A method and apparatus for dividing a power distribution network

Abstract

This invention provides a distribution network partitioning method and apparatus, relating to the field of power system planning technology. The invention employs the DPC algorithm to determine cluster centers and the number of partitions based on relative distance and local density, obtaining an initial distribution network partitioning network diagram. The LPA algorithm is then used to determine the distribution network partitioning results based on the initial distribution network partitioning network diagram. This invention ensures the objectivity and scientific rigor of distribution network partitioning initialization and solves problems such as partitioning difficulties and giant partitions, ensuring the rationality of distribution network partitioning.

Description

Technical Field

[0001] This invention relates to the field of power system planning technology, and specifically to a distribution network zoning method and apparatus. Background Technology

[0002] As the proportion of new energy sources in new power systems increases, countries around the world are accelerating the development of new energy sources such as wind power and photovoltaics. As a crucial part directly connected to users at the end of the power grid, the distribution network urgently needs transformation and upgrading as the penetration rate of new energy sources rises.

[0003] When upgrading the distribution network configuration, zoning can simplify the network structure, control short-circuit current, reduce computational complexity, and lower the difficulty of operation. It also plays an important role in power grid planning and optimization, which is conducive to load forecasting, asset management and fault location, and helps to upgrade new distribution equipment.

[0004] Distribution network zoning algorithms mainly include: heuristic algorithms, which rely on experts to zonify based on multiple factors, are subjective, and are difficult to guarantee optimal results; physical characteristic-based algorithms are divided into three categories: first, autonomous zoning based on switch boundaries, which is suitable for areas containing distributed renewable energy; second, clustering algorithms based on network physical characteristics to ensure close nodes and loose connections within the zoning; and third, using complex network theory to divide the community structure; and scientific planning algorithms, which transform the zoning problem into a multi-objective optimization problem, comprehensively find a balance point by integrating multiple performance indicators to achieve efficient optimization of zoning.

[0005] The current problems with power distribution network partitioning are: under distributed resource access, the partitioning criteria do not take into account the impact of distributed power generation output; partitioning initialization is too subjective, and the effect is highly dependent on the initialization; due to the constraints of the partitioning method, there are problems such as partitioning difficulties and giant partitions. Summary of the Invention

[0006] The purpose of this invention is to provide a distribution network partitioning method and apparatus to address the following issues: existing distribution network partitioning methods do not consider the impact of distributed power generation output; partitioning initialization is too subjective, and the effectiveness of partitioning is highly dependent on the quality of partitioning initialization; due to the constraints of the partitioning method, problems such as partitioning difficulties and giant partitions may arise. This invention can ensure the objectivity and scientific nature of distribution network partitioning initialization, and solve the problems of partitioning difficulties and giant partitions, thus ensuring the rationality of distribution network partitioning.

[0007] To achieve the above objectives, in a first aspect, the present invention provides a distribution network zoning method, comprising:

[0008] Determine the output power of distributed generation sources in the distribution network;

[0009] Calculate the electrical distance between any two nodes in the distribution network based on the node's location in the distribution network and the impact of changes in the output power of distributed power sources on the node.

[0010] The relative distance and local density of nodes are determined based on electrical distance;

[0011] The DPC algorithm is used to determine the cluster centers and the number of partitions based on relative distance and local density, thus obtaining the initial distribution network partition network diagram.

[0012] The LPA algorithm is used to determine the distribution network partitioning results based on the initialized distribution network partitioning diagram.

[0013] According to a distribution network zoning method provided by the present invention, calculating the electrical distance between any two nodes in a distribution network includes:

[0014] Calculate the impedance between any two nodes;

[0015] The Jacobian matrix is ​​constructed by power flow calculation, and the corrected equation of the distribution network is obtained by solving the Newton-Raphson iterative method, thereby obtaining different types of sensitivity matrices;

[0016] Determine the node voltage change based on different types of sensitivity matrices;

[0017] To address the node voltage variations caused by distributed generation access to the distribution network, the electrical distance between any two nodes is determined by comprehensively considering the node's location in the distribution network and the impact of output power variations of the accessed distributed generation node on the node.

