A method, system, medium and program product for clustering division of a distribution network

Abstract

The present invention discloses a method, system, medium and program product for clustering division of a distribution network in the technical field of clustering division of an active distribution network. The method includes: quantifying and normalizing the capacity characteristics of nodes based on the load characteristics of nodes and the control parameters of a distributed photovoltaic system, and obtaining the similarity between nodes by weighted calculation; quantifying and normalizing the coupling degree between nodes according to the electrical relevance between nodes, and forming an adjacency matrix by combining the similarity and coupling degree between the nodes; taking the adjacency matrix as an input, iteratively optimizing the value of modularity through a Fast-Unfolding algorithm, and gradually merging nodes based on the maximization of modularity increment, so as to realize the clustering division of the distribution network. The method for clustering division of a distribution network provided by the present invention comprehensively considers the similarity and electrical coupling degree between nodes, comprehensively reflects the actual relevance of each node in the distribution network, and realizes more accurate clustering division.

Description

Technical Field

[0001] The present invention belongs to the technical field of cluster partitioning of active distribution networks, and particularly relates to a method, system, medium and program product for partitioning distribution network clusters. Background Art

[0002] In recent years, a new power system mainly based on new energy sources such as distributed photovoltaics and wind power is accelerating its construction, and its installed capacity continues to grow. However, due to the characteristics of distributed photovoltaics with small single-unit capacity, large quantity and wide distribution, the workload of individual node operation and control is relatively large, and it is difficult to adopt the traditional centralized control method. Therefore, it is urgent to consider adopting a control mode mainly based on clusters to better solve the regulation problem of distribution networks with high photovoltaic penetration.

[0003] With the large-scale access of new energy sources such as distributed photovoltaics, the operation mode and load characteristics of distribution networks have become more complex. The distribution network cluster partitioning technology can effectively divide the grid area, facilitate hierarchical management and flexible control, and thus improve the operation stability of the grid. When dealing with this change, the traditional distribution network cluster partitioning method has certain limitations: firstly, the considered factors are single, resulting in the partitioning result being difficult to comprehensively reflect the multi-dimensional correlation characteristics between nodes; secondly, it is sensitive to the initial data, and the number of partitions needs to be given artificially, resulting in a strong subjectivity of the partitioning result; thirdly, the existing heuristic algorithms have poor local search ability, are easy to fall into local optima, and have a slow convergence speed and high computational complexity, resulting in inaccurate partitioning results. Therefore, it is urgent to solve the above problems to improve the accuracy and rationality of cluster partitioning. Summary of the Invention

[0004] The purpose of the present invention is to overcome the deficiencies in the prior art and provide a method, system, medium and program product for partitioning distribution network clusters, which can effectively improve the regulation accuracy of distribution networks and meet the requirements of precise and flexible partition management in complex power environments.

[0005] To achieve the above purpose, the present invention is implemented by adopting the following technical solutions:

[0006] In the first aspect, the present invention provides a method for partitioning distribution network clusters, which is characterized by including:

[0007] Quantify and normalize the capacity characteristics of nodes based on the load characteristics of nodes and the control parameters of distributed photovoltaic systems, and calculate the similarity between nodes by weighted calculation;

[0008] Quantify and normalize the coupling degree between nodes according to the electrical relevance between nodes, and form an adjacency matrix by combining the similarity and coupling degree between nodes;

[0009] Taking the adjacency matrix as the input, iteratively optimize the value of modularity through the Fast-Unfolding algorithm, and gradually merge nodes based on the maximization of modularity increment until the modularity increment no longer increases, thereby realizing the cluster division of the distribution network.

[0010] Combined with the first aspect, further, based on the load characteristics of nodes and the control parameters of the distributed photovoltaic system, quantify and normalize the capacity characteristics of nodes, including:

[0011] According to the net load of source-load nodes under multiple sections and the control parameters of the distributed photovoltaic system, quantify and normalize the absorption capacity and regulation capacity of nodes. The control parameters include the proportional coefficient of the power outer loop controller of the photovoltaic inverter at the node and the integral coefficient ;

[0012] The net load of source-load nodes under multiple sections is calculated by the following formula:

[0013] ;

[0014] In the formula, represents the net load of node under multiple sections; represents the power of the load at node at time ; represents the power of the photovoltaic at node at time

[0015] The normalization process is calculated by the following formula:

[0016] ;

[0017] In the formula, represents the original value of the th index of node i, represents the value of the th index of node i after normalization.

