A power distribution network distributed power source cluster dynamic division method based on multiple control targets

Abstract

This invention relates to the field of distribution network operation and control technology, and particularly to a dynamic partitioning method for distributed power generation clusters in distribution networks based on multiple control objectives. The method includes: collecting operational data from candidate nodes in the distribution network and constructing node state variables; calculating node cohesion based on the operational data and node state variables; calculating cluster efficiency based on node cohesion and node state variables, and determining the optimal cluster partitioning through recursive merging and efficiency verification; obtaining the total active power regulation demand at the cluster level based on the final cluster partitioning result; and converting the total active power regulation demand at the cluster level into the final active power regulation command for each node, based on the cluster efficiency value. This method solves the problems of traditional methods, such as partitioning results failing when load fluctuates or renewable energy output changes, partitioning results being difficult to directly use for control and allocation, and the potential for local overload or voltage exceeding limits. It also fails to meet the requirements for flexibility and stability of the distribution network under conditions of high-proportion distributed power generation access.

Description

Technical Field

[0001] This invention relates to the field of power distribution network operation and control technology, and in particular to a dynamic partitioning method for distributed power generation clusters in power distribution networks based on multiple control objectives. Background Technology

[0002] With the large-scale integration of distributed generation into distribution networks, traditional distribution networks characterized by unidirectional power supply are gradually evolving into active distribution networks with multi-source interaction between power sources, grids, and loads. The widespread deployment of distributed photovoltaic, decentralized wind power, and energy storage devices in distribution networks has led to bidirectional and more volatile power flows at nodes. Therefore, the operating status of distribution networks is no longer solely driven by load changes but is also significantly affected by the uncertainty of distributed generation output. Against this backdrop, the node voltage distribution, branch power flow, and local power balance relationships in distribution networks exhibit greater dynamism and complexity, posing higher demands on operation and control.

[0003] In actual operation, complex electrical coupling relationships are formed between different nodes due to differences in line impedance, topology, and load and power supply configurations. These coupling relationships directly affect the active power distribution path and reactive power regulation effect. At the same time, the output of distributed power sources has obvious random fluctuation characteristics, causing the net injected power of nodes and its changing trend to change continuously over time, thus having a continuous impact on the peak shaving, frequency regulation, and voltage control of the distribution network.

[0004] Traditional methods lack the ability to dynamically respond to changes in electrical coupling relationships and operating states between nodes, leading to the failure of partitioning results when load fluctuates or renewable energy output changes. There is also a lack of effective connection between cluster partitioning and scheduling control, making it difficult to directly use partitioning results for control and allocation, resulting in low regulation efficiency. Some methods do not fully consider the differences in node regulation capabilities and reactive power support capacity limitations, which can easily cause local overload or voltage over-limit problems. Furthermore, they lack an adaptive adjustment mechanism when the system operating state changes, making it difficult to meet the requirements for the flexibility and stability of the distribution network under the condition of high proportion of distributed power source access.

[0005] Therefore, based on the above-mentioned characteristics of power distribution network operation and control requirements, it is necessary to construct a distributed power generation cluster partitioning and coordinated regulation method that can reflect the electrical relationships, power characteristics and voltage regulation requirements of nodes, so as to achieve the rational allocation and stable regulation of power in the power distribution network. Summary of the Invention

[0006] This invention provides a dynamic partitioning method for distributed generation clusters in distribution networks based on multiple control objectives. This addresses the shortcomings of traditional methods, which lack dynamic response capabilities to changes in electrical coupling relationships and operating states between nodes, leading to partitioning failures when load fluctuates or renewable energy output changes. Furthermore, the lack of effective connection between cluster partitioning and dispatch control makes it difficult to directly use partitioning results for control allocation, resulting in low regulation efficiency. Some methods also fail to fully consider differences in node regulation capabilities and reactive power support limitations, easily causing local overloads or voltage exceedances. Finally, the lack of an adaptive adjustment mechanism when system operating states change makes it difficult to meet the requirements for flexibility and stability of the distribution network under conditions of high-proportion distributed generation access.

