Power grid dynamic double-layer partition optimization method considering regional voltage control capability

Abstract

The invention discloses a power grid dynamic double-layer partition optimization method considering regional voltage control capability, and relates to the field of power grid reactive power control. The method comprises the following steps: (1) data preparation; and determining the current power grid operation condition based on the historical operation power data of the power grid or the power data measured in real time. (2) calculating an expansion type reactive voltage sensitivity matrix; and according to the power grid power data, calculating the expansion type reactive voltage sensitivity among the nodes of the power grid. And (3) first-layer partitioning based on modularity. The connection weight between nodes is calculated through the expanded reactive voltage sensitivity matrix, a regional modularity function is constructed, and clustering is carried out by adopting an improved K-means algorithm. And (4) sub-layer partitioning based on the regional reactive power margin. And constructing a regional reactive power margin index according to the reactive power compensation capability of the reactive power equipment, and carrying out clustering by adopting an improved K-means algorithm. And repeating for a plurality of times to obtain a plurality of partitioning results. And (5) carrying out double-layer partition combination. And constructing a partition similarity index, taking the partition result with the highest similarity with the result in the step (3) in the step (4), and integrating the two partition results by adopting condensation clustering. And (6) performing partition optimization. And judging the region where the voltage out-of-limit node is located, and scheduling reactive power sources in the region for optimization. The method has the advantages that the reactive power margin and the reactive power voltage sensitivity are integrated into the field of power grid partitioning, the reactive power margin of each region after power grid partitioning is reasonably distributed, the regional reactive power voltage control capability is improved, and the power grid partitioning optimization efficiency is improved.

Description

Technical Field

[0001] The present invention relates to a power grid dynamic double-layer partition optimization method taking into account regional voltage control capability, belonging to the field of power grid reactive power control. Background Art

[0002] Reactive power optimization of power grid is an important means to improve the safety, stability and economy of power system. Reasonable configuration of reactive power can reduce active power loss of power grid, reduce reactive power dispatching cost, improve system stability and improve voltage quality. With the large-scale access of renewable energy and distributed energy and the growing demand for electricity, the management of voltage and reactive power has become particularly important. Grid zoning optimization is the basis for refined dispatching of reactive equipment, which is crucial to ensure the stable operation of power grid and improve power quality.

[0003] Under the existing power grid framework, the present invention partitions the power grid from two aspects: reactive voltage sensitivity between nodes and reactive margin of partitions, and integrates the two partition results. On this basis, the power grid is optimized to achieve refined scheduling of reactive equipment, which is of great significance for ensuring the stable operation of the power grid and improving the power quality. Summary of the invention

[0004] The technical problem to be solved by the present invention is to overcome the existing technical defects and propose a dynamic double-layer partition optimization method for power grids taking into account the regional voltage control capability. The method comprises the following steps: (1) Data preparation. Obtain the reactive power compensation capability data of the reactive power equipment of the power grid, and monitor whether the voltage of the power grid node exceeds the limit. If so, confirm the current power grid operation power situation based on the historical power data of the power grid or the real-time measured power data, and jump to step (2). Otherwise, continue to monitor the node voltage.

[0005] (2) Calculate the extended reactive voltage sensitivity matrix. According to the power data of the power grid, calculate the reactive voltage sensitivity between each PQ node of the power grid, then treat each PV node as a PQ node and calculate the corresponding reactive voltage sensitivity, and then obtain the extended reactive voltage sensitivity matrix including all nodes except the balancing node.

[0006] (3) First-level partitioning based on modularity: The node connection weights are calculated by using the extended reactive voltage sensitivity matrix, the regional modularity function is constructed, and the improved Kmeans algorithm is used for clustering.

[0007] (4) Sub-level partitioning based on regional reactive power margin. According to the reactive power compensation capacity of reactive equipment, the regional reactive power margin index is constructed, and combined with the grid topology, the improved Kmeans algorithm is used for clustering. Repeat several times to obtain multiple partitioning results.

[0008] (5) Combination of two-layer partitions. Construct a partition similarity index, calculate the similarity of the partition results in step (3) and step (4), take the partition result in step (4) with the highest similarity to the result in step (3), construct a similarity matrix for each node partition, and use agglomerative clustering to integrate the two partition results. The advantage of the present invention is that it integrates reactive margin and reactive voltage sensitivity into the field of power grid partitioning, reasonably allocates reactive margins in each area after power grid partitioning, and improves the reactive voltage control capability of the area.

