Power system partitioning method and device, computer equipment and storage medium

Abstract

The invention provides an electric power system partitioning method and device, computer equipment and a storage medium, and belongs to the field of electric power control, and the method comprises the steps: constructing an adjacent matrix based on a topological structure of an electric power system, and calculating a similarity distance matrix DSC of the distribution structure connectivity of nodes of the electric power system through the adjacent matrix; calculating a similarity distance matrix FRC based on the frequency response curve of each node; performing weighted fusion on the DSC and the FRC after normalization to generate a comprehensive similarity measurement distance matrix FRCDSC; taking the FRCDSC as a partition basis, randomly selecting an initial clustering center, and dividing each node into a region where the clustering center with the minimum FRCDSC is located; updating the clustering center to be the node with the minimum FRCDSC sum among the nodes in the region; and iteratively updating until the clustering center is unchanged, and outputting a partitioning result. Therefore, the node frequency deviation in the region can be reduced, and the oscillation amplitude of partition boundary nodes is reduced.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of power control, and particularly relates to a power system partitioning method and device, computer equipment and a storage medium. BACKGROUND

[0002] To achieve the goal of "carbon peak and carbon neutral", the installed capacity of new energy mainly including wind power and photovoltaic will be further increased. New energy units are connected to the power grid through power electronic converter equipment and cannot actively provide inertia support to the system, resulting in a continuous decrease in system inertia level and uneven distribution of inertia. Since inertia is an important indicator of power system frequency stability, it directly reflects the system frequency drop rate after disturbance. Therefore, estimating the inertia of the power system is of great significance to resist internal and external disturbances and stable operation of the system.

[0003] At present, inertia estimation has shifted from the system side to the regional side, and accurate partitioning of the system is a prerequisite for regional inertia estimation.

[0004] In the existing partitioning method, a relative correlation gain matrix is constructed through the frequency response correlation between generator units, wherein the size of the elements reflects the correlation degree of the frequency response characteristics between the nodes, and the units with similar sizes are divided into a region, but the analysis of the partitioning of other nodes is lacking. Alternatively, principal component analysis is used to reduce the dimensionality of the indicators affecting the inertia distribution of the system, and the generator units are partitioned according to the reduced indicators, while the remaining nodes are divided into the nearest region according to the electrical distance. In addition, the actual configuration of the Phasor Measurement Unit (PMU) in the power grid is used to passively divide the region. Furthermore, the concept of graph theory is used to regard the power system as a graph, and the electrical distance is used to represent the connection strength between nodes, and then the spectral clustering algorithm is used to partition the power system. The above partitioning methods only consider the connectivity of the distribution structure of the region, and divide the nodes closely connected into the same region, and can only obtain a fixed partitioning result, and cannot guarantee the similarity of the frequency under various disturbance types. SUMMARY

[0005] To solve the problem, the present application provides a power system partitioning method, device, computer equipment and storage medium.

[0006] To achieve the above-mentioned purpose, the present application provides the following technical solutions: A power system partitioning method, the method comprising: constructing an adjacency matrix based on the topological structure of the power system, and calculating a similarity distance matrix DSC of the distribution structure connectivity of the nodes of the power system through the adjacency matrix; calculating a similarity distance matrix FRC based on frequency response curves of each node of the power system; generating a comprehensive similarity distance matrix FRCDSC by normalizing and fusing the DSC and FRC; taking the FRCDSC as the partition basis, randomly selecting an initial clustering center, dividing each node into a region where the FRCDSC is the smallest, updating the clustering center to be a node with the smallest sum of FRCDSC between nodes in the region, and iteratively updating until the clustering center is unchanged, and outputting the partition result.

[0007] Optionally, the constructing an adjacency matrix based on the topology of the power system comprises: abstracting the power system as a graph G(V, E), wherein the node set V corresponds to the nodes of the power system, and the edge set E corresponds to the power transmission lines; for any nodes i and j, when there is a directly connected power transmission line, W(i, j) = 1, otherwise W(i, j) = 0.