[0007] The present invention provides a dynamic partitioning method for distributed generation clusters in a distribution network based on multiple control objectives, which specifically includes the following steps:

[0008] S1. Collect operational data of candidate nodes in the distribution network and construct node state variables; calculate node cohesion based on operational data and node state variables.

[0009] S2. Initialize the nodes as candidate clusters. Based on the node cohesion and the node state variables, calculate the cluster performance value. Determine the optimal cluster partition through recursive merging and performance verification to obtain the final cluster partition result. Based on the final cluster partition result, obtain the total active power regulation demand of the cluster layer. Combined with the cluster performance value, convert the total active power regulation demand of the cluster layer into the final active power regulation command of the node.

[0010] Preferably, the operational data of the candidate nodes specifically includes the active power of the distributed power source of the node, the load power of the node, the voltage amplitude of the node, the reactive load of the node, the impedance modulus of the line between nodes, and the rated capacity of the inverter; based on the operational data of the candidate nodes, a unified node state quantity is constructed, including the net injected power of the node, the net power change of the node, the voltage demand of the node, and the reactive power support margin of the node.

[0011] Preferably, based on the inter-node line impedance modulus, node net injected power, node net power change, node voltage demand, and node reactive power support margin, the corresponding normalized scaling parameters are calculated, and the corresponding normalized sub-items are constructed; based on the normalized sub-items, the node cohesion is calculated.

[0012] Preferably, the cluster performance value is calculated in the following way:

[0013] Based on node cohesion, the average cohesion within the cluster is constructed; based on the net injected power, net power change, and reactive power support margin of nodes within the candidate cluster, the net injected power, net power change, and reactive power support margin of the cluster are constructed; the external node set of the candidate cluster is constructed, and the average cohesion of cross-cluster node pairs is calculated to obtain the external coupling degree of the cluster; based on the average cohesion within the cluster, the net injected power, net power change, reactive power support margin, and external coupling degree of the cluster, the cluster efficiency value is calculated.

[0014] Preferably, the specific implementation process of the recursive merging and performance verification includes:

[0015] Calculate the average cohesion between any two candidate clusters and select the two clusters with the highest current average cohesion as candidate merge targets. Perform a trial merge on the candidate merge targets to obtain a new temporary cluster and calculate the efficiency value of the new cluster. Take a weighted average of the efficiency values ​​of the two original sub-clusters in the unmerged state and compare it with the efficiency value of the new cluster to judge the rationality of the cluster merge. If the cluster merge is reasonable, officially retain the merge and update the cluster structure. If the cluster merge is unreasonable, cancel the merge and mark the cluster pair as unmergeable. Continue to select the next pair of clusters with the highest average cohesion from the remaining mergeable cluster pairs and repeat the above process until the requirement of improved efficiency after merging cannot be met, then stop merging and output the final cluster partitioning result.

[0016] Preferably, the difference between the net injected power of the cluster and the target value of the net injected power of the cluster in the final cluster partitioning result is taken as the total active power regulation demand of the cluster layer, and corrected in combination with the cluster efficiency value to obtain the effective regulation amount after cluster efficiency correction; the target value of the net injected power of the cluster is calculated based on the node load power of the cluster.

[0017] Preferably, based on the effective adjustment amount after cluster performance correction, and combined with the node allocation weight, the total active power adjustment demand of the cluster layer is transformed into the power adjustment target value that the node actually needs to execute, thus obtaining the final active power adjustment command of the node.

[0018] Preferably, the specific calculation method for the node allocation weight is as follows:

[0019] For any node in the final cluster partitioning result, based on the node's operating data and node state variables, construct the node's residual active power regulation capability and equivalent active power regulation capability, and calculate the node's comprehensive regulation capability; based on the node's comprehensive regulation capability and the node's average cohesion, construct the node allocation weight; the node's average cohesion is obtained by calculating the arithmetic mean of the node's cohesion and that of all other nodes in the cluster.