[0009] (6) Zoning optimization: Determine the area where the voltage-limited node is located and optimize the reactive power resources in the dispatching area.

[0010] Furthermore, the current power condition of the power grid in step (1) is confirmed. Since the partitioning of the power grid in the present invention is based on the current operating state of the system, the partitioning results obtained under different operating states will be different, so it is necessary to obtain the current operating power parameters of the power grid. The power parameter determination method is as follows: If historical operating data is used, the daily power data is divided into one hour period, and the historical power data of the past 30 days in the current period is taken and processed according to the following formulas (1) and (2).

[0011] (1) (2) in: and To process the obtained power data, we regard it as Grid power during the period; and They are the power grid's performance in the past week and the past month. The average value of active power in the time period; and The past week and the past month of the power grid The average value of reactive power in the time period; and is the weight coefficient, , .

[0012] The average value is taken as the power grid power of the current period; if the real-time measured power data is used, the power data in the past 30 minutes is averaged as the power grid power data of the electrical appliance.

[0013] Furthermore, the calculation of the extended reactive voltage sensitivity matrix in step (2) is as follows: The sensitivity matrix can directly reflect the coupling relationship between nodes and is an important basis for the grid partitioning of the present invention. The calculation method of the extended reactive voltage sensitivity matrix is ​​as follows: First, the reactive voltage sensitivity matrix of the PQ node is calculated. The Newton method is used to calculate the power flow distribution of the power grid, and the power correction equation in the last iteration process is taken, and its expression is shown in formula (3): (3) in: is the Jacobian matrix, which can be expressed to illustrate the partial derivative relationship between power and node voltage; and They represent the deviation matrices of active power and reactive power injected into the node respectively; and Respectively represent the change matrices of node voltage phase angle and voltage amplitude.

[0014] Inverting the above formula gives the sensitivity matrix: (4) in: is the sensitivity matrix; and They represent active voltage and reactive voltage sensitivity factors, respectively, and are used to explain the changes in node voltage caused by changes in system node power; and It reflects the voltage phase angle change caused by the power change of the system node.

[0015] Grid node voltage change Changes in active and reactive power and The relationship can be expressed as: (5) in: , Ignore the active The impact of voltage change on the reactive power voltage sensitivity matrix of the PQ node is obtained : (6) Calculate the reactive voltage sensitivity matrix of the PV node. The PV node has the ability to maintain its own voltage, so the generalized reactive voltage sensitivity is defined to reflect the coupling characteristics between the PV node and the other nodes. The generalized reactive voltage sensitivity is defined as follows: for a certain PV node, replace the node with a PQ node with the same electrical characteristics, and the voltage change caused by the reactive power change of the other nodes in the power grid on the node. The calculation process is as follows: The PV node for which reactive voltage sensitivity calculation is required is regarded as a PQ node, and the power flow of the entire network is calculated. The reactive voltage sensitivity of the node can be obtained by using the above method of calculating the reactive voltage sensitivity of the PQ node. The remaining PV nodes are regarded as PQ nodes one by one to calculate the corresponding reactive voltage sensitivity.

[0016] The reactive voltage sensitivity of the PQ node is integrated with the reactive voltage sensitivity of the PV node to obtain the extended reactive voltage sensitivity matrix: (7) in: and are the number of PQ nodes and PV nodes respectively; the matrix Medium Element It means the reactive voltage sensitivity of PQ node i to PQ node j; matrix Medium Element It means the reactive voltage sensitivity of PQ node i to PV node j; matrix Medium Element It means the reactive voltage sensitivity of PV node i to PQ node j; matrix Medium Element It means the reactive voltage sensitivity of PV node i to PV node j.

[0017] Furthermore, the construction of the modularity function in step (3) is as follows: (8) (9) (10) (11) in: and is the first in the extended reactive voltage sensitivity matrix OK Columns and OK Two elements of the column; To connect nodes and nodes The right to connect between Represents the sum of the weights of all edges in the network; if the node With Node In the same partition, ,otherwise .

