using Gaussian elimination to obtain the equivalent system admittance matrix, and the equivalent system admittance matrix is ​​spliced ​​with the retained nodes to obtain a power grid coarsening model, which realizes the reasonable decomposition of the power grid structure, and realizes the modeling of the power grid coarsening model on the basis of retaining the key electrical characteristics, which can facilitate researchers to determine the retained nodes and line electrical characteristics according to research needs, and then realize the efficient, accurate and flexible analysis of large-scale power grid systems, which is helpful for large-scale power grid analysis such as optimization of power grid planning, operation and control. It can be seen that the present application effectively solves various problems existing in the existing power grid coarsening model construction technology.

[0176] It should be noted that, for the purpose of simple description, the method embodiments are expressed as a series of action combinations, but those skilled in the art should be aware that the embodiments of the present application are not limited to the described order of actions, because according to the embodiments of the present application, certain steps may be performed in other orders or simultaneously.

[0177] like Figure 7 As shown, the embodiment of the present application further provides a network coarsening model construction device 700 based on community partitioning, including:

[0178] A power grid network model building module 710 is used to determine the power system to be partitioned, and build a power grid network model based on the power grid parameters corresponding to the power system to be partitioned;

[0179] A power grid partitioning module 720 is used to partition the power grid network model using a preset algorithm to obtain a target community partition, wherein the target community partition includes at least two nodes to be analyzed and connection information between the nodes to be analyzed;

[0180] A matrix construction module 730, configured to construct a matrix based on the connection information and the nodes to be analyzed, to obtain an admittance matrix and retained nodes, wherein the admittance matrix includes the nodes to be eliminated;

[0181] The modeling optimization module 740 is used to perform modeling optimization processing based on the retained nodes, the admittance matrix and the nodes to be eliminated to obtain a coarsened model of the power grid.

[0182] It should be noted that the device provided in the embodiments of the present application can execute the method provided in any embodiment of the present application, and has the corresponding functions and beneficial effects of the execution method.

[0183] In a specific implementation, the above-mentioned community partition-based grid coarsening model construction device can be integrated into a device, so that the device can reasonably decompose the grid structure and retain key electrical characteristics, as an electronic device, to achieve the modeling of the grid coarsening model. The electronic device can be composed of two or more physical entities, or it can be composed of one physical entity, such as a personal computer (PC), a computer, a server, etc., and the embodiment of the present application does not impose specific restrictions on this.

[0184] In actual processing, an embodiment of the present application provides an electronic device, including a processor 111, a communication interface 112, a memory 113 and a communication bus 114, wherein the processor 111, the communication interface 112, and the memory 113 communicate with each other through the communication bus 114; the memory 113 is used to store computer programs; the processor 111 is used to execute the program stored in the memory 113, and implement the steps of the method for constructing a coarsening model of a power grid based on community partitioning provided in any of the aforementioned method embodiments. Exemplarily, the steps of the method for constructing a coarsening model of a power grid based on community partitioning may include the following steps: determining the power system to be partitioned, and constructing a power grid network model based on the power grid parameters corresponding to the power system to be partitioned; partitioning the power grid network model through a preset algorithm to obtain a target community partition, wherein the target community partition includes at least two nodes to be analyzed and connection information between the nodes to be analyzed; constructing a matrix based on the connection information and the nodes to be analyzed to obtain an admittance matrix and retained nodes, wherein the admittance matrix includes nodes to be eliminated; performing modeling optimization processing based on the retained nodes, the admittance matrix and the nodes to be eliminated to obtain a coarsening model of the power grid.

[0185] An embodiment of the present application further provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the steps of the method for constructing a coarsened power grid model based on community partitioning as provided in any of the aforementioned method embodiments are implemented.

[0186] It should be noted that, in this article, relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements.

[0187] The above description is only a specific implementation of the present application, so that those skilled in the art can understand or implement the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest range consistent with the principles and novel features applied for herein.