[0018] Combined with the first aspect, further, the similarity between nodes is obtained by weighted calculation, including:

[0019] The eigenvalue of different nodes is weighted and calculated by the entropy weight method to obtain the weights corresponding to the three indicators of the net load of source-load nodes under multiple sections, the proportional coefficient of the power outer loop controller of the photovoltaic inverter and the integral coefficient ;

[0020] The calculation steps of the entropy weight method include data normalization, calculating the entropy value of indicators, and determining weights. The calculation formulas for each step are as follows:

[0021] Normalization processing:

[0022] ;

[0023] In the formula, represents the proportion of the th node in the th indicator; represents the total number of nodes;

[0024] Calculating information entropy:

[0025] ;

[0026] In the formula, represents the entropy value of the th indicator; is a constant used to ensure that the value of is in the range of [0, 1];

[0027] Calculating the difference coefficient:

[0028] ;

[0029] In the formula, the difference coefficient represents the degree of importance of the weight of the th indicator;

[0030] Calculating weights:

[0031] ;

[0032] In the formula, represents the weight of the th indicator;

[0033] After obtaining the weights of each indicator, the similarity between nodes is obtained through the following formula:

[0034] ;

[0035] ;

[0036] In the formula, represents the similarity between node and node ; represents the value of the th indicator of node j after normalization; R represents the maximum value after weighting the product of the indicators of any two nodes.

[0037] In combination with the first aspect, further, according to the electrical correlation between nodes, the coupling degree between nodes is quantified and normalized, including:

[0038] Linearize the power flow equation based on the net load of the source-load nodes under multiple sections:

[0039] ;

[0040] ;

[0041] In the formula, represents the change in the active power injected into the node; represents the change in the reactive power injected into the node; represents the change in the node voltage phase angle; represents the change in the node voltage amplitude; represents the Jacobian matrix; represents the relationship between the fluctuation of the active power injected into the node and the change in the node voltage phase angle; represents the relationship between the fluctuation of the active power injected into the node and the change in the node voltage amplitude; represents the relationship between the fluctuation of the reactive power injected into the node and the change in the node voltage phase angle; represents the relationship between the fluctuation of the reactive power injected into the node and the changes in the node voltage amplitude and phase angle;

[0042] Multiply both sides of the linearized power flow equation by , to obtain:

[0043] ;

[0044] ;

[0045] In the formula, represents the inverse matrix of the Jacobian matrix; respectively represent the changes in the node voltage phase angle when unit active and reactive powers are injected into the grid nodes, represents the active power sensitivity matrix, , representing the sensitivity of the node voltage in the grid to the change in active power; represents the reactive power sensitivity matrix, , representing the sensitivity of the node voltage in the grid to the change in reactive power;

[0046] Calculate the relative influence of the power change of node j on its own voltage change and on the voltage change of node i based on the active power sensitivity matrix and the reactive power sensitivity matrix:

[0047] ;

[0048] In the formula, Represents a node When the reactive power and active power change, the sum of the ratio of its own voltage change value to the voltage change value of the node ; Respectively represent the node Reactive and active node voltage sensitivities of itself; Respectively represent the node and the node Reactive and active node voltage sensitivities between;

[0049] Based on the relative influence between the above nodes, considering the node relationship, its own characteristics and the correlation with other nodes in the network, the electrical distance between node i and node j is defined as:

[0050] ;

[0051] For Normalize to get:

[0052] ;

[0053] In the formula, Represents the electrical distance between the normalized node and the node ; The electrical distance between nodes characterizes the coupling degree between nodes in the power grid.

[0054] Combined with the first aspect, further, an adjacency matrix is formed by comprehensively considering the similarity and coupling degree between the nodes, including:

[0055] The similarity and coupling degree between nodes are weighted and calculated to obtain the adjacency matrix. The elements of the adjacency matrix are calculated by the following formula:

[0056] ;

[0057] In the formula, Represents the weight of the edge connecting node and node ; Respectively represent the weights of similarity and electrical distance.

[0058] Combined with the first aspect, further, taking the adjacency matrix as the input, the value of modularity is iteratively optimized through the Fast - Unfolding algorithm, and nodes are gradually merged based on the maximization of modularity increment until the modularity increment no longer increases, thereby realizing the clustering division of the distribution network, including:

[0059] Regard each node in the distribution network as a separate cluster, and calculate the initial modularity value;

[0060] In each iteration, starting from each node i in the distribution network, the modularity increment after moving it to other modules is examined in turn ;

[0061] If > 0, then merge the node i with the clusters to which the corresponding adjacent nodes belong to form a new cluster;

[0062] After each merge, update the modularity of the distribution network and parameters, and calculate the modularity increment according to the updated parameters , and the parameters include the edge weight ratio between clusters and the degree ratio within clusters;

[0063] Continue the iterative merging process until there is no further modularity increment;

[0064] When the modularity of the entire distribution network can no longer increase, the algorithm terminates and outputs the final optimal cluster division result;

[0065] The calculation formula of the modularity Q is as follows:

[0066] ;

[0067] ;

[0068]

[0069] ;