[0018] Furthermore, the step (3) adopts an improved K-means modularity-based power grid partitioning method. The calculation of the modularity index requires knowing the partitioning status of all nodes. The traditional K-means algorithm is not applicable and its iterative process needs to be improved. The steps of improving the K-means algorithm for modularity-based power grid partitioning are as follows: ① Select the initial partition point. Determine the number of partition targets under the guidance of experts . Randomly select Nodes, calculate the sum of the connection weights between nodes as the basis for judgment. Perform several random selections and take a group of random nodes with the lowest sum of connection weights as the initial partition points.

[0019] ② Node partition iteration. Calculate the corresponding modularity when the remaining nodes are merged into different partitions, and record the partition with the largest modularity.

[0020] ③ Update the partition results. According to the partition results with the largest modularity recorded in step ②, update the node partition situation. Less than the maximum number of iterations If the result is half, jump to step ②, otherwise jump to step ④.

[0021] ④Randomly select Nodes are merged into different partitions, and their corresponding modularity is calculated. The partition with the largest modularity is recorded. Repeat step ④ until the number of iterations reaches the maximum value. The value of is shown in formula (12): (12) in: is the ceiling rounding function; and are the current iteration number and the maximum iteration number respectively.

[0022] ⑤ Output the partitioning results. The balanced node is assigned to the node partition with the largest connection weight.

[0023] Furthermore, the regional reactive power margin index of step (4) is The conventional reactive margin definition method of the ratio of reactive load to maximum reactive compensation has limitations, ignoring that different partitions contain different numbers of nodes and corresponding reactive margin weights are also different. The present invention proposes a regional reactive margin index , defined as follows: (13) in: For Region The corresponding regional reactive power margin; and Respectively, the regions Reactive load and maximum reactive compensation amount; For Region The number of nodes you have; is the penalty coefficient, which is a large positive number.

[0024] Furthermore, the step (4) adopts the improved K-means method for regional reactive power margin-based power grid partitioning. The reactive power margin data distribution of nodes is different from the reactive voltage sensitivity. In order to prevent the problem of regional disconnection in the partition results, it is necessary to improve the partition selection part of the algorithm iteration process. The steps of regional reactive power margin-based power grid partitioning using improved K-means are as follows: ①Construct a connection matrix. According to the power grid topology diagram, construct a connection matrix .

[0025] (14) in: The connection matrix Middle Line Column elements; Reaction Node With Node Is it directly connected? If so, ,otherwise .

[0026] ② Select the initial partition point. Use the initial partition in "Partitioning based on modularity". If the number of nodes with reactive power sources is Greater than the number of partitions , jump to step ③, otherwise terminate the partitioning operation.

[0027] ③ Node partition iteration. According to the connection matrix , the corresponding regional reactive power margin index when each node is merged into the adjacent partition , record the partition with the largest modularity.

[0028] ④Update the partition results. According to the maximum regional reactive power margin index recorded in step ③ , update the node partition situation. If the number of partition iterations Less than the maximum number of iterations If the result is half, jump to step ③, otherwise jump to step ⑤.

[0029] ⑤Randomly select Nodes, calculate the corresponding modularity when they are incorporated into different partitions, and record the regional reactive power margin index The maximum partition situation. Repeat step ⑤ until the number of iterations reaches the maximum value. The value of is shown in formula (15): (15) in: is the ceiling rounding function; and are the current iteration number and the maximum iteration number respectively.

[0030] ⑥ Repeat steps ② to ⑤ several times to obtain multiple partition results.

[0031] Furthermore, the integration of the two-layer partitioning results in step (5) is performed by defining the partitioning result similarity based on the partitions of the nodes in different partitioning results, and defining the similarity between nodes based on the partitioning results of multiple nodes in the same partition. The steps for integrating the two-layer partitioning results are as follows: ① Select the integration base. Define the similarity index of the partition results, and select the partition result with the highest similarity to the first-level partition "grid partition based on modularity" from the second-level partition "grid partition based on regional reactive power margin" as the integration base. The definition is as follows: (16) in: Represents the partition result Midpoint The set of nodes to which the partition belongs; Represents the partition result Midpoint The set of nodes to which the partition belongs; Represents a partition Midpoint Partition and zone Midpoint The number of intersections of the partitions to which it belongs; Represents a partition Midpoint Partition and zone Midpoint The number of unions of the partitions to which it belongs.