[0020] The beneficial effects of the technical solution of the present invention are:

[0021] 1. This invention introduces multiple control objectives such as active power matching degree, power change complementarity and reactive power support capability during the cluster formation process. By verifying the results through a unified performance criterion (cluster performance value), the resulting cluster is not only structurally compact, but also has higher consistency and coordination in operation functions such as frequency regulation, peak regulation and voltage regulation, thereby significantly improving the actual effectiveness of the cluster as a regulation unit.

[0022] 2. In the control allocation phase, this invention adaptively corrects the adjustment commands based on the cluster structure quality. When the internal coupling of the cluster is high and the operating state is stable, the adjustment capability is fully utilized. When the cluster boundary does not match the operating state, the adjustment commands automatically converge, thereby avoiding excessive control over local unstable areas and improving the system's operational safety.

[0023] 3. This invention achieves a reasonable distribution of regulation tasks within the cluster through a joint allocation mechanism that combines node regulation capabilities with inter-node coupling relationships. This prioritizes nodes with higher regulation capabilities and better matching with the cluster structure to undertake regulation tasks. Under complex operating conditions such as rapid changes in source load, fluctuations in distributed power output, and changes in network topology, it maintains the continuity between cluster partitioning and regulation strategies, enabling the partitioning results to directly serve scheduling and control. Furthermore, it achieves boundary adaptive optimization through rolling updates, enhancing the distribution network's adaptability to uncertainties. Attached Figure Description

[0024] Figure 1 This is a flowchart of a dynamic partitioning method for distributed power generation clusters in a distribution network based on multiple control objectives, as described in this invention. Detailed Implementation

[0025] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0026] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0027] The following description, in conjunction with the accompanying drawings, details a specific scheme for a dynamic partitioning method for distributed power generation clusters in a distribution network based on multiple control objectives, provided by this invention.

[0028] See attached document Figure 1 The diagram illustrates a flowchart of a dynamic partitioning method for distributed power generation clusters in a distribution network based on multiple control objectives, provided by an embodiment of the present invention. The method includes the following steps:

[0029] S1. Collect operational data of candidate nodes in the distribution network and construct node state variables; calculate node cohesion based on operational data and node state variables.

[0030] Operational data of a candidate node set is collected through a distribution automation system. This candidate node set includes nodes already connected to distributed generation sources, energy storage devices, or with adjustable loads in the current distribution network's basic operational structure, as well as adjacent nodes directly electrically connected to these nodes via switching equipment, reflecting the actual electrical coupling relationships. For each candidate node, the active power of the distributed generation source is read by the monitoring module built into the distributed generation inverter. The node load power is read by the smart meter on the low-voltage side of the distribution transformer. The node voltage amplitude is read through the distribution automation terminal. The reactive load at the nodes is collected through the reactive power metering module. Offline reading of inter-node line impedance modulus values ​​from the distribution network line parameter database. And read the inverter's rated capacity from the equipment nameplate. The above and Representing nodes respectively and , Indicates the time.

[0031] Based on all acquired operational data, a unified node state variable is constructed: node net injection power. Net power change at nodes ,in, Represents a node At any moment Net injected power at the node; node voltage requirement , The target voltage is the rated voltage of the distribution network; the reactive power support margin at the nodes is also considered. To ensure that different physical quantities can be included in a unified calculation process, the average absolute value of each quantity within the candidate node set is taken as the normalization scale parameter. Specifically, within the current candidate node set, the average value of the line impedance magnitude between all node pairs (i.e., the average line impedance magnitude) is calculated. The average of the absolute values ​​of net injected power across all nodes (i.e., the average absolute value of net injected power). The average of the absolute values ​​of the net power change at all nodes (i.e., the average absolute value of the net power change). The average of the absolute values ​​of the reactive power support margin of all nodes (i.e., the average absolute value of the reactive power support margin). And the average voltage demand of all nodes (i.e., average voltage demand). The above-mentioned normalized scaling parameters can all be obtained by calculating the arithmetic mean based on data collected or further constructed within the current scheduling period.