[0008] Optionally, the calculating a similarity distance matrix DSC of the connectivity of the distribution structure of the nodes of the power system through the adjacency matrix comprises: calculating a degree matrix D and a normalized Laplacian matrix L based on the adjacency matrix W; extracting the first R eigenvectors of the normalized Laplacian matrix L to form a matrix U; calculating the DSC distance between nodes i and j based on a preset formula to obtain the similarity distance matrix DSC; wherein the preset formula is: ; wherein R represents the number of columns of U, 、 representing the elements in U.

[0009] Optionally, the calculating a similarity distance matrix FRC based on frequency response curves of each node of the power system comprises: collecting the frequency response curves of each node after the disturbance; applying a dynamic time warping algorithm to calculate the warping distance between the frequency response curves of any two nodes to construct the FRC.

[0010] Optionally, after outputting the partition result, verification is further performed, comprising: based on the adjacency matrix, verifying whether there is an infinite distance between any two nodes in each region using the Floyd algorithm; when there is no infinite distance in all regions, it is determined that the distribution structure connectivity is satisfied.

[0011] Optionally, the verification further comprises: calculating the average value A of the FRC distance between nodes in the regionin ; Computing the average value A of the inter-node FRC distance in the region out ; When A out > A in , it is determined that the frequency similarity is satisfied.

[0012] Optionally, the iteratively updating until the clustering centers are unchanged comprises: calculating the silhouette coefficient under the current number of clustering centers during each iteration; taking the number of partitions closest to 1 as the optimal number of partitions.

[0013] A real-time inertia partitioning device for a power system, comprising: a construction module configured to construct an adjacency matrix based on the topological structure of the power system, to calculate a similarity distance matrix DSC of the connectivity of the distribution structure of the nodes of the power system through the adjacency matrix, and to calculate a similarity distance matrix FRC based on the frequency response curves of the nodes; a fusion module configured to fuse the normalized DSC and FRC after weighting to generate a comprehensive similarity measure distance matrix FRCDSC; an optimization module configured to take the FRCDSC as the basis for partitioning, to randomly select initial clustering centers, to divide each node into the region in which the clustering center is located with the minimum FRCDSC, to update the clustering centers to the nodes with the minimum sum of the FRCDSC of the nodes in the region, and to iteratively update until the clustering centers are unchanged to output the partitioning result.

[0014] A computer readable storage medium, the storage medium storing a computer program, the computer program being executed by a processor to implement the power system partitioning method.

[0015] A computer device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, the processor implementing the power system partitioning method when executing the program.

[0016] The power system partitioning method provided by the present application has the following beneficial effects: The DSC distance matrix constructed by the adjacency matrix calculation can retain the physical connection characteristics of the power grid, and the FRC distance matrix generated based on the frequency response curve can quantify the coherent characteristics of the nodes under dynamic disturbance, and the normalized and weighted fusion of the two makes the FRCDSC matrix have both static topology and dynamic response information, which fundamentally changes the defects of single-dimensional partition. Then, the partition process based on the criterion of minimizing the comprehensive distance can automatically form an optimal partition that meets the electrical connection closeness and ensures the frequency response synergy. The partition result of this dynamic fusion method can reduce the frequency deviation of the nodes in the region and the oscillation amplitude of the boundary nodes under different disturbance scenarios such as short-circuit fault and load mutation, significantly improves the stability of regional autonomous control and the accuracy of inertia estimation; and the matrix weighting mechanism can flexibly adjust the structure or frequency response weight according to the actual power grid demand, so that the partition is more suitable for specific operation modes. BRIEF DESCRIPTION OF DRAWINGS

[0017] In order to more clearly illustrate the embodiments of the present application and the design scheme thereof, the following will briefly introduce the drawings required by the embodiments. The drawings in the following description are only part of the embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor on the basis of these drawings.

[0018] Figure 1 A flowchart of a power system partitioning method according to an exemplary embodiment of the present application.

[0019] Figure 2 A flowchart of a real-time inertia partitioning process of a power system according to an exemplary embodiment of the present application.

[0020] Figure 3 A block diagram of a power system partitioning device according to an exemplary embodiment of the present application. DETAILED DESCRIPTION

[0021] In order for those skilled in the art to better understand the technical solutions of the present application and to be able to implement them, the present application will be described in detail below in conjunction with the drawings and specific embodiments. The following embodiments are only used to more clearly illustrate the technical solutions of the present application, and cannot be used to limit the protection scope of the present application.