[0032] ②Construct the similarity matrix between nodes. With Node In different partition results, whether they belong to the same region, define the similarity matrix between nodes : (17) (18) (19) in: and Representation Node With Node In the partition The set of partitions to which it belongs; Reaction Node With Node Whether they belong to the same area in the partition; is the number of integration bases adopted; Reaction Node With Node The similarity of the partitions in the integrated base ranges from [0 to 1]. The node similarity here is defined by whether different nodes belong to the same partition in multiple partition results, which is different from the similarity in step ① above.

[0033] ③ Agglomerative clustering. Using similarity matrix The elements of each row are used as the attributes of the corresponding node, the number of partition targets As the termination condition, agglomerative clustering is used to cluster the similarity matrix between nodes.

[0034] Furthermore, the partition optimization in step (6) is as follows: ① Data reading. Obtain the grid operation parameters and detect whether any node voltage exceeds the limit. If yes, jump to step ②.

[0035] ② Determine the partition to which the voltage-over-limit node belongs as the main partition, and dispatch the reactive equipment in the main partition for optimization simulation. If the voltage-over-limit problem is successfully solved, end the optimization and output the optimization plan. Otherwise, jump to step ③.

[0036] ③Traverse the node voltages in the adjacent partitions of the main partition, select the partition with the highest voltage node and merge it into the main partition, dispatch the reactive sources in the main partition to perform reactive optimization simulation, and if the voltage over-limit problem is successfully solved, end the optimization and output the optimization plan; otherwise, repeat step ③. If all reactive sources have participated in the dispatch and there are still voltage over-limit nodes, output the current control plan and issue a warning. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 Overall flow chart Figure 2 Improved K-means partitioning process based on modularity Figure 3 Improved K-means partitioning process based on regional reactive power margin index Figure 4 Partition result integration process Figure 5 Partition Optimization Process DETAILED DESCRIPTION

[0038] The present invention is further described in conjunction with the given embodiments, but is not limited thereto. A method for optimizing a power grid with dynamic two-layer partitioning taking into account the regional voltage control capability, the process is as follows: Figure 1 As shown, the following steps are included: (1) Data preparation. Obtain the reactive power compensation capability data of the reactive power equipment of the power grid, and monitor whether the voltage of the power grid node exceeds the limit. If so, confirm the current power grid operation power situation based on the historical power data of the power grid or the real-time measured power data, and jump to step (2). Otherwise, continue to monitor the node voltage.

[0039] If historical operating data is used, the daily power data is divided into one hour period, and the historical power data of the past 30 days in the current period is taken and processed according to the following formulas (1) and (2).

[0040] (1) (2) in: and To process the obtained power data, we regard it as Grid power during the period; and They are the power grid's performance in the past week and the past month. The average value of active power in the time period; and The past week and the past month of the power grid The average value of reactive power in the time period; and is the weight coefficient, , .

[0041] The average value is taken as the power grid power of the current period; if the real-time measured power data is used, the power data in the past 30 minutes is averaged as the power grid power data of the electrical appliance.

[0042] (2) Calculate the extended reactive voltage sensitivity matrix. According to the power data of the power grid, calculate the reactive voltage sensitivity between each PQ node of the power grid.

[0043] First, the reactive voltage sensitivity matrix of the PQ node is calculated. The Newton method is used to calculate the power flow distribution of the power grid, and the power correction equation in the last iteration process is taken, and its expression is shown in formula (3): (3) in: is the Jacobian matrix, which can be expressed to illustrate the partial derivative relationship between power and node voltage; and They represent the deviation matrices of active power and reactive power injected into the node respectively; and Respectively represent the change matrices of node voltage phase angle and voltage amplitude.

[0044] Inverting the above formula gives the sensitivity matrix: (4) in: is the sensitivity matrix; and They represent active voltage and reactive voltage sensitivity factors, respectively, and are used to explain the changes in node voltage caused by changes in system node power; and It reflects the voltage phase angle change caused by the power change of the system node.

[0045] Grid node voltage change Changes in active and reactive power and The relationship can be expressed as: (5) in: , Ignore the active The impact of voltage change on the reactive power voltage sensitivity matrix of the PQ node is obtained : (6) After that, the PV nodes are regarded as PQ nodes one by one to calculate the corresponding reactive voltage sensitivity, and then the extended reactive voltage sensitivity matrix including all nodes except the balancing node is obtained. Each PV node is still regarded as a PV node in the sensitivity calculation process of other PV nodes.