[0032] Based on the above normalized scaling parameters, the normalized sub-items are constructed as follows: First, the inter-node line impedance magnitude is... With the average impedance modulus of the line Ratio calculations are performed to construct an electrical distance normalization term, which characterizes the strength of electrical coupling between nodes; a larger value indicates a greater electrical distance. Subsequently, the node... With nodes The absolute value of the difference in net injected power at the nodes and compared with the absolute value of average net injected power A ratio calculation is performed to construct a net injected power difference term, which characterizes the degree of consistency between the two nodes in terms of active power supply and demand structure. Represents a node At any moment The node net injected power; then, by calculating the node With nodes The absolute value of the difference between the net power changes at the nodes and the absolute value of the average net power change A ratio calculation is performed to construct a net power change consistency term, which reflects the degree of synchronization between the two nodes in their peak-shaving and frequency-modulation behaviors. Represents a node At any moment The change in net power at the nodes; further, by calculating the nodes With nodes The absolute value of the difference between the reactive power support margins of the nodes and the absolute value of the average reactive power support margin. A ratio calculation is performed to construct a reactive power support capability difference term. This term characterizes the consistency of the two nodes' capabilities when participating in voltage regulation. Represents a node At any moment The reactive power support margin of the nodes; finally, by calculating the sum of the voltage demands of the two nodes. and twice the average voltage requirement By performing a ratio calculation, a dimensionless voltage demand intensity term is obtained. This term characterizes the voltage regulation pressure level experienced by the two nodes at the current moment. Represents a node At any moment The node voltage requirement.

[0033] After constructing the normalized components as described above, all dimensionless components are summed, and a constant term of 1 is added to form a unified denominator. This constructs the formula for calculating nodal cohesion, compressing multidimensional state variables with different physical meanings into a single dimensionless scalar, and ensuring that each component is numerically comparable and contributes in the same direction to the overall result. The specific formula for calculating nodal cohesion is as follows: , in, Indicates time node With nodes The degree of cohesion between them; It is used to prevent positive numbers with a denominator of 0, and can take values ​​of... ; Represents the electrical distance normalization term; This indicates the net injected power difference item; This indicates a consistency term for net power change; This indicates the difference in reactive power support capability. This represents the voltage demand intensity. Node cohesion comprehensively reflects the electrical distance between nodes, net active power difference, consistency of power change, difference in reactive power support capacity, and voltage regulation pressure. The higher the cohesion, the more suitable the two nodes are to be assigned to the same cluster.

[0034] The above-mentioned node cohesion The calculation formula can extend the single distance in traditional electrical distance methods and similarity clustering methods (such as clustering based on Euclidean distance or correlation coefficient) to a fusion of multiple physical quantities, including electrical distance, active power difference, ramp difference, non-functional capacity difference, and voltage demand, after unified normalization.

[0035] S2. Initialize the nodes as candidate clusters. Based on the node cohesion and the node state variables, calculate the cluster performance value. Determine the optimal cluster partition through recursive merging and performance verification to obtain the final cluster partition result. Based on the final cluster partition result, obtain the total active power regulation demand of the cluster layer. Combined with the cluster performance value, convert the total active power regulation demand of the cluster layer into the final active power regulation command of the node.