[0070] In the formula, represents the modularity; represents the sum of the weights of all edges connected to the node ; represents the sum of the weights of all edges connected to the node j; represents the sum of the weights between all pairs of nodes in the network; is a 0-1 variable, representing the relationship between node i and node j, being 1 if in the same cluster and 0 otherwise;

[0071] The modularity increment is calculated as follows:

[0072] ;

[0073] ;

[0074] ;

[0075] In the formula, represents the edge weight ratio between clusters, obtained by calculating the ratio of the edge weight connected to cluster and cluster to the total edge weights in the network; represents the degree ratio within a cluster, obtained by calculating the ratio of the sum of the degrees of the nodes within cluster to the sum of the degrees of all nodes; represents the degree ratio within a cluster, obtained by calculating the ratio of the sum of the degrees of the nodes within cluster to the sum of the degrees of all nodes.

[0076] In a second aspect, the present invention provides a distribution network clustering system, including:

[0077] A similarity analysis module, configured to: based on the load characteristics of nodes and the control parameters of distributed photovoltaic systems, quantify and normalize the ability characteristics of nodes, and use weighted calculation to obtain the similarity between nodes;

[0078] A coupling degree analysis module, configured to: according to the electrical relevance between nodes, quantify and normalize the coupling degree between nodes;

[0079] A weighted fusion module, configured to: comprehensively form an adjacency matrix based on the similarity and coupling degree between the nodes;

[0080] A cluster division module, configured to: take the adjacency matrix as input, iteratively optimize the value of modularity through the Fast-Unfolding algorithm, and gradually merge nodes based on maximizing the modularity increment until the modularity increment no longer increases, thereby realizing the clustering of the distribution network.

[0081] In a third aspect, the present invention provides a computer-readable storage medium, on which a computer program is stored, characterized in that when the program is executed by a processor, the steps of the above-mentioned distribution network clustering method are realized.

[0082] In a fourth aspect, the present invention provides a computer program product, including computer programs / instructions, characterized in that when the computer programs / instructions are executed by a processor, the steps of the above-mentioned distribution network clustering method are realized.

[0083] Compared with the prior art, the present invention has the following beneficial effects:

[0084] 1. The distribution network clustering method provided by the present invention can more comprehensively reflect the actual relevance of each node in the distribution network by comprehensively considering the similarity and electrical coupling degree between nodes, and achieve more accurate clustering. In the present invention, the consumption and regulation capabilities of nodes are quantified and normalized through the load characteristics of nodes and the control parameters of distributed photovoltaic systems, so as to obtain the similarity between each node. At the same time, the electrical coupling degree is introduced to evaluate the physical connection relationship and electrical influence between nodes through electrical relevance. Combining these two dimensions, the generated adjacency matrix can fully express the multi-dimensional characteristics between each node.

[0085] Based on the similarity and coupling degree, an adjacency matrix is constructed and input into the Fast-Unfolding algorithm. The dynamic merging of nodes is realized by maximizing the modularity increment, avoiding falling into local optimal solutions, and improving the effect and accuracy of power grid clustering. Compared with the partitioning method based on a single feature only, the clustering scheme of the present invention improves the flexibility and accuracy of the distribution network, and significantly enhances the adaptability of distributed photovoltaic access and the power coordination performance between nodes.

[0086] 2. The distribution network clustering method provided by the present invention quantifies and normalizes the ability characteristics of each node by based on the load characteristics of nodes and the control parameters of distributed photovoltaic systems, making the clustering more comprehensive and accurate. Selecting the net load of source-load nodes and the control parameters of distributed photovoltaic systems under multiple cross-sections as indicators not only considers the net load characteristics of nodes under different working conditions, but also combines the proportional coefficient and integral coefficient of distributed photovoltaic inverters, accurately reflecting the consumption and regulation capabilities of nodes. This selection of multi-dimensional and multi-condition indicators enables the evaluation of the ability characteristics of nodes not to be limited to a specific state, but to take into account various operating conditions, thus providing a more representative and adaptable basis for clustering.

[0087] Through the fine quantization and normalization of node characteristics, the solution provided by the present invention effectively improves the accuracy of clustering, enabling the distribution network to better achieve load balancing and power regulation in a complex application environment, which helps to enhance the access ability of distributed photovoltaics and the operation stability of the distribution network. BRIEF DESCRIPTION OF THE DRAWINGS

[0088] Figure 1 is a flowchart of the power grid clustering method provided in Embodiment 1 of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0089] The technical solutions of the present invention will be described in detail below with reference to the drawings and specific embodiments. It should be understood that the specific features in the embodiments of the present application and the embodiments are detailed descriptions of the technical solutions of the present application, rather than limitations on the technical solutions of the present application. Without conflict, the technical features in the embodiments of the present application and the embodiments can be combined with each other.