[0046] The reactive voltage sensitivity of the PQ node is integrated with the reactive voltage sensitivity of the PV node to obtain the extended reactive voltage sensitivity matrix: (7) in: and are the number of PQ nodes and PV nodes respectively; the matrix Medium Element It means the reactive voltage sensitivity of PQ node i to PQ node j; matrix Medium Element It means the reactive voltage sensitivity of PQ node i to PV node j; matrix Medium Element It means the reactive voltage sensitivity of PV node i to PQ node j; matrix Medium Element It means the reactive voltage sensitivity of PV node i to PV node j.

[0047] (3) First-level partitioning based on modularity. Calculate the connection weights between nodes through the extended reactive voltage sensitivity matrix : (8) in: and is the first in the extended reactive voltage sensitivity matrix OK Columns and OK Two elements of the column; To connect nodes and nodes The right to connect.

[0048] Constructing regional modularity function , defined as follows: (9) (10) (11) in: Represents the sum of the weights of all edges in the network; if the node With Node In the same partition, ,otherwise .

[0049] The improved K-means algorithm is used for clustering. The process is as follows Figure 2 As shown, the steps are as follows: ① Select the initial partition point. Determine the number of partition targets under the guidance of experts . Randomly select Nodes, calculate the sum of the connection weights between nodes as the basis for judgment. Perform several random selections and take a group of random nodes with the lowest sum of connection weights as the initial partition points.

[0050] ② Node partition iteration. Calculate the corresponding modularity when the remaining nodes are merged into different partitions, and record the partition with the largest modularity.

[0051] ③ Update the partition results. According to the partition results with the largest modularity recorded in step ②, update the node partition situation. Less than the maximum number of iterations If the result is half, jump to step ②, otherwise jump to step ④.

[0052] ④Randomly select Nodes are merged into different partitions, and their corresponding modularity is calculated. The partition with the largest modularity is recorded. Repeat step ④ until the number of iterations reaches the maximum value. The value of is shown in formula (12): (12) in: is the ceiling rounding function; and are the current iteration number and the maximum iteration number respectively.

[0053] ⑤ Output the partitioning results. The balanced node is assigned to the node partition with the largest connection weight.

[0054] (4) Sub-division based on regional reactive power margin. Construct regional reactive power margin index according to the reactive power compensation capacity of reactive equipment. : (13) in: For Region The corresponding regional reactive power margin; and Respectively, the regions Reactive load and maximum reactive compensation amount; For Region The number of nodes you have; is the penalty coefficient, which is a large positive number.

[0055] Combined with the power grid topology, the improved K-means algorithm is used for clustering. The process is as follows: Figure 3 As shown, the steps are as follows: ①Construct a connection matrix. According to the power grid topology diagram, construct a connection matrix .

[0056] (14) in: The connection matrix Middle Line Column elements; Reaction Node With Node Is it directly connected? If so, ,otherwise .

[0057] ② Select the initial partition point. Use the initial partition in "Partitioning based on modularity". If the number of nodes with reactive power sources is Greater than the number of partitions , jump to step ③, otherwise terminate the partitioning operation.

[0058] ③ Node partition iteration. According to the connection matrix , the corresponding regional reactive power margin index when each node is merged into the adjacent partition , record the partition with the largest modularity.

[0059] ④Update the partition results. According to the maximum regional reactive power margin index recorded in step ③ , update the node partition situation. If the number of partition iterations Less than the maximum number of iterations If the result is half, jump to step ③, otherwise jump to step ⑤.

[0060] ⑤Randomly select Nodes, calculate the corresponding modularity when they are incorporated into different partitions, and record the regional reactive power margin index The maximum partition situation. Repeat step ⑤ until the number of iterations reaches the maximum value. The value of is shown in formula (15): (15) in: is the ceiling rounding function; and are the current iteration number and the maximum iteration number respectively.

[0061] ⑥ Repeat steps ② to ⑤ several times to obtain multiple partition results.

[0062] (5) Double-layer partition combination. The process is as follows Figure 4 As shown, the partition result similarity index is defined based on the partition of the node in different partition results. : (16) in: Represents the partition result Midpoint The set of nodes to which the partition belongs; Represents the partition result Midpoint The set of nodes to which the partition belongs; Represents a partition Midpoint Partition and zone Midpoint The number of intersections of the partitions to which it belongs; Represents a partition Midpoint Partition and zone Midpoint The number of unions of the partitions to which it belongs.