[0018] According to a power distribution network partitioning method provided by the present invention, the expression for the electrical distance between any two nodes is:

[0019] (10)

[0020] in, (4)

[0021] (8)

[0022] (9)

[0023] In the formula, Let Re be the electrical distance between nodes i and j, and let Re represent the real part and Im represent the imaginary part. This represents the element in the i-th row and j-th column of the impedance matrix of the distribution network nodes. This represents the impedance between nodes i and j. and The reactive voltage sensitivity and active voltage sensitivity between nodes i and j are respectively represented. The superscripts ii, jj, ij, and ji in equations (8) and (9) represent different transmission directions.

[0024] According to a distribution network zoning method provided by the present invention, the corrected equation for the distribution network is as follows:

[0025] (5)

[0026] In the formula, , , and Here are the Jacobian matrices of the corresponding variables in the distribution network, representing the active power P as a function of the phase angle. The rate of change of active power P with voltage amplitude V, and the rate of change of reactive power Q with phase angle The rate of change of reactive power Q with voltage amplitude V;

[0027] Different types of sensitivity matrices are represented as follows:

[0028] (6)

[0029] In the formula, the sensitivity matrix Represents the change in phase angle at the node Changes in active power injected by the node The influence of sensitivity matrix Represents the change in phase angle at the node Changes in reactive power injected by the node The influence of sensitivity matrix Represents the change in node voltage Changes in active power injected by the node The influence of sensitivity matrix Represents the change in node voltage Changes in reactive power injected by the node The impact;

[0030] Node voltage change The relationship between the power change and the power change at all nodes in the distribution network is as follows:

[0031] (7)

[0032] in, , N represents the total number of nodes in the distribution network. This represents the change in active power at node N. This represents the change in reactive power at node N.

[0033] According to a distribution network partitioning method provided by the present invention, the DPC algorithm is used to determine the cluster centers and the number of partitions based on relative distance and local density, thereby obtaining an initial distribution network partitioning network diagram, including:

[0034] First, a density peak decision map is plotted using local density and relative distance as the horizontal and vertical axes to determine the number of clusters and their centers. Then, a preset allocation distance is set to assign nodes that meet the conditions to their corresponding cluster centers, while nodes that do not meet the conditions are not assigned, thus obtaining the initial distribution network zoning network diagram.

[0035] According to a distribution network zoning method provided by the present invention, the LPA algorithm is used to determine the distribution network zoning result based on the initialized distribution network zoning network diagram, including:

[0036] Step 1: Determine the initial labeled nodes, initial unlabeled nodes, and label information based on the initialized distribution network partition network diagram, and determine the edge weights by the distances of the edges in the initialized distribution network partition network diagram;

[0037] Step 2: Calculate the probability matrix of label information transmission between nodes based on the edge weights;

[0038] Step 3: Define the matrix that records the label classification information of all nodes in the distribution network as the label classification matrix;

[0039] Step 4: The label information transmitted by all nodes is propagated according to the propagation probability. A label influence matrix is ​​defined to represent the impact of this round of propagation on the label information update of each node. After the propagation is completed, the values ​​of the initial labeled nodes are restored.

[0040] Step 5: Repeat the transfer process until the label classification matrix remains unchanged;

[0041] Step 6: Calculate the weights of edges between different nodes based on electrical distance;

[0042] Step 7: Calculate the partition size matrix, load balance matrix, and historical influence factors, and then calculate and improve the transmission probability matrix;

[0043] Step 8: Transfer the tag information. Taking into account the partition size matrix, load balance matrix, historical influence factors and tag influence matrix, update the tag classification matrix to complete one round of transfer.

[0044] Step 9: After the current round of transmission ends, determine whether the label classification matrix of this round is the same as that of the previous round. If the label classification matrix is ​​different, return to step 7; if the label classification matrix of this round is the same, output the distribution network zoning result.

[0045] In a second aspect, the present invention provides a power distribution network zoning device, comprising:

[0046] The first determining unit is used to determine the output power of distributed generation in the distribution network;

[0047] The calculation unit is used to calculate the electrical distance between any two nodes in the distribution network based on the node's location in the distribution network and the impact of changes in the output power of distributed power sources on the node.