[0022] The application provides a real-time partitioning method for inertia of a power system based on an improved Floyd k-medoids algorithm and a comprehensive similarity measure. Firstly, the application analyzes the definition of the equivalent inertia of a region of the power system and illustrates the principle of estimating the regional inertia. Secondly, the comprehensive similarity measure distance is obtained by simultaneously considering the connectivity of the distribution structure and the frequency similarity, so that the similarity between nodes can be better measured. Thirdly, the improved k-medoids algorithm is used for partitioning the system according to the comprehensive similarity measure distance, including introducing the Floyd algorithm to calculate the shortest distance length between any two points in the regional adjacency matrix, and calculating the average value of the FRC quantization index in the region and between the regions, so that the partitioning result can meet the frequency similarity and the connectivity of the distribution structure. Further, the profile coefficient is used to determine the number of partitions and measure the good and bad of the partitioning. Finally, simulation verification is carried out in the IEEE39 system and the IEEE118 system, and the results show that the method can update the partitioning of the power system in real time under different disturbance types, ensure the frequency similarity and the connectivity of the distribution structure in the region, improve the accuracy and rationality of the system partitioning, and obtain relatively accurate inertia estimation results.

[0023] In the application, the power system partitioning is for the regional inertia estimation of the power system. The inertia of the power system refers to the ability of the power system to resist frequency changes after being disturbed. Currently, the inertia constant is commonly used to represent the size of the inertia, and the unit is s. For a region with only synchronous machines, the equivalent inertia can be defined as the ratio of the weighted sum of the inertia of each synchronous generator according to its rated capacity to the total rated capacity of the system:

[0024] In the formula, is the equivalent inertia of the region with only synchronous machines; is the total rated capacity of the system; and are the inertia constant and rated capacity of the Ith generator, respectively; and N is the number of synchronous generators.

[0025] With the large-scale access of new energy to the power system, the source of inertia is no longer limited to synchronous generators. The equivalent inertia of a region containing new energy units can be expressed as:

[0026] In the formula, is the equivalent inertia of the region containing new energy units; is the virtual inertia of the pth new energy unit, is the capacity; M and Q are the number of synchronous generators and new energy units, respectively.

[0027] At present, the inertia estimation has been transferred from the system side to the regional side, and the equivalent formula can be used to calculate the system equivalent inertia with the regional inertia, while the regional inertia cannot be derived from the system inertia, so the regional inertia estimation is more flexible than the system inertia estimation.

[0028] In the scenario of the power system regional inertia estimation, for the divided region, the regional inertia can be estimated by obtaining the tie-line power between the regions after the disturbance and the center of inertia (COI) frequency in the region. The frequency dynamic response of the region is shown in equation (3):

[0029] In the formula: is the inertia of the region l; is the center of inertia frequency of the region l; is the change amount of the tie-line active power, and is the unbalanced power of the region, which is the sum of the change amounts of all tie-line active powers.

[0030] After the inertia of each region is estimated, the inertia of the entire system can be calculated using equation (4).

[0031] In the formula: l represents the number of regions, and the inertia of each region has been reduced to the system rated capacity and can be directly added. In order to more accurately and quickly estimate the regional inertia, the power system inertia partition result needs to be quickly and accurately obtained.

[0032] The technical solutions provided by the embodiments of the present application will be described in detail below in combination with the drawings.

[0033] Firstly, the present application provides a power system partitioning method, as shown in Figure 1 The method comprises the following steps: S101, constructing an adjacency matrix based on the topological structure of the power system, and calculating a similarity distance matrix DSC of the connectivity of the distribution structure of the nodes of the power system through the adjacency matrix.

[0034] In this step, first, the power system is abstracted as a graph G(V, E), wherein the node set V corresponds to the nodes of the power system, and the edge set E corresponds to the power transmission lines; for any node i and j, when there is a directly connected power transmission line, W(i, j) = 1, otherwise W(i, j) = 0; then the degree matrix D and the normalized Laplacian matrix L are calculated based on the adjacency matrix W; the first R eigenvectors of the normalized Laplacian matrix L are extracted to form a matrix U; the DSC distance between nodes i and j is calculated based on a preset formula to obtain the similarity distance matrix DSC.