[0063] Calculate the similarity between the partition results in step (3) and step (4), and take the partition result in step (4) that has the highest similarity with the result in step (3).

[0064] Construct the similarity matrix of each node partition to represent the similarity of different nodes in the same partition and define the similarity matrix between nodes The definition is as follows: (17) (18) (19) in: and Representation Node With Node In the partition The set of partitions to which it belongs; Reaction Node With Node Whether they belong to the same area in the partition; is the number of integration bases adopted; Reaction Node With Node Similarity of partitions in the integration base, in the range [0, 1].

[0065] Similarity Matrix The elements of each row are used as the attributes of the corresponding node, the number of partition targets As the termination condition, agglomerative clustering is used to cluster the similarity matrix between nodes.

[0066] (6) Partition optimization. The process is as follows Figure 5 As shown in the figure, the area where the voltage limit node is located is determined, and the reactive power source in the dispatching area is optimized. The steps are as follows: ① Data reading. Obtain the grid operation parameters and detect whether any node voltage exceeds the limit. If yes, jump to step ②.

[0067] ② Determine the partition to which the voltage-over-limit node belongs as the main partition, and dispatch the reactive equipment in the main partition for optimization simulation. If the voltage-over-limit problem is successfully solved, end the optimization and output the optimization plan. Otherwise, jump to step ③.

[0068] ③Traverse the node voltages in the adjacent partitions of the main partition, select the partition with the highest voltage node and merge it into the main partition, dispatch the reactive sources in the main partition to perform reactive optimization simulation, and if the voltage over-limit problem is successfully solved, end the optimization and output the optimization plan; otherwise, repeat step ③. If all reactive sources have participated in the dispatch and there are still voltage over-limit nodes, output the current control plan and issue a warning.

Claims

1. A method for optimizing the dynamic double-layer partitioning of a power grid taking into account the regional voltage control capability, characterized in that: The following steps are involved: (1) Data preparation: Obtain reactive power compensation capability data of reactive power equipment in the power grid, and monitor whether the voltage of power grid nodes exceeds the limit. If so, confirm the current power grid operation power situation based on the historical power data of the power grid or the real-time measured power data, and jump to step (2); otherwise, continue to monitor the node voltage; (2) Calculate the extended reactive voltage sensitivity matrix: According to the power data of the power grid, calculate the reactive voltage sensitivity between each PQ node of the power grid, then treat each PV node as a PQ node and calculate the corresponding reactive voltage sensitivity, and then obtain the extended reactive voltage sensitivity matrix including all nodes except the balancing node; (3) First-level partitioning based on modularity: The node connection weights are calculated through the extended reactive voltage sensitivity matrix, the regional modularity function is constructed, and the improved Kmeans algorithm is used for clustering; (4) Sub-level partitioning based on regional reactive power margin: According to the reactive power compensation capacity of reactive equipment, construct regional reactive power margin index, combine with the grid topology, and use improved Kmeans algorithm for clustering. Repeat several times to obtain multiple partitioning results; (5) Combination of two-layer partitions: construct a partition similarity index, calculate the similarity of the partition results in step (3) and step (4), take the partition result in step (4) with the highest similarity to the result in step (3), construct a similarity matrix for each node partition, and use agglomerative clustering to integrate the two partition results. The advantage of the present invention is that it integrates reactive margin and reactive voltage sensitivity into the field of power grid partitioning, reasonably allocates reactive margins in each area after power grid partitioning, and improves the reactive voltage control capability of the area; (6) Zoning optimization: Determine the area where the voltage exceeds the limit node and optimize the reactive power resources in the dispatching area.

2. A method for optimizing the dynamic double-layer partitioning of a power grid taking into account the regional voltage control capability according to claim 1, characterized in that: Confirmation of the current power grid operating power status in step (1): If historical operation data is used, the daily power data is divided into one hour period, and the historical power data of the past 30 days in the current period is taken and processed according to the following formulas (1) and (2): (1) (2) in: and To process the obtained power data, we regard it as Grid power during the period; and They are the power grid's performance in the past week and the past month. The average value of active power in the time period; and The past week and the past month of the power grid The average value of reactive power in the time period; and is the weight coefficient, , , take the average value as the power grid power in the current period; If real-time measured power data is used, the power data within the past 30 minutes is averaged and used as the electrical appliance power grid power data.