[0048] The second determining unit is used to determine the relative distance and local density of nodes based on electrical distance;

[0049] The initialization unit is used to determine the cluster centers and number of partitions based on the relative distance and local density using the DPC algorithm, thereby obtaining the initial distribution network partition network diagram.

[0050] The partitioning unit is used to determine the distribution network partitioning result based on the initial distribution network partitioning network diagram using the LPA algorithm.

[0051] The technical solution of the present invention has at least the following technical effects:

[0052] 1. When dividing the distribution network, common meteorological factors in the area are considered to calculate the common output of distributed power sources. Simultaneously, the common output and node location are combined to create an electrical distance index. This approach is applicable to the current development trend of large-scale integration of distributed resources into the distribution network, improving the scientific accuracy of electrical distance calculations.

[0053] 2. During partition initialization, the DPC algorithm can objectively determine the cluster centers and number of partitions through a visualized decision graph, and can also roughly classify the remaining nodes, ensuring the objectivity and scientific nature of the distribution network partition initialization.

[0054] 3. When using the LPA algorithm to partition the initialization results, introducing the partition size matrix, load balance matrix and historical influence factor can make the partition results meet the partition size balance and load balance principles, and also have a better modularity evaluation index. Attached Figure Description

[0055] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0056] In the attached diagram:

[0057] Figure 1 This is a diagram of the low-voltage AC / DC power distribution system in this invention;

[0058] Figure 2 This is the density peak decision map of the low-voltage AC / DC distribution network based on the DPC algorithm in this invention;

[0059] Figure 3 This is the final zoning result diagram of the low-voltage AC / DC distribution network in this invention;

[0060] Figure 4 This is a flowchart of the power distribution network zoning method of the present invention. Detailed Implementation

[0061] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0062] The following detailed description of some embodiments of the present invention will be provided in conjunction with the accompanying drawings. Unless otherwise specified, the following embodiments and features can be combined with each other.

[0063] Please see Figure 4 This invention provides a method for dividing a power distribution network, comprising the following steps:

[0064] Step 100: Determine the output power of distributed generation in the distribution network;

[0065] Specifically, historical output data of distributed power sources are obtained, and cluster analysis is performed on the historical output data. Based on the capacity of the distributed power sources, the output power of wind turbines and photovoltaic generators is determined.

[0066] In the power output model construction of a wind farm, the output power of the wind turbine is... The functional relationship with wind speed is approximately as follows:

[0067] (1)

[0068] In the formula, This refers to the cut-in wind speed of the wind turbine. The rated wind speed of the wind turbine unit. This refers to the cut-off wind speed of the wind turbine. denoted as the rated output power of the wind turbine, and v as the wind speed.

[0069] In the construction of the photovoltaic power output model, the functional relationship between the output power of the photovoltaic generator and the solar irradiance is approximated as follows:

[0070] (2)

[0071] In the formula, Indicates the rated power of the photovoltaic generator set. denoted as rated solar irradiance, and s as solar irradiance.

[0072] The technical solution of the present invention will be described below with reference to a specific embodiment. First, the IEEE 69-node distribution network is adjusted to form a medium- and low-voltage AC / DC distribution system connected by Electric Energy Routers (EERs), such as... Figure 1 As shown in Table 1.

[0073]

[0074] Based on the rated power of distributed generation in the distribution network and the common meteorological conditions in the region, the planned output power of distributed generation for subsequent zoning calculations is calculated, and the calculation results are shown in Table 2.

[0075]

[0076] Step 200: Calculate the electrical distance between any two nodes in the distribution network based on the node's location in the distribution network and the impact of changes in the output power of distributed power sources on the node.

[0077] Specifically, calculating the electrical distance between any two nodes in a distribution network includes:

[0078] Calculate the impedance between any two nodes;

[0079] The Jacobian matrix is ​​constructed by power flow calculation, and the corrected equation of the distribution network is obtained by solving the Newton-Raphson iterative method, thereby obtaining different types of sensitivity matrices;

[0080] The node voltage change is determined based on the different types of sensitivity matrices.