[0090] The terms "first", "second", etc. are for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, features defined with "first", "second", etc. may explicitly or implicitly include one or more of such features. In the description of the present disclosure / application, unless otherwise stated, the meaning of "a plurality" is two or more.

[0091] Embodiment 1:

[0092] Figure 1 is a flowchart of a method for partitioning a distribution network cluster in Embodiment 1 of the present invention. This flowchart only shows the logical sequence of the method described in this embodiment. On the premise of non-conflict, in other possible embodiments of the present invention, the steps shown or described may be completed in a different Figure 1 order than that shown.

[0093] See Figure 1 , the method of this embodiment specifically includes the following steps:

[0094] Quantify and normalize the capacity characteristics of nodes based on the load characteristics of nodes and the control parameters of distributed photovoltaic systems, and calculate the similarity between nodes using weighted calculation;

[0095] Quantify and normalize the coupling degree between nodes according to the electrical relevance between nodes, and form an adjacency matrix by combining the similarity and coupling degree between the nodes;

[0096] Take the adjacency matrix as the input, iterate and optimize the value of modularity through the Fast-Unfolding algorithm, and gradually merge nodes based on the maximization of modularity increment until the modularity increment no longer increases, thereby realizing the partitioning of the distribution network into clusters.

[0097] Based on the load characteristics of nodes and the control parameters of distributed photovoltaic systems, that is, select the net load of source-load nodes under multiple sections, the proportional coefficient and integral coefficient of the power outer-loop controller of the photovoltaic inverter

[0098] as the indicators for cluster partitioning in the power grid.

[0099] Multi-section refers to the load conditions of nodes at different time periods or different operating states. For example, the light conditions are different in the morning, at noon, and in the evening, resulting in changes in the photovoltaic power generation, and the net load conditions will also change accordingly. Therefore, examining the net load of nodes under multiple "sections" can conduct a more comprehensive assessment.

[0100] An inverter is a device that converts the direct current generated by solar panels into alternating current. A distributed photovoltaic inverter refers to an inverter distributed in each photovoltaic power generation device. The control of the inverter has multiple levels, divided into an "inner loop" and an "outer loop". "Outer loop control" is usually used to adjust the overall output of the inverter, such as controlling the output voltage or frequency, so as to better cooperate with the power grid. These outer loop control parameters can help the system maintain stability under different photovoltaic power generation levels and avoid excessive fluctuations affecting the power grid. The "proportional coefficient" and the "integral coefficient" are two important parameters in the inverter control system, especially used to adjust the response speed and stability of the system in the control of the inverter.

[0101] The proportional coefficient Kp controls the "speed" of the system response. When the system detects a deviation, that is, the gap between the target value and the actual value, the controller will adjust the output by multiplying this deviation by the proportional coefficient Kp. In a photovoltaic inverter, the proportional coefficient Kp can help regulate the power output of the photovoltaic system and make it respond more quickly to changes in photovoltaic power generation.

[0102] The integral coefficient Ki is used to eliminate the "steady-state error". Even if the deviation value is very small, the integral controller will continuously accumulate the deviation and multiply it by Ki, gradually reducing the error until the target value is reached. In a photovoltaic inverter, the integral coefficient can help the system eliminate the long-term deviation between the photovoltaic output power and the expected power and improve the steady-state performance of the system.

[0103] Selecting the net load of source-load nodes and the control parameters of the distributed photovoltaic system under multiple sections as indicators not only considers the net load characteristics of nodes under different working conditions but also combines the proportional coefficient and integral coefficient of the distributed photovoltaic inverter, accurately reflecting the absorption and regulation capabilities of nodes. This selection of multi-dimensional and multi-condition indicators enables the assessment of the capacity characteristics of nodes not to be limited to a certain specific state but to take into account multiple operating conditions, thus providing a more representative and adaptable basis for cluster division.

[0104] The net load of source-load nodes under multiple sections is calculated by the following formula:

[0105] ;

[0106] In the formula, represents the net load of node under multiple sections; represents the node at time Power of the load at Indicates Moment node Power of the photovoltaic at Represents a time period.

[0107] After obtaining the net load value, proportionality coefficient Kp, and integral coefficient Ki value of each node in the power grid, normalize them through the following formula:

[0108] ;

[0109] In the formula, Indicates the original value of the th index of node i, Indicates the value of the th index of node i after normalization.