Claims

1. A method for constructing a coarsened power grid model based on community partitioning, characterized in that: include: Determine the power system to be partitioned, and construct a power grid network model based on the power grid parameters corresponding to the power system to be partitioned; By using a preset algorithm, the power grid network model is partitioned to obtain a target community partition, wherein the target community partition includes at least two nodes to be analyzed and connection information between the nodes to be analyzed; Performing matrix construction based on the connection information and the nodes to be analyzed to obtain an admittance matrix and retained nodes, wherein the admittance matrix includes the nodes to be eliminated; Modeling optimization processing is performed based on the retained nodes, the admittance matrix and the nodes to be eliminated to obtain a coarsened model of the power grid.

2. The method according to claim 1, characterized in that The step of constructing a power grid network model based on power grid parameters corresponding to the power system to be partitioned includes: Constructing an initial unweighted network graph based on the power system to be analyzed, wherein the initial unweighted network graph includes at least two power grid nodes and connection relationships between the power grid nodes; Performing an intimacy analysis based on the at least two power grid nodes and a connection relationship corresponding to the at least two power grid nodes to obtain weight information; The initial unweighted network graph is updated by using the weight information to obtain a target unweighted network graph; A power grid network model is constructed based on the target unweighted network graph.

3. The method according to claim 2, characterized in that The performing of intimacy analysis based on the at least two power grid nodes and the connection relationship corresponding to the at least two power grid nodes to obtain weight information includes: Calculating a reactive voltage sensitivity matrix L between the grid nodes; According to the formula Calculate the weight information between power grid nodes; Among them, ΔQ i is the reactive power unbalance vector injected into the grid node i, U i is the voltage amplitude of grid node i, U j is the voltage amplitude of grid node j, ΔU i is the amplitude change vector of grid node i, ΔU j is the amplitude change vector of grid node j, θ ij is the voltage phase angle difference between grid node i and grid node j, G ij is the real part of the corresponding element of the grid node admittance matrix, B ij is the imaginary part of the corresponding element of the grid node admittance matrix, L ij is the weight of the edge between grid node i and grid node j.

4. The method according to claim 1, characterized in that: The performing grid partitioning on the grid network model to obtain target community partitions includes: Performing an initial division on all grid nodes in the grid network model to obtain at least two initial grid communities, wherein the initial grid community includes at least one of the grid nodes; Randomly sorting the power grid nodes to obtain a community traversal table, wherein the community traversal table includes nodes to be traversed, and the nodes to be traversed correspond to the power grid nodes one by one; Performing gain analysis on each of the nodes to be traversed to obtain gain information, wherein the gain information is used to characterize the modularity gain of the node to be traversed from the current initial power grid community to the neighboring power grid community; Perform secondary division on the power grid nodes corresponding to the to-be-traversed nodes according to the gain information to obtain secondary division communities; The secondary partitioned communities are optimized by a preset simulated annealing algorithm to obtain target community partitions.

5. The method according to claim 4, characterized in that The step of performing gain analysis on each of the nodes to be traversed to obtain gain information includes: according to Calculate the modularity Q, which is used to characterize the quality of the initial power grid community; visit each power grid node in turn, and calculate the modularity gain ΔQ of power grid node i leaving the current community d and joining the neighbor community c i_fd2c ; Where m is the sum of all value information in the network, 2m is the sum of all grid node degrees, W ij is the weight from grid node i to grid node j, d i is the degree of grid node i, δ(a, b) is the judgment symbol, which is used to characterize the expected value of the weight of the grid node and the edge connecting the grid node; The secondary division of the grid nodes corresponding to the to-be-traversed nodes according to the gain information to obtain the secondary division community includes: visiting each grid node i in turn, determining the modularity gain ΔQ i_fd2c Is it greater than zero? If the modularity gain ΔQ i_fd2c If it is greater than zero, grid node i is added from community d to neighbor community c; The method optimizes the secondary partition community by a preset simulated annealing algorithm to obtain a target community partition, including: taking the secondary partition community as an equivalent node and combining the connection relationship between the secondary partition communities to generate a community sequence; traversing all communities in the community sequence to determine the energy change between the current community and each neighbor community; determining the number of community mergers based on the energy change; determining a merger stop threshold; if the number of community mergers is less than the merger stop threshold, merging the current community and the neighbor community to obtain the target community partition.