[0081] To address the node voltage variations caused by distributed generation access to the distribution network, the electrical distance between any two nodes is determined by comprehensively considering the node's location in the distribution network and the impact of output power variations of the accessed distributed generation node on the node.

[0082] Based on the Jacobian matrix constructed using the power flow model, it can be seen that the corrected equations for both AC and DC distribution networks obtained by the Newton-Raphson method can be expressed as follows:

[0083] (5)

[0084] In the formula, , , and Let be the Jacobian matrices of the corresponding variables in AC and DC distribution networks, representing the active power P as a function of the phase angle. The rate of change of active power P with voltage amplitude V, and the rate of change of reactive power Q with phase angle The rate of change of reactive power Q with respect to voltage amplitude V. Inverting the Jacobian matrix yields different types of sensitivity matrices:

[0085] (6)

[0086] In the formula, the sensitivity matrix , , , These represent the changes in the phase angle at the nodes, respectively. or node voltage change Changes in active power injected by the node or change in reactive power The impact;

[0087] As can be seen from the above, in the presence of In a distribution network with 1 node, the voltage variation at each node It has a certain relationship with the power change of all nodes in the distribution network, and can be expressed as:

[0088] (7)

[0089] In the formula, and The matrix form representing the power changes at all nodes in the distribution network, i.e. , N is the total number of nodes in the distribution network. This represents the change in active power at node N. This represents the change in reactive power at node N.

[0090] As shown in equation (7), in a distribution network with distributed generation (DG) access, the voltage impact on the node caused by changes in the output power of DG connected to other nodes needs to be considered. Therefore, for the node voltage changes caused by DG access in the distribution network, an electrical distance based on voltage sensitivity is defined, which can be expressed as:

[0091] (8)

[0092] (9)

[0093] In distribution networks with distributed generation integration, it is necessary to comprehensively consider the node's location within the network and the impact of changes in the output of other nodes connected to distributed resources on that node. and , To achieve uniformity, the integrated electrical distance is therefore redefined. , can be represented as:

[0094] (10)

[0095] In the formula, Let Re be the electrical distance between nodes i and j, and let Re represent the real part and Im represent the imaginary part. This represents the element in the i-th row and j-th column of the impedance matrix of the distribution network nodes. This represents the impedance between nodes i and j. and The reactive voltage sensitivity and active voltage sensitivity between nodes i and j are respectively represented. The superscripts ii, jj, ij, and ji in equations (8) and (9) represent different transmission directions.

[0096] Step 300: Determine the relative distance and local density of nodes based on electrical distance;

[0097] Step 400: Using the DPC algorithm (Density Peak Clustering), determine the cluster centers and number of partitions based on the relative distance and local density to obtain the initial distribution network partition network diagram.

[0098] In this embodiment, step 400 specifically includes:

[0099] A density peak decision map is plotted with local density as the horizontal axis and relative distance as the vertical axis. The number of clusters and their centers are determined through visualization. The remaining nodes of the cluster are clustered at a short preset allocation distance. Nodes that meet the distance condition are assigned to the corresponding cluster center, while those that do not meet the condition are not assigned. This yields the initial distribution network partition network diagram. Finally, based on the classification situation and in conjunction with equation (17), the initialization of the label classification matrix F is completed.

[0100] DPC is an unsupervised clustering algorithm whose advantage lies in its ability to automatically identify density peaks (i.e., cluster centers) in the data. DPC is based on two assumptions: in the samples to be clustered, points with low local density and small relative distances usually surround the cluster centers; points with high local density and large relative distances are often the cluster centers.

[0101] Assume the samples to be clustered are For any sample point in the sample Local density For large samples, the calculation using a truncated kernel function can be expressed as:

[0102] (11)

[0103] (12)

[0104] In the formula, For sample points and The distance between them To cut off the distance.