[0035] In one embodiment, the present application is based on real-time partitioning of power system inertia using Floyd improved k-medoids algorithm and comprehensive similarity metric. The present application provides a comprehensive similarity metric index to measure the similarity between nodes from two indexes of distribution structure connectivity and FRC. The comprehensive similarity metric distance between nodes is calculated, and the smaller the distance, the higher the similarity between nodes, i.e. the frequency of the nodes is more similar and the structure connection based on electrical distance is more compact, and these nodes can be divided into the same area. Therefore, the comprehensive similarity metric distance is used as the basis for partitioning, and the improved k-medoids algorithm is used to perform real-time partitioning of power system inertia.

[0036] For example, first, the similarity index based on distribution structure connectivity is calculated. Based on the distribution structure connectivity, the topology of the power system can be used to mine regions of any shape, and it can be ensured that nodes with strong electrical connection are divided into the same region. The steps for calculating the similarity distance based on distribution structure connectivity are as follows.

[0037] First, the power system can be regarded as a graph G(V, E), which includes a vertex set V and an edge set E. The n nodes of the system are vertices, and each edge in the edge set can be represented by E(i, j), which represents the power transmission line between node i and node j. According to the graph G(V, E), the adjacency matrix W of the system can be obtained, and the weight between nodes i and j is As shown in equation (5):

[0038] That is, in W, the elements of two nodes directly connected are 1, and the other elements are 0, and the adjacency matrix is a symmetric matrix.

[0039] Secondly, the spectral clustering algorithm is a clustering method based on graph theory, which can construct the similarity distance of connectivity according to the adjacency matrix. The degree matrix D is constructed based on the adjacency matrix, and the degree of the degree matrix is And The relationship is shown in equation (6):

[0040] That is, the degree matrix D is shown in equation (7): Non-normalized Laplacian matrix As shown in equation (8): Then the normalized Laplacian matrix L is shown in equation (9): Finally, the eigenvector matrix U of L is solved, and the similarity distance matrix DSC of the distribution structure connectivity is obtained according to the Euclidean formula, and is defined as formula (10) as follows: In the formula, R=3 represents the column number of U, 、 represents the element in U.

[0041] S102, similarity distance matrix FRC is calculated based on the frequency response curve of each node.

[0042] In this step, the frequency response curve of each node after disturbance can be collected, and the dynamic time warping algorithm is applied to calculate the bending distance of the frequency response curve between any two nodes to construct FRC.

[0043] For example, based on the similarity index calculation of the frequency response curve, the FRC of each node in the system will change under different disturbances, which can ensure the real-time performance of the partition result and the frequency similarity in the region. Therefore, FRC can be defined as a similarity measurement index, and the DTW algorithm is suitable for calculating the distance between two time series. Then the similarity distance matrix FRC of FRC is defined as formula (11) as follows:

[0044] Among them, is the frequency difference of the two curves, and S is the number of sampling points.

[0045] S103, after normalization, the DSC and FRC are weighted and fused to generate the comprehensive similarity measurement distance matrix FRCDSC.

[0046] In this step, in order to reflect the comprehensive similarity between the nodes of the system, the above two similarity distances are normalized and combined to obtain the comprehensive similarity measurement distance matrix FRCDSC, and the specific formula is formula (12) as follows: In the formula, 、 are weight coefficients, which are 0.5 respectively.

[0047] The smaller the comprehensive similarity measurement distance is, the higher the similarity degree between the nodes is, that is, the more similar the frequency between the nodes is and the more closely the structure connection based on the electrical distance is, and these nodes can be divided into the same region. Therefore, on the basis of the comprehensive similarity measurement distance, the inertia real-time partition of the power system can be completed through the clustering algorithm.

[0048] S104, taking FRCDSC as the partition basis, randomly selecting an initial clustering center, dividing each node to the area where the clustering center with the minimum FRCDSC is located; updating the clustering center to the node with the minimum sum of FRCDSC between nodes in the area; iteratively updating until the clustering center is unchanged, and outputting the partition result.