3. A method for optimizing the dynamic double-layer partitioning of a power grid taking into account the regional voltage control capability according to claim 1, characterized in that: Calculation of the extended reactive voltage sensitivity matrix in step (2): First, the reactive voltage sensitivity matrix of the PQ node is calculated, and the power flow distribution of the power grid is calculated using the Newton method. The power correction equation in the last iteration is taken, and its expression is shown in formula (3): (3) in: is the Jacobian matrix, which can be expressed to illustrate the partial derivative relationship between power and node voltage; and They represent the deviation matrices of active power and reactive power injected into the node respectively; and Respectively represent the change matrices of node voltage phase angle and voltage amplitude; Inverting the above formula gives the sensitivity matrix: (4) in: is the sensitivity matrix; and They represent active voltage and reactive voltage sensitivity factors, respectively, and are used to explain the changes in node voltage caused by changes in system node power; and It reflects the voltage phase angle change caused by the power change of the system node; Grid node voltage change Changes in active and reactive power and The relationship can be expressed as: (5) in: , Ignore the active The impact of voltage change on the reactive power voltage sensitivity matrix of the PQ node is obtained : (6) Calculate the reactive voltage sensitivity matrix of the PV node. The PV node has the ability to maintain its own voltage, so the generalized reactive voltage sensitivity is defined to reflect the coupling characteristics between the PV node and other nodes; The generalized reactive voltage sensitivity is defined as follows: for a certain PV node, replace the node with a PQ node with the same electrical characteristics, and the voltage change caused by the reactive power change of the remaining nodes in the power grid at the node; The calculation process is as follows: The PV node for which reactive voltage sensitivity calculation is required is regarded as a PQ node, and the power flow of the entire network is calculated. The reactive voltage sensitivity of the node can be obtained by using the above method of calculating the reactive voltage sensitivity of the PQ node. The remaining PV nodes are regarded as PQ nodes one by one to calculate the corresponding reactive voltage sensitivity. The reactive voltage sensitivity of the PQ node is integrated with the reactive voltage sensitivity of the PV node to obtain the extended reactive voltage sensitivity matrix: (7) in: and are the number of PQ nodes and PV nodes respectively; the matrix Medium Element It means the reactive voltage sensitivity of PQ node i to PQ node j; matrix Medium Element It means the reactive voltage sensitivity of PQ node i to PV node j; matrix Medium Element It means the reactive voltage sensitivity of PV node i to PQ node j; matrix Medium Element It means the reactive voltage sensitivity of PV node i to PV node j.

4. A method for optimizing the dynamic double-layer partitioning of a power grid taking into account the regional voltage control capability according to claim 1, characterized in that: The construction of the modularity function in step (3) is: The modularity function is applied in weighted networks to measure the community structure characteristics of complex networks and determine the optimal number of partitions. The modularity function is defined as follows: (8) (9) (10) (11) in: and is the first in the extended reactive voltage sensitivity matrix OK Columns and OK Two elements of the column; To connect nodes and nodes The right to connect between Represents the sum of the weights of all edges in the network; if the node With Node In the same partition, ,otherwise .

5. A method for optimizing the dynamic double-layer partitioning of a power grid taking into account the regional voltage control capability according to claim 1, characterized in that: The step (3) adopts an improved K-means modularity-based power grid partitioning method: The steps of improving K-means algorithm for modularity-based power grid partitioning are as follows: ① Select the initial partition point: determine the number of partition targets under the guidance of experts , randomly selected nodes, calculate the sum of the connection weights between nodes as the basis for judgment, make several random selections, and take a group of random nodes with the lowest sum of connection weights as the initial partition points; ② Node partition iteration: Calculate the corresponding modularity when the remaining nodes are merged into different partitions, and record the partition with the largest modularity; ③ Update partition results: Update node partitions according to the partition results with the largest modularity recorded in step ②. Less than the maximum number of iterations If the value is half, jump to step ②, otherwise jump to step ④; ④Randomly select Nodes are merged into different partitions, and their corresponding modularity is calculated. The partition with the largest modularity is recorded. Step ④ is repeated until the number of iterations reaches the maximum value. The value of is shown in formula (12): (12) in: is the ceiling rounding function; and are the current iteration number and the maximum iteration number respectively; ⑤Output partition results: The balanced node is assigned to the node partition with the largest connection weight.