[0036] After calculating the node cohesion, the cluster construction and verification process is initiated using the node cohesion as input. Candidate clusters are gradually formed, and their compliance with the distribution network operation and regulation requirements is evaluated. The specific implementation process is as follows:

[0037] First, define the average cohesion within the cluster for any candidate cluster. Let its set of nodes be . The number of nodes is Any two different nodes within the candidate cluster The node cohesion is By summing the node cohesion of all node pairs and calculating the average cohesion within the cluster, the higher the value, the greater the degree of coordination among nodes within the candidate cluster in terms of power, reactive power, and voltage regulation requirements; for candidate clusters The net injected power of all nodes within the cluster is summed to obtain the cluster's net injected power. To evaluate the source-load matching degree within the cluster, the ratio of the absolute value of the cluster's net injected power to the sum of the absolute values ​​of the net injected power of each node is calculated to reflect the source-load complementarity within the cluster. For candidate clusters... The net power change of all nodes within the cluster is summed to obtain the cluster's net power change. To measure the synchronization of power changes within the cluster, the ratio of the absolute value of the cluster's net power change to the sum of the absolute values ​​of the net power changes of each node is calculated. When the node changes are complementary, this ratio decreases, indicating that the candidate cluster has better peak-shaving and frequency-modulating capabilities. The candidate cluster is then... The reactive power support margin of all nodes within the cluster is summed to obtain the cluster reactive power support margin. The ratio of the absolute value of the cluster reactive power support margin to the sum of the absolute values ​​of the reactive power support margins of each node is defined as the reactive power coordination evaluation metric, used to reflect the degree of reactive power supply and demand matching within the cluster. For candidate clusters... Its external node set Calculate the average cohesion of all cross-cluster node pairs and define it as the cluster external coupling degree; the external node set refers to the candidate node set excluding candidate clusters. All remaining nodes other than the included nodes;

[0038] Furthermore, based on the fact that all the aforementioned intermediate variables are clearly defined and calculable, the node-level cohesion results in the modularity index and source-load balance evaluation method (such as community-based modularity + power balance index) of the power system are elevated to the cluster-level evaluation. "Net active power offset, ramp offset, reactive power imbalance, and external coupling" are introduced as penalty terms, and all these factors are integrated to form a unified cluster performance value. This is used as the sole criterion for determining whether cluster partitioning is reasonable and whether the current cluster structure should be retained. The expression is: , in, Candidate clusters At any moment The performance value; This represents the number of nodes outside the cluster. A set of external nodes of the cluster; Indicates the average cohesion within the cluster; Net power injection into the cluster; This represents the change in the cluster's net power. This provides a reactive power support margin for the cluster. This is a measure for reactive power coordination evaluation. For external coupling of the cluster, Indicates time node With nodes The degree of cohesion between them.

[0039] Based on the definitions of the intermediate variables and performance values ​​mentioned above, the specific steps for cluster partitioning are as follows: First, initialize each node as an independent cluster and calculate the average cohesion between any two clusters; then, select the two clusters with the highest current average cohesion as candidate merge targets, perform a trial merge to obtain a new temporary cluster, and calculate the performance value of the new cluster; simultaneously, calculate the weighted average of the performance values ​​of the original two sub-clusters in the unmerged state, which can be weighted by the number of nodes or the total power. For example, the formula for weighting by the number of nodes is as follows: (Number of nodes in cluster 1 × Performance value of cluster 1 + Number of nodes in cluster 2 × Performance value of cluster 2) / (Number of nodes in cluster 1 + Number of nodes in cluster 2). If the new... If the cluster efficiency value is greater than the weighted value, it indicates that the cluster merger is reasonable. After the merger, the overall performance is improved in terms of active power matching, ramp complementarity, reactive power balance, and external decoupling. In this case, the merger is officially retained, and the cluster structure is updated. Otherwise, it indicates that the cluster merger is unreasonable. The merger will weaken the cluster's ability as a regulating unit. In this case, the merger is canceled, and the cluster pair is marked as no longer eligible for merging. Then, the next pair of clusters with the highest average cohesion is selected from the remaining mergeable cluster pairs, and the above process is repeated. When all candidate cluster pairs fail to pass the "efficiency improvement after merger" criterion, it indicates that the current partition has reached the optimal cluster structure under this operating state. At this point, merging is stopped, and the final cluster partition result is output.