[0049] In this step, the connectivity clustering algorithm of k-medoids based on Floyd needs to be improved.

[0050] Firstly, the present application introduces Floyd algorithm on the basis of k-medoids. Floyd algorithm is an algorithm for solving the shortest distance between any two points, which can calculate the shortest distance between any two points in the system, and the distance between two points is infinite, thereby being conducive to finding a solution meeting the connectivity of distribution structure. After obtaining the partition result by using k-medoids, the specific steps of ensuring the partition result meeting the connectivity by using Floyd algorithm are as follows:

[0051] (1) Matrix initialization. After obtaining the partition result by using k-medoids each time, the adjacency matrix of each area is imported, and the elements not adjacent are initialized to infinity.

[0052] (2) Traversing the adjacency matrix. Starting from the first node, the traversal is performed until the nth node is reached. If there are two nodes not connected in the area, there will be infinity distance, and it will be determined that the partition result is not connected.

[0053] (3) Repeating the traversal until the nodes in each area are traversed.

[0054] (4) Outputting the partition result meeting the connectivity.

[0055] In addition, the present application also verifies after outputting the partition result. For example, based on the adjacency matrix, whether there is an infinite distance between any two nodes in each area is verified by using Floyd algorithm; when there is no infinite distance in all areas, it is determined that the distribution structure connectivity is met; the average value A in of FRC distance between nodes in the area is calculated; the average value A out of FRC distance between nodes in the area is calculated; when A out >A in , it is determined that the frequency similarity is met.

[0056] Secondly, in order to make the frequency response curves in the same area more similar, and the frequency response curve similarity between different areas as small as possible, the present application defines the FRC quantitative index, that is, the average value of the distance matrix FRC between nodes in the area is calculated, and the average value of the distance matrix FRC between nodes in each area is calculated. If the average value between areas is greater than the average value in the area, it is considered that the partition result meets the frequency similarity.

[0057] Finally, the present application automatically determines the optimal number of clusters using the silhouette coefficient. The range of the silhouette coefficient is [-1, 1], and the closer to 1, the better the separation between clusters. Therefore, the silhouette coefficient is calculated under the current number of cluster centers during each iteration; the number of partitions closest to 1 is taken as the optimal number of partitions.

[0058] For example, based on the integrated similarity distance, the power system inertia real-time partitioning step based on the improved k-medoids algorithm is as follows.

[0059] Step one: Calculate the integrated similarity distance matrix FRCDSC from formulas (5)-(12).

[0060] Step two: Divide the system using the improved k-medoids clustering algorithm, randomly select a set of nodes as initial cluster centers, and divide the nodes to the region most similar (i.e., the minimum FRCDSC) according to the integrated similarity of each node and the cluster center.

[0061] Step three: Update the cluster center. After the nodes are divided into the region in the last step, the sum of FRCDSC of each point in the region and other points in the region is calculated, and then the node with the minimum sum of distances is selected as the new cluster center.

[0062] Step four: Repeat steps two and three until the cluster center is unchanged, and obtain the partitioning result.

[0063] Step five: Keep the solution that meets the frequency similarity and distribution structure connectivity, and compare the size of the silhouette coefficient to obtain the number of partitions and the partitioning result at this time. The real-time partitioning flowchart is shown in Figure 2 .

[0064] Using the above method, the DSC distance matrix calculated by the adjacency matrix can retain the physical connection characteristics of the power grid, and the FRC distance matrix generated based on the frequency response curve can quantify the coherent characteristics of the nodes under dynamic disturbance. The normalized weighted fusion of the two makes the FRCDSC matrix have both static topology and dynamic response information, fundamentally changing the defects of single-dimensional partitioning. Then, the partitioning process based on the minimum integrated distance criterion can automatically form the optimal partitioning that meets the electrical connection tightness and ensures the frequency response synergy. The partitioning result of this dynamic fusion method can reduce the frequency deviation of the nodes in the region and the oscillation amplitude of the boundary nodes under different disturbance scenarios such as short-circuit fault and load mutation, significantly improving the stability of regional autonomous control and the accuracy of inertia estimation; and the matrix weighting mechanism can flexibly adjust the structure or frequency response weight according to the actual power grid demand, making the partitioning more suitable for specific operating modes.