[0110] Calculate the eigenvalue of different nodes by weighted calculation through the entropy weight method, and obtain the net load of the source-load nodes, the proportionality coefficient and integral coefficient of the power outer loop controller of the photovoltaic inverter

[0111] The calculation steps of the entropy weight method include data normalization, calculation of index entropy value, and determination of weight. The calculation formulas for each step are as follows:

[0112] Normalization processing:

[0113] ;

[0114] In the formula, Indicates the proportion of the th node in the th index; Indicates the total number of nodes;

[0115] Calculation of information entropy:

[0116] ;

[0117] In the formula, Indicates the entropy value of the th index; Is a constant used to ensure that the value of is in the range of [0,1];

[0118] Calculation of the difference coefficient:

[0119] ;

[0120] In the formula, the difference coefficient Indicates the The weight importance degree of each index;

[0121] Calculate the weights:

[0122] ;

[0123] In the formula, represents the weight of the th index;

[0124] After obtaining the weights of each index, the similarity between nodes is obtained through the following formula:

[0125] ;

[0126] ;

[0127] In the formula, represents the similarity between node and node ; represents the value of the th index of node j after normalization; R is the maximum value after weighting the product of the indexes of any two nodes.

[0128] After obtaining the similarity between nodes, further quantify and normalize the coupling degree between nodes according to the electrical correlation between nodes.

[0129] First, linearize the power flow equation based on the net load of the source-load nodes under multiple sections:

[0130] ;

[0131] ;

[0132] In the formula, represents the change in active power injection at the node; represents the change in reactive power injection at the node; represents the change in the phase angle of the node voltage; represents the change in the amplitude of the node voltage; represents the Jacobian matrix; represents the relationship between the fluctuation of active power injection at the node and the change in the phase angle of the node voltage; represents the relationship between the fluctuation of active power injection at the node and the change in the amplitude of the node voltage; represents the relationship between the fluctuation of reactive power injection at the node and the change in the phase angle of the node voltage; represents the relationship between the fluctuation of reactive power injection at the node and the changes in the amplitude and phase angle of the node voltage;

[0133] Active power is the part that actually does work in the power system and is used to drive motors, lighting equipment, etc. Active power directly affects the energy transmission and consumption of the power system.

[0134] Reactive power is the part that does not do actual work in the power system and is used to establish and maintain electric and magnetic fields. Reactive power mainly affects the voltage level and the stability of the power system. The sensitivity matrix is used to describe how the change of certain variables in the system affects other variables. In this embodiment, the active sensitivity matrix and the reactive sensitivity matrix are mainly used to describe how the changes of active power and reactive power affect the voltage. In the distribution network, the sensitivity matrix helps engineers understand and predict the system's response to different operations, so as to optimize the control strategy.

[0135] Multiply both sides of the linearized power flow equation by to obtain the active sensitivity matrix and the reactive sensitivity matrix:

[0136] ;

[0137] ;

[0138] In the formula, represents the inverse matrix of the Jacobian matrix; respectively represent the changes in the phase angles of the node voltages when injecting unit active and reactive powers into the grid nodes, represents the active sensitivity matrix, , representing the sensitivity of the node voltage in the grid to the change in active power; represents the reactive sensitivity matrix, , representing the sensitivity of the node voltage in the grid to the change in reactive power;

[0139] Calculate the relative influence of the power change of node j on its own voltage change and on the voltage change of node i based on the active sensitivity matrix and the reactive sensitivity matrix:

[0140] ;

[0141] In the formula, represents the sum of the ratios of the change in its own voltage value to the change in the voltage value of node when the reactive power and active power change; respectively represent the reactive and active node voltage sensitivities of node ; its own reactive and active node voltage sensitivities; respectively represent the reactive and active node voltage sensitivities between node and node ;

[0142] Based on the relative influence between the above nodes, considering the relevance of node relationships to their own characteristics and other nodes in the network, define the electrical distance between node i and node j as:

[0143] ;

[0144] For perform normalization to obtain:

[0145] ;

[0146] In the formula, represents the electrical distance between node and node after normalization; the electrical distance between nodes characterizes the coupling degree between nodes in the power grid.

[0147] Based on the similarity and coupling degree between nodes obtained above, assign different weights to the similarity and coupling degree between nodes and then add them to obtain the adjacency matrix. The elements of the adjacency matrix are calculated by the following formula:

[0148] ;

[0149] In the formula, represents the weight of the edge connecting node and node ; respectively represent the weights of similarity and electrical distance.

[0150] In this embodiment, the absorption and regulation capabilities of nodes are quantified and normalized through the load characteristics of nodes and the control parameters of the distributed photovoltaic system, so as to obtain the similarity between nodes. At the same time, the electrical coupling degree is introduced to evaluate the physical connection relationship and electrical influence between nodes through electrical relevance. Combining these two dimensions, the generated adjacency matrix fully expresses the multi-dimensional characteristics between nodes, making the subsequent cluster division more comprehensive and accurate.