[0105] For small samples, to reduce statistical bias and improve clustering performance, the Gaussian kernel function is used, which can be expressed as:

[0106] (13)

[0107] For sample points relative distance , can be represented as:

[0108] (14)

[0109] In this embodiment, the impedance, active and reactive voltage sensitivity between any two nodes in 10kV AC, 10kV DC, 380V AC and 750V DC distribution networks are calculated respectively, and then the electrical distance is calculated and normalized.

[0110] Taking a 10kV AC distribution network as an example, according to the principle of the DPC algorithm, larger local density and relative distance often indicate a cluster center. Figure 2 In the data, three nodes have large local and relative distances, therefore, the number of partitions and cluster centers can both be determined as three. Based on the truncation distance, the remaining nodes are clustered to complete the partition initialization. The initialization results are shown in Table 3.

[0111]

[0112] Step 500: Using the LPA algorithm (Label Propagation Algorithm), determine the distribution network partitioning result based on the initialized distribution network partitioning network diagram.

[0113] The computation process of the LPA algorithm is as follows:

[0114] Step 1: Determine the initial labeled nodes, initial unlabeled nodes, and label information based on the initialized distribution network partition network diagram. Determine the edge weights by analyzing the distances between edges in the initialized distribution network partition network diagram. This allows us to determine the similarity between nodes and the weight of the edge between node i and node j. It can be represented as:

[0115] (15)

[0116] In the formula, Let represent the Euclidean distance between node i and node j. This represents the parameters that affect the weights, where D is the total dimension of the feature vector. Let d be the value of the d-th feature dimension of node i.

[0117] Step 2: Calculate the probability matrix of label information transmission between nodes based on edge weights. The propagation probability between node i and node j It can be represented as:

[0118] (16)

[0119] In the formula, This indicates the number of initially labeled nodes. This indicates the initial number of unlabeled nodes.

[0120] Step 3: Define the matrix that records the label classification information of all nodes in the distribution network as the label classification matrix. Tag classification matrix elements It can be represented as:

[0121] (17)

[0122] In the formula, Represents a node. Indicates the label category.

[0123] Step 4: The label information transmitted by all nodes is propagated according to the propagation probability, and the label influence matrix is ​​defined. To represent the impact of this round of propagation on the update of label information of each node, it can be expressed as:

[0124] (18)

[0125] In the formula, This indicates the updated tag information. Let T represent the label classification matrix from the previous round, and T represent the propagation probability matrix. After each propagation, the values ​​of the initial labeled nodes need to be restored.

[0126] Step 5: Continue repeating the transmission process until the label classification matrix is ​​obtained. By keeping these settings unchanged, we can obtain the final tag information and community division.

[0127] Step 6: Calculate the weights of edges between different nodes based on electrical distance.

[0128] Step 7: Calculate the partition size matrix M, load balance matrix P, and historical influence factor H, and then calculate and improve the transmission probability matrix T.

[0129] (1) Partition size matrix

[0130] This invention uses a partition size matrix to reflect the size of the community. The larger the community size corresponding to a tag, the smaller its corresponding partition size factor. Partition Size Factor It can be represented as:

[0131] (19)

[0132] In the formula, Represented as The sum of the degrees of all nodes in the class label. This represents the total number of branches in the distribution network.

[0133] Partition Scale Matrix The diagonal elements are the corresponding partition size factors, which can be represented as:

[0134] (20)

[0135] (2) Load balance matrix

[0136] When updating tags, considering the proportion of load in each zone to the total load can bring the loads in different zones closer together, resulting in a more balanced power supply. The load balancing factor can be expressed as:

[0137] (twenty one)

[0138] In the formula, This represents the load balancing factor of the k-th partition. Represented as nodes The load.

[0139] Assuming the system is divided into s partitions, the load balancing matrix... The diagonal element is the load balancing factor, which can be expressed as:

[0140] (twenty two)

[0141] (3) Historical Influence Factors

[0142] When updating labels, considering historical label information can reduce label oscillations and improve algorithm stability. A defined historical influence factor is used. It can be represented as:

[0143] (twenty three)

[0144] In the formula, Represented as node at time t-1 of Historical impact factor of class labels Represented as Time's up -1 time interval nodes of The changes in the historical influence factor of class labels can be represented as follows:

[0145] (twenty four)

[0146] In the formula, and All are label classification matrices The elements in express -1 time-time label influence matrix The elements in.