[0040] After calculating the cluster performance value and determining the final cluster partitioning result at the current moment, it is necessary to convert the total active power regulation requirement of the cluster layer into active power regulation instructions that can be executed at the node layer.

[0041] Specifically, the clusters in the final cluster partitioning result any node within Determine its remaining active power regulation capacity. This refers to the current active power margin available for upward or downward adjustment at a node. This amount can be calculated by comparing the node's rated capacity with the active power of its distributed generation sources. The difference is obtained from the device nameplate, and is the inverter's rated capacity. Limits in the active power direction; subsequently, reactive power support margins at nodes. Perform equivalent conversion to include the reactive power support margin at the nodes. The smaller of the inverter's rated capacity and its remaining capacity is used as the equivalent active power regulation capability to eliminate the dimensional difference between reactive and active power. The inverter's remaining capacity is determined by adjusting the inverter's rated capacity. Active power of distributed power sources at nodes We obtain the square root of the difference of squares, denoted as . The sum of the remaining active power regulation capacity and the equivalent active power regulation capacity is defined as the node's comprehensive regulation capacity; then, the node's comprehensive regulation capacity is calculated. With cluster The arithmetic mean of the nodal cohesion of all other nodes is used to obtain the node. In the cluster Average cohesion within , or average node cohesion, is used to represent the node's average cohesion. With cluster The average coupling strength of other nodes within the cluster; the product of the node's comprehensive regulation capacity and the node's average cohesion is defined as the node's allocation weight, representing the comprehensive weight of the node's regulation task within the cluster; the difference between the cluster's net injected power and the target value of the cluster's net injected power is taken as the total active power regulation demand of the cluster layer, and the target value of the cluster's net injected power is obtained by calculating the sum of the load power of all nodes within the cluster; based on the total active power regulation demand of the cluster layer... Combined with the cluster performance value and the node allocation weights, the final active power adjustment command of the node is calculated, and its expression is: , in, It is a node At any moment The final active power regulation command is the actual power regulation target value that the node needs to execute; Represents a node At any moment Comprehensive regulatory capacity; This indicates the weights assigned to the nodes; This indicates the proportion of nodes allocated within the cluster; This represents the effective adjustment amount after cluster performance correction. The closer to 1, the more stable the internal structure of the cluster and the better the matching of adjustment capabilities; adjustment commands are basically not reduced. When the value is reduced, the denominator increases, indicating that the cluster boundary matches the current operating state less, thereby automatically reducing the actual amount of adjustment sent to the nodes and avoiding applying excessive control intensity to unstable clusters.

[0042] The final active power regulation command of the above nodes The calculation formula expands the allocation weight in the proportional allocation strategy and droop control concept (allocating power according to capacity or regulation capability) in power dispatch from "single capacity" to "regulation capability × cohesion", and introduces a cluster effectiveness value. Adaptive reduction of total instructions.

[0043] In summary, a dynamic partitioning method for distributed power generation clusters in distribution networks based on multiple control objectives has been developed.

[0044] The order of the embodiments is for illustrative purposes only and does not represent the superiority or inferiority of the embodiments. The processes depicted in the drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0045] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0046] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A dynamic partitioning method for distributed generation clusters in a distribution network based on multiple control objectives, characterized in that, Specifically, the following steps are included: S1. Collect operational data of candidate nodes in the distribution network and construct node state variables; calculate node cohesion based on operational data and node state variables. S2. Initialize the nodes as candidate clusters, calculate the cluster performance value based on the node cohesion and the node state variables, and determine the optimal cluster partitioning through recursive merging and performance verification to obtain the final cluster partitioning result. Based on the final cluster partitioning results, the total active power regulation requirement of the cluster layer is obtained. By combining the cluster performance value, the total active power regulation requirement of the cluster layer is transformed into the final active power regulation command of the node.