[0151] Taking the obtained adjacency matrix as the input, iteratively optimize the modularity value through the Fast - Unfolding algorithm, and gradually merge nodes based on the maximization of the modularity increment until the modularity increment no longer increases, so as to realize the cluster division of the distribution network, including:

[0152] Regard each node in the distribution network as a separate cluster and calculate the initial modularity value;

[0153] In each iteration, starting from each node i in the distribution network, successively examine the modularity increment ;

[0154] If > 0, then merge the node i with the cluster to which the corresponding adjacent node belongs to form a new cluster;

[0155] After each merge, update the modularity and parameters of the distribution network, and calculate the modularity increment according to the updated parameters. The parameters include the edge weight ratio between clusters and the degree ratio within clusters;

[0156] Continue to iterate the merging process until there is no further modularity increment;

[0157] When the modularity of the entire network cannot be increased anymore, the algorithm terminates and outputs the final optimal cluster division result;

[0158] The calculation formula of the modularity Q is as follows:

[0159] ;

[0160] ;

[0161]

[0162] ;

[0163] In the formula, represents the modularity; represents the sum of the weights of all edges connected to the node ; represents the sum of the weights of all edges connected to the node j; represents the sum of the weights between all pairs of nodes in the network; is a 0-1 variable, representing the relationship between node i and node j. It is 1 if they are in the same cluster, otherwise 0;

[0164] The modularity increment is calculated as follows:

[0165] ;

[0166] ;

[0167] ;

[0168] In the formula, represents the edge weight ratio between clusters, which is obtained by calculating the ratio of the edge weight connected between cluster and cluster to the total edge weight in the network; represents the degree ratio within clusters, which is obtained by calculating the cluster Obtained by the ratio of the sum of the degrees of the internal nodes to the sum of the degrees of all nodes; Denote the degree ratio within the cluster, which is obtained by calculating the ratio of the sum of the degrees of the internal nodes to the sum of the degrees of all nodes within the cluster Obtained by the ratio of the sum of the degrees of the internal nodes to the sum of the degrees of all nodes.

[0169] The Fast - Unfolding algorithm iteratively optimizes the cluster partition in the direction of maximizing the modularity increment. By gradually optimizing the modularity increment, this algorithm can find the global optimum or a partition closer to the optimum more smoothly, avoiding getting trapped in the local optimum at an early stage. At each update, it determines whether to continue merging based on the modularity increment, reducing unnecessary merging operations and enhancing the stability of the partition result. And the continuously updated edge weight sum and ratio ensure the real - time accuracy of the modularity. This update mechanism is more in line with the real community structure of the network than the static calculation of edge weights, making the partition result more accurate.

[0170] In summary, the Fast - Unfolding algorithm takes the modularity increment as the optimization objective and dynamically updates the edge weight ratio, making the partition of the power grid clusters more stable, effectively avoiding the local optimum problem, and thus obtaining a partition result closer to the global optimum, making the partition result not only accurate but also having good generalization and applicability.

[0171] Example Two:

[0172] Based on the power distribution network cluster partition method described in Example One, this example provides a power distribution network cluster partition system, including:

[0173] Similarity analysis module, configured to: Quantify and normalize the ability characteristics of nodes based on the load characteristics of nodes and the control parameters of the distributed photovoltaic system, and obtain the similarity between nodes by weighted calculation;

[0174] Coupling degree analysis module, configured to: Quantify and normalize the coupling degree between nodes according to the electrical relevance between nodes;

[0175] Weighted fusion module, configured to: Synthesize the similarity and coupling degree between the nodes to form an adjacency matrix;

[0176] Cluster partition module, configured to: Take the adjacency matrix as the input, iteratively optimize the value of the modularity through the Fast - Unfolding algorithm, and gradually merge nodes based on the maximization of the modularity increment until the modularity increment no longer increases, thereby realizing the cluster partition of the power distribution network.

[0177] Example Three:

[0178] An embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, it first implements the steps of the distribution network cluster division method described in the above-mentioned Embodiment 1.

[0179] The computer-readable storage medium includes: various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical discs.

[0180] Embodiment 4:

[0181] An embodiment of the present invention further provides a computer program product, including a computer program / instructions. When the computer program / instructions are executed by a processor, they implement the steps of the distribution network cluster division method described in Embodiment 1.

[0182] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memories, CD-ROMs, optical memories, etc.) containing computer-usable program codes.

[0183] The present application is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or block in the flowchart and / or block diagram can be implemented by computer program instructions, and the combination of processes and / or blocks in the flowchart and / or block diagram can also be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate a device for implementing the specified functions in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0184] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory generate a manufactured product including an instruction device, and the instruction device implements the specified functions in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0185] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus, so that a series of operation steps are executed on the computer or other programmable apparatus to generate a computer-implemented process, thereby providing instructions for implementing the steps specified in one process or a plurality of processes and / or boxes Figure 1 one process or a plurality of processes and / or boxes Figure 1 steps for implementing the functions specified in one box or a plurality of boxes.

[0186] The above are only the preferred embodiments of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the technical principle of the present invention, several improvements and modifications can be made, and these improvements and modifications should also be regarded as the protection scope of the present invention.

