Multi-index coupling-based dc power distribution network scenario division method and system

Abstract

The method and system for dividing scenes of direct current distribution network based on multi-index coupling first extract the voltage amplitude, injected active power, voltage ripple ratio and voltage fluctuation of each node, calculate the comprehensive index and classify the nodes; then obtain the electrical distance between nodes based on the similarity of the comprehensive index, and calculate the dynamic equivalent electrical distance between the power supply node and the adjacent load node; the load nodes meeting the electrical distance condition are preferentially clustered, and the remaining nodes complete secondary clustering with other power supply nodes to form clusters consistent with the number of power supply nodes. Finally, compare the electrical distance in the cluster with the threshold value to complete the division of typical operation scenes. The application can accurately identify the typical scenes in the direct current distribution network, and make up for the shortcomings of the traditional method, such as lack of dynamic adaptability, difficulty in reflecting the difference of power quality and unclear characteristics of different operation modes, so as to improve the scientificity, operation efficiency and safety of system scheduling.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of power systems, and in particular relates to a DC distribution network scene division method and system based on multi-index coupling. BACKGROUND

[0002] With the rapid development of renewable energy and the increasing demand for smart grid construction, DC systems have attracted widespread attention due to their high efficiency, reliability and other advantages. However, how to reasonably divide the typical scenes of DC systems to adapt to different load demands and operating conditions has become a problem to be solved.

[0003] Traditional scene division is often based on voltage levels, topological structures, new energy access, etc. However, different types of power sources and loads have different sensitivities to voltage fluctuations and harmonics, and unified power quality evaluation methods have not been perfected. The output characteristics of new energy, energy storage, grid-side rectifiers and electric vehicle charging piles in DC systems are quite different. The traditional scene division method cannot be well applied to DC distribution networks with complex source and load characteristics.

[0004] The patent document with publication number CN118539428A discloses a method for dividing clusters in an AC / DC hybrid distribution network, which divides the nodes and branches of the AC / DC hybrid distribution network system into sub-regions to obtain a sub-region set; a data analysis model is established to generate a pre-cluster division evaluation coefficient; the sub-regions are classified according to the pre-cluster division evaluation coefficient to determine the cluster regions. However, this method cannot flexibly adjust the clustering results according to the real-time operating state, especially when facing large-scale or complex systems, this lack of adaptability of the division method cannot meet the needs of efficient scheduling and optimal resource allocation of the system. For example, the prior art (A Cluster Division Method for Large-Scale Distributed Photovoltaic Access Distribution Network, Kou Lingfeng, et al., Renewable Energy, April 2019) analyzes the relevant characteristics of distributed photovoltaic power sources and loads in the distribution network, selects appropriate cluster division indicators, considers the influence of electrical distance between nodes, forms a network with photovoltaic power source nodes in the distribution network, introduces and optimizes the complex network module function, and uses clustering algorithm for photovoltaic cluster division. However, this method only performs scene division in grid-connected mode and does not dynamically consider the influence of different operating modes on cluster division. SUMMARY

[0005] The present application is to solve the problems of lack of dynamic adaptability in traditional scene division, difficulty in reflecting power quality differences and unclear characteristics under different operating modes. The following technical solutions are adopted.

[0006] The present application proposes a DC distribution network scene division method based on multi-index coupling, which includes:

[0007] extracting voltage amplitude, injected active power, voltage ripple ratio and voltage fluctuation of each node in the direct current system as the index of node clustering and scenario division;

[0008] calculating the comprehensive index of node i at time t based on the four indexes of the node; classifying the nodes by the positive and negative of the comprehensive index of each node, the types including load nodes and power supply nodes; calculating the similarity degree of the comprehensive index of each pair of nodes, and calculating the electrical distance between each pair of nodes based on the similarity degree;

[0009] for each power supply node, obtaining its adjacent nodes, comparing the electrical distance between the power supply node and each node in the adjacent nodes with the equivalent electrical distance between the power supply node and all the adjacent nodes, and preferentially clustering the nodes in the adjacent nodes that meet the conditions with the power supply node; for the nodes in the adjacent nodes that are not clustered, performing secondary clustering, and finally obtaining clusters with the same number of power supply nodes as the center of each power supply node;

[0010] calculating the intra-cluster electrical distance between each node in each cluster and the power supply node in the cluster, and dividing the cluster according to the intra-cluster electrical distance.

[0011] Further preferably, the calculation formula of the comprehensive index of node i at time t is as follows:

[0012]

[0013] wherein, the comprehensive index of node i at time t, positive indicating that the node is a load node, and negative indicating that the node is a power supply node, the power quality normalization coefficient, the voltage power coupling coefficient at time t, related to the change rate of voltage amplitude with time and the change rate of actual injected power with time, 、 、 and voltage amplitude, injected active power, voltage ripple ratio and voltage fluctuation value of node i respectively.