[0065] Secondly, the application further provides a power system partition device, as shown in the accompanying drawings, comprising: Figure 3 A construction module 201 is configured to construct an adjacency matrix based on a topological structure of the power system, calculate a similarity distance matrix DSC of the distribution structure connectivity of the nodes of the power system through the adjacency matrix, and calculate a similarity distance matrix FRC based on frequency response curves of the nodes of the power system. The construction module 201 is configured to construct an adjacency matrix based on a topological structure of the power system, calculate a similarity distance matrix DSC of the distribution structure connectivity of the nodes of the power system through the adjacency matrix, and calculate a similarity distance matrix FRC based on frequency response curves of the nodes of the power system.

[0066] A fusion module 202 is configured to fuse the DSC and the FRC after normalization and weighting to generate a comprehensive similarity distance matrix FRCDSC.

[0067] An optimization module 203 is configured to take the FRCDSC as a partition basis, randomly select an initial clustering center, divide each node into a region where the clustering center is located, and update the clustering center to be a node with the smallest sum of the FRCDSC between the nodes in the region.

[0068] The DSC distance matrix calculated by the above device can retain the physical connection characteristics of the power grid, and the FRC distance matrix generated based on the frequency response curve can quantify the coherent characteristics of the nodes under dynamic disturbance. The normalization and weighting fusion of the two makes the FRCDSC matrix have both static topological and dynamic response information, and fundamentally changes the defects of single-dimensional partition. The partition process taking the minimum comprehensive distance as the criterion can automatically form an optimal partition that meets the electrical connection closeness and guarantees the frequency response cooperativity. The partition result of the dynamic fusion method can reduce the frequency deviation of the nodes in the region and the oscillation amplitude of the boundary nodes under different disturbance scenarios such as short-circuit faults and load mutations, significantly improves the stability of regional autonomous control and the accuracy of inertia estimation, and the matrix weighting mechanism can flexibly adjust the structure or frequency response weight according to the actual power grid demand, so that the partition is more suitable for specific operation modes.

[0069] The application further provides a computer readable storage medium, which stores a computer program, and the computer program can be used to execute the steps of the power system partition method. Figure 1 The application further provides a computer readable storage medium, which stores a computer program, and the computer program can be used to execute the steps of the power system partition method.

[0070] The application further provides a computer device, which comprises a processor, an internal bus, a network interface, a memory and a non-volatile memory at the hardware level, and can further comprise other hardware required by business. The processor reads the corresponding computer program from the non-volatile memory into the memory and then runs to implement the steps of the power system partition method. Figure 1 The application further provides a computer device, which comprises a processor, an internal bus, a network interface, a memory and a non-volatile memory at the hardware level, and can further comprise other hardware required by business. The processor reads the corresponding computer program from the non-volatile memory into the memory and then runs to implement the steps of the power system partition method.

[0071] Those skilled in the art will appreciate that embodiments of the present application can be readily used as a method, a system or a computer program product. Accordingly, the present application can take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application can take the form of a computer program product on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROMs, optical storage devices, etc.) embodying computer readable program code.

[0072] The present application is described in reference to the flowchart illustrations and / or block diagrams of the methods, apparatus (systems) and computer program products according to embodiments of the application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general purpose computer, special purpose computer, embedded processor or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart illustrations and / or block diagrams block or blocks. Figure 1 one or more functions specified in the flowchart illustrations and / or block diagrams block or blocks. Figure 1 one or more functions specified in the flowchart illustrations and / or block diagrams block or blocks.

[0073] These computer program instructions can also be stored in a computer- readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instructions which implement the function specified in the flowchart illustrations and / or block diagrams block or blocks. Figure 1 one or more functions specified in the flowchart illustrations and / or block diagrams block or blocks. Figure 1 one or more functions specified in the flowchart illustrations and / or block diagrams block or blocks.

[0074] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart illustrations and / or block diagrams block or blocks. Figure 1 one or more functions specified in the flowchart illustrations and / or block diagrams block or blocks. ​ one or more functions specified in the flowchart illustrations and / or block diagrams block or blocks.