Claims

1. A method for partitioning a distribution network cluster, characterized in that Including: Quantify and normalize the capacity characteristics of nodes based on the load characteristics of nodes and the control parameters of the distributed photovoltaic system, and calculate the similarity between nodes by weighted calculation; Quantify and normalize the coupling degree between nodes according to the electrical correlation between nodes, and form an adjacency matrix by synthesizing the similarity and coupling degree between the nodes; Take the adjacency matrix as the input, iteratively optimize the value of modularity through the Fast-Unfolding algorithm, and gradually merge nodes based on the maximization of modularity increment until the modularity increment no longer increases, thereby realizing the clustering division of the distribution network; Among them, quantifying and normalizing the capacity characteristics of nodes based on the load characteristics of nodes and the control parameters of the distributed photovoltaic system includes: Quantify and normalize the absorption capacity and regulation capacity of nodes according to the net load of source-load nodes under multiple cross-sections and the control parameters of the distributed photovoltaic system, where the control parameters include the proportional coefficient of the power outer-loop controller of the photovoltaic inverter at the node and the integral coefficient ; The net load of the source-load nodes under multiple sections is calculated by the following formula: ； In the formula, represents the net load of the node under multiple cross-sections ; represents the power of the load at the node at time ; represents the power of the photovoltaic at the node at time ; represents a time period. The normalization process is calculated by the following formula: ； In the formula, represents the original value of the -th index of node i, represents the value of the -th index of node i after normalization; Calculating the similarity between nodes by weighted calculation includes: The eigenvalue of different nodes is calculated by the entropy weight method to obtain the net load of the source-load nodes under multiple cross-sections, the proportional coefficient of the power outer loop controller of the photovoltaic inverter, and the integral coefficient. The weights corresponding to these three indicators; The calculation steps of the entropy weight method include data normalization, calculation of index entropy value, and determination of weight. The calculation formulas for each step are as follows: Normalization process: ； In the formula, represents the proportion of node i in the th index; represents the total number of nodes; Calculating information entropy: ； In the formula, represents the entropy value of the th index; is a constant used to ensure that the value of is within the range of [0, 1]; Calculating the coefficient of variation: ； Wherein, the difference coefficient represents the weight importance degree of the th index; Calculating weight: ； wherein, represents the weight of the th index; After obtaining the weights of each index, the similarity between nodes is obtained through the following formula: ； ； In the formula, represents the similarity between nodes and node ; represents the value of the -th index of node j after normalization; R represents the maximum value after weighting the product of the indexes of any two nodes.

2. The distribution network cluster division method according to claim 1, wherein Quantifying and normalizing the coupling degree between nodes according to the electrical correlation between nodes includes: Linearize the power flow equation based on the net load of the source-load nodes under multiple sections; ； ； In the formula, represents the change in active power injected at the node; represents the change in reactive power injected at the node; represents the change in the phase angle of the node voltage; represents the change in the amplitude of the node voltage; represents the Jacobian matrix; represents the relationship between the fluctuation of active power injected at the node and the change in the phase angle of the node voltage; represents the relationship between the fluctuation of active power injected at the node and the change in the amplitude of the node voltage; represents the relationship between the fluctuation of reactive power injected at the node and the change in the phase angle of the node voltage; represents the relationship between the fluctuation of reactive power injected at the node and the changes in the amplitude and phase angle of the node voltage; Multiply both sides of the linearized power flow equation by , and we get: ； ； In the formula, represents the inverse matrix of the Jacobian matrix; respectively represent the changes in the node voltage phase angles when injecting unit active and reactive powers into the power grid nodes, represents the active power sensitivity matrix, , representing the sensitivity of the node voltage in the power grid to the change in active power; represents the reactive power sensitivity matrix, , representing the sensitivity of the node voltage in the power grid to the change in reactive power; Calculate the relative influence of the power change of node j on its own voltage change and on the voltage change of node i based on the active sensitivity matrix and the reactive sensitivity matrix; ； In the formula, represents a node which is the sum of the ratio of the self-voltage change value when the reactive power and active power change to the voltage change value of node ; respectively represent the self-reactive and active node voltage sensitivities of node ; respectively represent the reactive and active node voltage sensitivities between node and node ; Based on the relative influence between the above nodes, considering the node relationship, its own characteristics, and the correlation with other nodes in the network, the electrical distance between node i and node j is defined as follows as follows: ； Pair After normalization, we get: ； In the formula, represents the electrical distance between the normalized node and the node ; the electrical distance between nodes characterizes the coupling degree between nodes in the power grid.

3. The distribution network cluster division method according to claim 2, wherein Forming an adjacency matrix by synthesizing the similarity and coupling degree between the nodes includes: Perform weighted calculation on the similarity and coupling degree between nodes to obtain the adjacency matrix. The elements of the adjacency matrix are calculated by the following formula: ； In the formula, represents the weight of the edge between the connection node and the node ; respectively represent the weights of the similarity and the electrical distance.