6. A method for optimizing the dynamic double-layer partitioning of a power grid taking into account the regional voltage control capability according to claim 1, characterized in that: The step (4) regional reactive power margin index Definition: (13) in: For Region The corresponding regional reactive power margin; and Respectively, the regions Reactive load and maximum reactive compensation amount; For Region The number of nodes you have; is the penalty coefficient, which is a large positive number.

7. A method for optimizing the dynamic double-layer partitioning of a power grid taking into account the regional voltage control capability according to claim 1, characterized in that: The step (4) adopts an improved K-means method for power grid partitioning based on regional reactive power margin: The steps of improving K-means for power grid partitioning based on regional reactive power margin are as follows: ①Construct connection matrix: Construct connection matrix according to the power grid topology diagram : (14) in: The connection matrix Middle Line Column elements; Reaction Node With Node Is it directly connected? If so, ,otherwise ; ② Select the initial partition point: Use the initial partition in "Partitioning based on modularity". If the number of nodes with reactive power sources is Greater than the number of partitions , jump to step ③, otherwise terminate the partitioning operation; ③Node partition iteration: according to the connection matrix , the corresponding regional reactive power margin index when each node is merged into the adjacent partition , record the partition with the largest modularity; ④Update the partition results: according to the maximum regional reactive power margin index recorded in step ③ , update the node partition situation. If the number of partition iterations Less than the maximum number of iterations If the value is half, jump to step ③, otherwise jump to step ⑤; ⑤Randomly select Nodes, calculate the corresponding modularity when they are incorporated into different partitions, and record the regional reactive power margin index The largest partition situation. Repeat step ⑤ until the number of iterations reaches the maximum. The value of is shown in formula (15): (15) in: is the ceiling rounding function; and are the current iteration number and the maximum iteration number respectively; ⑥ Repeat steps ② to ⑤ several times to obtain multiple partition results.

8. A method for optimizing the dynamic double-layer partitioning of a power grid taking into account the regional voltage control capability according to claim 1, characterized in that: Integration of the two-layer partitioning results in step (5): The steps for integrating the two-layer partition results are as follows: ① Select the integration base: define the similarity index of the partition results, and select the partition result with the highest similarity to the first-level partition "grid partition based on modularity" from the second-level partition "grid partition based on regional reactive power margin" as the integration base. Partition similarity index The definition is as follows: (16) in: Represents the partition result Midpoint The set of nodes to which the partition belongs; Represents the partition result Midpoint The set of nodes to which the partition belongs; Represents a partition Midpoint Partition and zone Midpoint The number of intersections of the partitions to which it belongs; Represents a partition Midpoint Partition and zone Midpoint The number of unions of the partitions to which it belongs; ②Construct the node similarity matrix: for nodes With Node In different partition results, whether they belong to the same region, define the similarity matrix between nodes : (17) (18) (19) in: and Representation Node With Node In the partition The set of partitions to which it belongs; Reaction Node With Node Whether they belong to the same area in the partition; is the number of integration bases adopted; Reaction Node With Node The similarity of the partitions in the integrated base ranges from [0 to 1]. The node similarity here is defined by whether different nodes belong to the same partition in multiple partition results, which is different from the similarity in step ① above; ③ Agglomerative clustering: similarity matrix The elements of each row are used as the attributes of the corresponding node, the number of partition targets As the termination condition, agglomerative clustering is used to cluster the similarity matrix between nodes.

9. A method for optimizing the dynamic double-layer partitioning of a power grid taking into account the regional voltage control capability according to claim 1, characterized in that: The partition optimization in step (6) is as follows: ① Data reading: obtain the grid operation parameters and detect whether any node voltage exceeds the limit. If yes, jump to step ②; ② Determine the partition to which the voltage-over-limit node belongs as the main partition, and dispatch the reactive equipment in the main partition for optimization simulation. If the voltage-over-limit problem is successfully solved, the optimization ends and the optimization plan is output. Otherwise, jump to step ③; ③ Traverse the node voltages in the adjacent partitions of the main partition, select the partition with the highest voltage node and merge it into the main partition, dispatch the reactive sources in the main partition to perform reactive optimization simulation. If the voltage over-limit problem is successfully solved, end the optimization and output the optimization plan. Otherwise, repeat step ③. If all reactive sources have participated in the dispatch and there are still voltage over-limit nodes, output the current control plan and issue a warning.

Images