6. The method according to claim 1, characterized in that The step of constructing a matrix based on the connection information and the nodes to be analyzed to obtain an admittance matrix and reserved nodes includes: Analyzing the target community partition according to the equivalent requirements to obtain information of retained nodes and to-be-eliminated nodes, wherein the to-be-eliminated node information includes connection relationships between the to-be-eliminated nodes and connection relationships between the to-be-eliminated nodes and the retained nodes, and the retained nodes include at least one of boundary nodes, scene setting nodes, and partition leading nodes; The mutual admittance is performed based on the information of the nodes to be eliminated to generate the admittance matrix.

7. The method according to claim 6, characterized in that The connection relationship includes the connection relationship of the line model and the connection relationship of the transformer model, and the generating of the admittance matrix based on the to-be-eliminated node information includes: According to Y i =Y i_load +Y i_shunt +Y i_line +Y i_tran Calculate the self-admittance according to Y ij Calculate the mutual admittance and get the admittance matrix, where Y i_load is the self-admittance of the node load to be eliminated, Y i_shunt To compensate for the self-admittance of the capacitor, Y i_line is the connection relationship self-admittance of the line model, Y i_tran is the connection relationship self-admittance of the transformer model.

8. The method according to claim 1, characterized in that The modeling optimization process is performed based on the retained nodes, the admittance matrix and the nodes to be eliminated to obtain a coarsened model of the power grid, including: Eliminating the to-be-eliminated nodes from the admittance matrix using Gaussian elimination to obtain an equivalent system admittance matrix; The equivalent system admittance matrix is ​​concatenated with the reserved nodes to obtain the coarsened model of the power grid.

9. The method according to claim 8, characterized in that The retained nodes include boundary nodes, and the Gaussian elimination is used to eliminate the nodes to be eliminated from the admittance matrix to obtain an equivalent system admittance matrix, including: according to Calculate the equivalent system admittance matrix, where Y ee4 is the mutual admittance between the boundary node and the node to be eliminated, Y et is the mutual admittance between the boundary node and the internal node, is the mutual admittance between internal nodes, Y te is the mutual admittance between the internal node and the boundary node; The method of systematically splicing the equivalent system admittance matrix with the power grid parameters to obtain the power grid coarsening model includes: adding each row of the equivalent system admittance matrix, taking the added value as the ground admittance of the retained node corresponding to the row, adding the original ground admittance of the node and updating the ground admittance of the retained node; and Generate additional ground admittance Y of boundary node i i ;according to Generate an equivalent power grid system, where P iload_new is the active power load of the boundary node i of the equivalent power grid system, Q iload_new is the reactive power load of node i at the boundary of the equivalent power grid system, P iload is the original active power load of boundary node i, Q iload is the original reactive power load of boundary node i.

10. A device for constructing a coarsened model of a power grid based on community partitioning, characterized in that: include: A power grid network model building module, used to determine the power system to be partitioned, and build a power grid network model based on the power grid parameters corresponding to the power system to be partitioned; A power grid partitioning module, used to partition the power grid network model by a preset algorithm to obtain a target community partition, wherein the target community partition includes at least two nodes to be analyzed and connection information between the nodes to be analyzed; A matrix construction module, used to construct a matrix based on the connection information and the nodes to be analyzed to obtain an admittance matrix and retained nodes, wherein the admittance matrix includes the nodes to be eliminated; The modeling optimization module is used to perform modeling optimization processing based on the retained nodes, the admittance matrix and the nodes to be eliminated to obtain a coarsened model of the power grid.