4. The distribution network cluster division method according to claim 3, characterized in that, Taking the adjacency matrix as the input, iteratively optimize the value of modularity through the Fast-Unfolding algorithm, and gradually merge nodes based on the maximization of modularity increment until the modularity increment no longer increases, thereby realizing the clustering division of the distribution network, including: Treat each node in the distribution network as a separate cluster and calculate the initial modularity value; In each iteration, starting from each node i in the distribution network, the increment of modularity after moving it into the cluster to which other adjacent nodes belong is examined in turn. ; If > 0, then merge the node i with the cluster to which the corresponding adjacent node belongs to form a new cluster; After each merge, update the modularity of the distribution network and parameters, and calculate the modularity increment according to the updated parameters , where the parameters include the edge weight ratio between clusters and the degree ratio within clusters; Continue the iterative merging process until there is no further modularity increment; When the modularity of the entire power distribution network can no longer be increased, the algorithm terminates and outputs the final optimal cluster division result; The calculation formula of modularity Q is as follows: ； ； ； ； In the formula, represents modularity; represents the sum of the weights of all edges connected to node ; represents the sum of the weights of all edges connected to node j; represents the sum of the weights between all pairs of nodes in the network; is a 0-1 variable, representing the relationship between node i and node j, which is 1 if they are in the same cluster and 0 otherwise; Modularity increment The calculation formula is as follows: ； ； ； In the formula, represents the edge weight ratio between clusters, which is obtained by calculating the ratio of the edge weight connecting cluster and cluster to the sum of all edge weights in the network; represents the degree ratio within a cluster, which is obtained by calculating the ratio of the sum of the degrees of the nodes in cluster to the sum of the degrees of all nodes; represents the degree ratio within a cluster, which is obtained by calculating the ratio of the sum of the degrees of the nodes in cluster to the sum of the degrees of all nodes.

5. A distribution network cluster division system, characterized in that, Including: A similarity analysis module configured to: quantify and normalize the capacity characteristics of nodes based on the load characteristics of nodes and the control parameters of the distributed photovoltaic system, and calculate the similarity between nodes by weighted calculation; Among them, quantifying and normalizing the capacity characteristics of nodes based on the load characteristics of nodes and the control parameters of the distributed photovoltaic system includes: Quantify and normalize the absorption capacity and regulation capacity of nodes according to the net load of source-load nodes under multiple cross-sections and the control parameters of the distributed photovoltaic system, where the control parameters include the proportional coefficient of the power outer-loop controller of the photovoltaic inverter at the node and the integral coefficient ; The net load of the source-load nodes under multiple sections is calculated by the following formula: ； Wherein, represents the net load of the node under multiple cross-sections ; represents the power of the load at the node at time ; represents the power of the photovoltaic at the node at time ; represents a time period; The normalization process is calculated by the following formula: ； In the formula, represents the original value of the -th index of node i, represents the value of the -th index of node i after normalization; Calculating the similarity between nodes by weighted calculation includes: The eigenvalue of different nodes is calculated by entropy weight method to obtain the net load of the source-load nodes under multiple sections, the proportional coefficient of the power outer-loop controller of the photovoltaic inverter and the integral coefficient The weights corresponding to these three indicators; The calculation steps of the entropy weight method include data normalization, calculation of index entropy value, and determination of weight. The calculation formulas for each step are as follows: Normalization process: ； In the formula, represents the proportion of node i in the th index; represents the total number of nodes; Calculating information entropy: ； In the formula, represents the entropy value of the th index; is a constant used to ensure that the value of is within the range of [0, 1]; Calculating the coefficient of variation: ； Wherein, the difference coefficient represents the weight importance degree of the th index; Calculating weight: ； In the formula, represents the weight of the th index; After obtaining the weights of each index, the similarity between nodes is obtained through the following formula: ； ； In the formula, represents the similarity between nodes and node ; represents the value of the th index of node j after normalization; R represents the maximum value after weighting the product of the indexes of any two nodes; A coupling degree analysis module configured to: quantify and normalize the coupling degree between nodes according to the electrical correlation between nodes; The weighted fusion module is configured to: synthesize the similarity and coupling degree between the nodes to form an adjacency matrix; The cluster division module is configured to: take the adjacency matrix as an input, iteratively optimize the value of modularity through the Fast-Unfolding algorithm, and gradually merge nodes based on maximizing the modularity increment until the modularity increment no longer increases, thereby realizing the cluster division of the distribution network.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by a processor, it implements the steps of the distribution network cluster division method according to any one of claims 1 to 4.

7. A computer program product comprising a computer program / instructions, characterized in that, When the computer program / instructions are executed by a processor, it implements the steps of the distribution network cluster division method according to any one of claims 1 to 4.

Images