[0147] For a partition size matrix M, the improved label delivery strategy, nodes Tag information It can be represented as:

[0148] (25)

[0149] In the formula, Represented as nodes The set of adjacent nodes, Represented as nodes and The sum of edge weights of adjacent nodes, Represented as The partition size factor.

[0150] Based on consideration of the historical influence factors of nodes, the transmission probability matrix T is improved and can be expressed as:

[0151] (26)

[0152] Step 8: Transmit the label information. Taking into account the partition scale matrix M, load balance matrix P, historical influence factor H and label influence matrix Y, update the label classification matrix F, which can be expressed as Equation (27), and complete one round of transmission.

[0153] (27)

[0154] Step 9: After the current round of transmission ends, determine whether the label classification matrix F of this round is the same as that of the previous round. If the label classification matrix F is different, return to step 7 for looping. If the label classification matrix F of this round is the same, the distribution network partitioning is completed, and the distribution network partitioning result is output.

[0155] Taking a 10kV AC distribution network as an example, the DPC algorithm is used for initialization to complete the above-mentioned electrical distance and initialization partitioning results. Then, the LPA algorithm of this invention is used for final partitioning. Introducing evaluation indicators such as partition size matrix, load balance matrix, and historical influence factors can make the partitioning as consistent as possible with the power grid partitioning concept. To verify the beneficial effects of the LPA algorithm of this invention, the following two partitioning schemes are used:

[0156] Option 1: Use the traditional LPA algorithm to partition the initial partitions based on the calculation results of electrical distance.

[0157] Option 2: Use the LPA algorithm of this invention to partition the initialized partitions based on the calculation results of electrical distance.

[0158] The partitioning results for the two schemes are shown in Table 4:

[0159]

[0160] Table 4 shows that the partitioning results of the two schemes differ, with different partitioning arrangements for nodes 5, 12, and 13. In Scheme 1, all three nodes are located in partition 2, resulting in a larger partition 2 and a smaller partition 3. Scheme 2 places node 5 in partition 1 and nodes 12 and 13 in partition 3. This approach demonstrates the role of the partition size matrix, making the partitioning more consistent with the principle of balanced size.

[0161] From the load zoning results, although both zoning outcomes deviate significantly from the load balancing requirements, this is due to the large differences in node loads. Some nodes have loads close to 200kW, while others have zero load; in the presence of distributed generation, some nodes may even have negative loads. Therefore, the load imbalance is determined by the inherent characteristics of the 10kV AC distribution network. Scheme 2 demonstrates a better load zoning effect than Scheme 1, highlighting the role of the load balancing matrix.

[0162] In terms of modularity index, the two schemes have similar values ​​and are both within a good range of modularity values.

[0163] It should be noted that modularity is a metric for measuring the density of nodes within a community, and it is the most commonly used evaluation indicator for assessing the effectiveness of community partitioning. This invention will use the modularity index to evaluate the quality of power grid partitioning. The modularity value is typically between 0.3 and 0.7, with a higher value indicating a better power grid partitioning effect. The modularity Q can be expressed as:

[0164] (28)

[0165] In the formula, This represents the elements of the unweighted adjacency matrix. This represents the total number of branches in the network. and Representing nodes respectively and nodes The degree, Represents a node and nodes The partition affiliation of the node, if the node and nodes If they belong to the same partition, then ,otherwise .

[0166] The remaining distribution network is divided into zones, and the zoning results are as follows: Figure 3 As shown in Table 5, the information is as follows.

[0167] Table 5 Remaining Distribution Network Zone Information

[0168]

[0169] As shown in Table 5, the load proportions among the zones of each distribution network are similar, which verifies the effectiveness of the load balancing matrix. Regarding the modularity index, the modularity values ​​of all distribution network zones are similar and fall within a good range. In summary, the distribution network zoning method proposed in this invention has good zoning performance.