[0075] It should be noted that the above detailed description of the specific embodiments of the present application is not intended to limit the present application in any way. Thus, while the present application has been described above with a certain degree of particularity, one skilled in the art will understand that the present application can be carried out by other techniques and modifications are possible without departing from the spirit and scope of the present application. Accordingly, all such modifications are intended to be included within the scope of the present application. No limitation is intended to the scope of the claims based on any embodiment illustrated in the figures.

Claims

1. A method of power system partitioning, characterized by, The method comprises: constructing an adjacency matrix based on the topology of the power system, and calculating a similarity distance matrix DSC of the distribution structure connectivity of the nodes of the power system through the adjacency matrix; calculating a similarity distance matrix FRC based on the frequency response curves of the nodes of the power system; after normalization, the DSC and the FRC are weighted and fused to generate a comprehensive similarity measurement distance matrix FRCDSC; taking the FRCDSC as the partition basis, randomly selecting an initial clustering center, and dividing each node into the region where the clustering center with the minimum FRCDSC is located; updating the clustering center to the node with the minimum sum of the FRCDSC of the nodes in the region; iteratively updating until the clustering center is unchanged, and outputting the partition result.

2. The method of claim 1, wherein, The constructing of the adjacency matrix based on the topology of the power system comprises: abstracting the power system as a graph G(V, E), wherein the node set V corresponds to the nodes of the power system, and the edge set E corresponds to the power transmission lines; for any nodes i and j, when there is a directly connected power transmission line, W(i, j) = 1, otherwise W(i, j) = 0.

3. The method of claim 1, wherein, The calculating of the similarity distance matrix DSC of the distribution structure connectivity of the nodes of the power system through the adjacency matrix comprises: calculating a degree matrix D and a normalized Laplacian matrix L based on the adjacency matrix W; extracting the first R eigenvectors of the normalized Laplacian matrix L to form a matrix U; calculating the DSC distance of nodes i and j based on a preset formula to obtain the similarity distance matrix DSC; wherein the preset formula is: ； where R represents the number of columns in U, , represents an element in U.

4. The method of claim 1, wherein, The calculating of the similarity distance matrix FRC based on the frequency response curves of the nodes of the power system comprises: collecting the frequency response curves of the nodes after disturbance; applying a dynamic time warping algorithm to calculate the warping distance of the frequency response curves between any two nodes to construct the FRC.

5. The method of claim 1, wherein, After outputting the partition result, verification is further performed, comprising: based on the adjacency matrix, using the Floyd algorithm to verify whether there is an infinite distance between any two nodes in each region; when there is no infinite distance in all regions, it is determined that the distribution structure connectivity is satisfied.

6. The method of claim 5, wherein, The verification further comprises: A is the average value of the FRC distance between nodes in the calculation region in ; Computing the inter-region inter-node FRC distance average A out ; When A out > A in , it is determined that the frequency similarity is satisfied.

7. The method of claim 1, wherein, The iteratively updating until the clustering center is unchanged comprises: calculating the silhouette coefficient under the current number of clustering centers in each iteration; taking the number of partitions with the silhouette coefficient closest to 1 as the optimal number of partitions.

8. A real-time inertia partitioning device for power systems, characterized by, The device comprises: a construction module configured to construct an adjacency matrix based on the topology of the power system, and calculate a similarity distance matrix DSC of the distribution structure connectivity of the nodes of the power system through the adjacency matrix; and calculate a similarity distance matrix FRC based on the frequency response curves of the nodes of the power system; a fusion module configured to, after normalization, weight and fuse the DSC and the FRC to generate a comprehensive similarity measurement distance matrix FRCDSC; an optimization module configured to take the FRCDSC as the partition basis, randomly select an initial clustering center, divide each node into the region where the clustering center with the minimum FRCDSC is located, update the clustering center to the node with the minimum sum of the FRCDSC of the nodes in the region, iteratively update until the clustering center is unchanged, and output the partition result.

9. A computer-readable storage medium, characterized in that, The storage medium stores a computer program, and the computer program is executed by a processor to implement the method in any one of claims 1-7.

10. A computer device, comprising: A computer program product, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the method in any one of claims 1-7 when executing the program.

Images