2. The method for dynamic partitioning of distributed power generation clusters in a distribution network based on multiple control objectives as described in claim 1, characterized in that, The operational data of the candidate nodes specifically includes the active power of the distributed power source at the node, the load power at the node, the voltage amplitude at the node, the reactive load at the node, the impedance modulus of the line between nodes, and the rated capacity of the inverter. Based on the operational data of the candidate nodes, a unified node state quantity is constructed, including the net injected power at the node, the net power change at the node, the voltage demand at the node, and the reactive power support margin at the node.

3. The method for dynamic partitioning of distributed power generation clusters in a distribution network based on multiple control objectives as described in claim 2, characterized in that, Based on the inter-node line impedance modulus, node net injected power, node net power change, node voltage demand, and node reactive power support margin, the corresponding normalized scaling parameters are calculated and the corresponding normalized sub-items are constructed. The nodal cohesion is calculated based on the normalized components.

4. The method for dynamic partitioning of distributed power generation clusters in a distribution network based on multiple control objectives as described in claim 3, characterized in that, The specific calculation method for the cluster performance value is as follows: Based on node cohesion, the average cohesion within the cluster is constructed; based on the net injected power, net power change, and reactive power support margin of nodes within the candidate cluster, the net injected power, net power change, and reactive power support margin of the cluster are constructed; the external node set of the candidate cluster is constructed, and the average cohesion of cross-cluster node pairs is calculated to obtain the external coupling degree of the cluster; based on the average cohesion within the cluster, the net injected power, net power change, reactive power support margin, and external coupling degree of the cluster, the cluster efficiency value is calculated.

5. The method for dynamic partitioning of distributed power generation clusters in a distribution network based on multiple control objectives as described in claim 4, characterized in that, The specific implementation process of the recursive merging and performance verification includes: Calculate the average cohesion between any two candidate clusters and select the two clusters with the highest current average cohesion as candidate merge targets. Perform a trial merge on the candidate merge targets to obtain a new temporary cluster and calculate the efficiency value of the new cluster. Take a weighted average of the efficiency values ​​of the two original sub-clusters in the unmerged state and compare it with the efficiency value of the new cluster to judge the rationality of the cluster merge. If the cluster merge is reasonable, officially retain the merge and update the cluster structure. If the cluster merge is unreasonable, cancel the merge and mark the cluster pair as unmergeable. Continue to select the next pair of clusters with the highest average cohesion from the remaining mergeable cluster pairs and repeat the above process until the requirement of improved efficiency after merging cannot be met, then stop merging and output the final cluster partitioning result.

6. The method for dynamic partitioning of distributed power generation clusters in a distribution network based on multiple control objectives as described in claim 5, characterized in that, The difference between the net injected power of the cluster and the target value of the net injected power of the cluster in the final cluster partitioning result is taken as the total active power regulation demand of the cluster layer, and corrected in combination with the cluster efficiency value to obtain the effective regulation amount after cluster efficiency correction; the target value of the net injected power of the cluster is calculated based on the node load power of the cluster.

7. The method for dynamic partitioning of distributed power generation clusters in a distribution network based on multiple control objectives as described in claim 6, characterized in that, Based on the effective adjustment amount after cluster performance correction, and combined with the node allocation weight, the total active power adjustment demand of the cluster layer is transformed into the power adjustment target value that the node actually needs to execute, thus obtaining the final active power adjustment command of the node.

8. The method for dynamic partitioning of distributed power generation clusters in a distribution network based on multiple control objectives as described in claim 7, characterized in that, The specific calculation method for the node allocation weight is as follows: For any node in the final cluster partitioning result, based on the node's operating data and node state variables, construct the node's residual active power regulation capacity and equivalent active power regulation capacity, and calculate the node's comprehensive regulation capacity. Based on the comprehensive adjustment capability of nodes and the average cohesion of nodes, a node allocation weight is constructed; The average node cohesion is obtained by calculating the arithmetic mean of the node cohesion of the node and all other nodes in the cluster.

Images