[0170] Based on the same inventive concept, another embodiment of the present invention provides a power distribution network partitioning device, which corresponds to the method of the foregoing embodiment, and the device includes:

[0171] The first determining unit is used to determine the output power of distributed generation in the distribution network;

[0172] The calculation unit is used to calculate the electrical distance between any two nodes in the distribution network based on the node's location in the distribution network and the impact of changes in the output power of distributed power sources on the node.

[0173] The second determining unit is used to determine the relative distance and local density of nodes based on electrical distance;

[0174] The initialization unit is used to determine the cluster centers and number of partitions based on the relative distance and local density using the DPC algorithm, thereby obtaining the initial distribution network partition network diagram.

[0175] The partitioning unit is used to determine the distribution network partitioning result based on the initial distribution network partitioning network diagram using the LPA algorithm.

[0176] In summary, the distribution network partitioning method and apparatus provided by this invention include: determining the output power of distributed generation sources in the distribution network; calculating the electrical distance between any two nodes in the distribution network based on the node's location in the distribution network and the impact of changes in the output power of the distributed generation sources on the node; determining the relative distance and local density of the nodes based on the electrical distance; using the DPC algorithm to determine the cluster centers and the number of partitions based on the relative distance and local density to obtain an initialized distribution network partitioning network diagram; and using the LPA algorithm to determine the distribution network partitioning result based on the initialized distribution network partitioning network diagram. This invention ensures the objectivity and scientific nature of distribution network partitioning initialization and solves problems such as partitioning difficulties and giant partitions, ensuring the rationality of distribution network partitioning.

[0177] Other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the embodiments disclosed herein. This invention is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein. It should be understood that the invention is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is limited only by the appended claims.

Claims

1. A method for dividing a power distribution network, characterized in that, include: Determine the output power of distributed generation sources in the distribution network; Calculate the electrical distance between any two nodes in the distribution network based on the node's location in the distribution network and the impact of changes in the output power of distributed power sources on the node. The relative distance and local density of nodes are determined based on the electrical distance; Density peak clustering algorithm is used to determine the cluster centers and number of partitions based on the relative distance and local density, thereby obtaining the initial distribution network partition network diagram; The label transfer algorithm is used to determine the distribution network partitioning result based on the initialized distribution network partitioning network diagram; The calculation of the electrical distance between any two nodes in the distribution network includes: Calculate the impedance between any two nodes; The Jacobian matrix is ​​constructed by power flow calculation, and the corrected equation of the distribution network is obtained by solving the Newton-Raphson iterative method, thereby obtaining different types of sensitivity matrices; The node voltage change is determined based on the different types of sensitivity matrices. To address the node voltage variations caused by distributed generation access to the distribution network, the electrical distance between any two nodes is determined by comprehensively considering the node's location in the distribution network and the impact of the output power variation of the accessed distributed generation node on the node. The expression for the electrical distance between any two nodes is: (10) in, (4) (8) (9) In the formula, Let Re be the electrical distance between nodes i and j, and let Re represent the real part and Im represent the imaginary part. This represents the element in the i-th row and j-th column of the impedance matrix of the distribution network nodes. This represents the impedance between nodes i and j. and The reactive voltage sensitivity and active voltage sensitivity between nodes i and j are respectively represented. The superscripts ii, jj, ij, and ji in equations (8) and (9) represent different transmission directions.

2. The power distribution network zoning method according to claim 1, characterized in that, The distributed power sources include wind turbines and photovoltaic generators. Determining the output power of the distributed power sources in the distribution network includes: Cluster analysis is performed on the historical output data of the distributed power source to obtain the distribution of wind speed and solar irradiance. Then, based on the capacity of the distributed power source, the output power of the wind turbine and photovoltaic generator is obtained.

3. The power distribution network zoning method according to claim 2, characterized in that, The output power of the wind turbine is: (1) In the formula, This refers to the cut-in wind speed of the wind turbine. The rated wind speed of the wind turbine unit. This refers to the cut-off wind speed of the wind turbine. denoted as the rated output power of the wind turbine, and v as the wind speed.

4. The power distribution network zoning method according to claim 2, characterized in that, The output power of the photovoltaic generator set is: (2) In the formula, This indicates the rated output power of the photovoltaic generator set. denoted as rated solar irradiance, and s as solar irradiance.

5. The power distribution network zoning method according to claim 1, characterized in that, The corrected equation for the distribution network is: (5) In the formula, , , and Here are the Jacobian matrices of the corresponding variables in the distribution network, representing the active power P as a function of the phase angle. The rate of change of active power P with voltage amplitude V, and the rate of change of reactive power Q with phase angle The rate of change of reactive power Q with voltage amplitude V; The different types of sensitivity matrices are represented as follows: (6) In the formula, the sensitivity matrix Represents the change in phase angle at the node Changes in active power injected by the node The influence of sensitivity matrix Represents the change in phase angle at the node Changes in reactive power injected by the node The influence of sensitivity matrix Represents the change in node voltage Changes in active power injected by the node The influence of sensitivity matrix Represents the change in node voltage Changes in reactive power injected by the node The impact; Node voltage change The relationship between the power change and the power change at all nodes in the distribution network is as follows: (7) in, , N represents the total number of nodes in the distribution network. This represents the change in active power at node N. This represents the change in reactive power at node N.

6. The power distribution network zoning method according to claim 1, characterized in that, The density peak clustering algorithm is used to determine the cluster centers and the number of partitions based on the relative distance and local density, thereby obtaining an initial distribution network partition network diagram, including: A density peak decision map is plotted with local density as the x-axis and relative distance as the y-axis to determine the number of clusters and cluster centers. The allocation distance is preset, and the remaining nodes are clustered: nodes that meet the distance conditions are assigned to the corresponding cluster centers, and those that do not meet the conditions are not assigned for the time being, and finally the initial distribution network partition network diagram is obtained.

7. The power distribution network zoning method according to claim 6, characterized in that, The step of using the label-transfer algorithm to determine the distribution network partitioning result based on the initialized distribution network partitioning network diagram includes: Step 1: Determine the initial marked nodes, initial unmarked nodes, and label information based on the initialized distribution network partition network diagram, and determine the weight of the edges by the distance of the edges in the initialized distribution network partition network diagram; Step 2: Based on the weights of the edges, calculate the probability matrix of label information transmission between nodes; Step 3: Define the matrix that records the label classification information of all nodes in the distribution network as the label classification matrix; Step 4: The label information transmitted by all nodes is propagated according to the propagation probability. A label influence matrix is ​​defined to represent the impact of this round of propagation on the label information update of each node. After the propagation is completed, the values ​​of the initial labeled nodes are restored. Step 5: Repeat the transfer process until the label classification matrix remains unchanged; Step 6: Calculate the weights of edges between different nodes based on electrical distance; Step 7: Calculate the partition size matrix, load balance matrix, and historical influence factors, and then calculate and improve the transmission probability matrix; Step 8: Transfer the tag information. Taking into account the partition size matrix, load balance matrix, historical influence factors and tag influence matrix, update the tag classification matrix to complete one round of transfer. Step 9: After the current round of transmission ends, determine whether the label classification matrix of this round is the same as that of the previous round. If the label classification matrix is ​​different, return to step 7; if the label classification matrix of this round is the same, output the distribution network zoning result.

8. A distribution network zoning device, characterized in that, For implementing the distribution network zoning method as described in any one of claims 1-7, the apparatus comprises: The first determining unit is used to determine the output power of distributed generation in the distribution network; The calculation unit is used to calculate the electrical distance between any two nodes in the distribution network based on the node's location in the distribution network and the impact of changes in the output power of distributed power sources on the node. The second determining unit is used to determine the relative distance and local density of nodes based on the electrical distance; An initialization unit is used to determine the number of cluster centers and partitions based on the relative distance and local density using a density peak clustering algorithm, thereby obtaining an initial distribution network partition network diagram. The partitioning unit is used to determine the distribution network partitioning result based on the initialized distribution network partitioning network diagram using a label transfer algorithm.

Images