[0014] Further preferably, the formula for calculating the similarity degree of the comprehensive index of two nodes is:

[0015]

[0016]

[0017] wherein, is the similarity degree of the comprehensive index between node i and node j, the value closer to 1 means the two nodes are more similar, is the maximum value of the product of the comprehensive index of any two nodes in the DC system at time t.

[0018] Further preferably, based on the similarity degree, the electrical distance between each pair of nodes is calculated, and the formula is:

[0019]

[0020] In the formula, is the electrical distance between node i and node j, when is 0, the similarity degree of the two nodes i and j is the highest.

[0021] Further preferably, the method of priority clustering is:

[0022] Suppose there are N nodes in the system, of which there are m power nodes, select a power node , and let the adjacent load node set be N( ), calculate the equivalent electrical distance between the power node and its adjacent load nodes ;

[0023] If the electrical distance between node i and the power node is , then node i is clustered with the power node , otherwise not. Repeat until the electrical distance between all nodes in N( ) and is calculated.

[0024] Go to the next power node and repeat the above steps until the power node is clustered, and the priority clustering is completed.

[0025] Further preferably, suppose the adjacent nodes of node p are nodes 1~n, calculate the equivalent electrical distance between node p and its adjacent nodes at time t, and the formula is:

[0026]

[0027] In the formula, is the equivalent electrical distance between node p and its adjacent nodes, is the corresponding weight of the electrical distance , and is the electrical distance between node p and i.

[0028] Further preferably, the calculation formula of the weight is:

[0029]

[0030] In the formula, For time-series weighting functions, The operating state weight function has different values ​​for daytime and nighttime. During the day, the time-series weight function is related to the node's voltage ripple rate, rated voltage, and voltage amplitude. At night, the time-series weight function is related to the node's rated power and injected active power. Similarly, the operating state weight function has different values ​​for grid-connected and islanded operation. During grid-connected operation, the operating state weight function is related to the node's rated power, voltage ripple rate, and voltage fluctuation. During islanded operation, the operating state weight function is related to the node's rated power, rated voltage, voltage amplitude, and injected active power.

[0031] Further preferably, for the nodes among the adjacent nodes that have not been clustered, a secondary clustering is performed, the method being:

[0032] For m power nodes in the system, from power node Begin by calculating the power supply sections separately. Click to With power nodes The set of adjacent load nodes N ( Equivalent electrical distance of unclustered nodes in ) arrive In equivalent electrical distance arrive Take the minimum value Corresponding power node , will N( Unclustered nodes and power nodes Clustering, then proceeding to the next power node. ;

[0033] Repeat the above steps to calculate the power nodes. to To set N ( Equivalent electrical distance of unclustered nodes in ) arrive , will N( The unclustered nodes in the load are clustered with the power supply nodes corresponding to the minimum equivalent electrical distance, until all load nodes are clustered, forming m clusters with the power supply nodes as the cluster centers, and the secondary clustering is completed.

[0034] Further preferred, based on the calculated intra-cluster electrical distances of m clusters, and the strong coupling threshold... and equivalent electrical distance weak coupling threshold Comparing the intra-cluster electrical distance of the i-th cluster wherein, if the cluster center is node , the cluster is divided into the data center scenario; if if the injection power of node is positive and the voltage fluctuation value is greater than the first fluctuation threshold, the cluster with the cluster center being node is divided into the new energy direct current gathering and sending out scenario, if the injection power of node is negative and the voltage fluctuation value is less than the second fluctuation threshold, the cluster with the cluster center being node is divided into the off-grid hydrogen production scenario; if the cluster with the cluster center being node is divided into the rail transit scenario.

[0035] The application further provides a direct current power distribution network scenario division system based on the method.

[0036] The index extraction module extracts the voltage amplitude, injection active power, voltage ripple rate and voltage fluctuation of each node in the direct current system as the index for node clustering and scenario division.

[0037] The node electrical distance calculation module calculates the comprehensive index of node i at time t based on the four indexes of the node; the node is classified through the positive and negative of the comprehensive index of each node, and the types include load nodes and power supply nodes; the similarity degree of the comprehensive index of each pair of nodes is calculated, and the electrical distance between each pair of nodes is calculated based on the similarity degree.

[0038] The load clustering module obtains the adjacent nodes of each power supply node, compares the electrical distance between each node in the adjacent nodes and the power supply node with the equivalent electrical distance between the power supply node and all the adjacent nodes, and preferentially clusters the nodes in the adjacent nodes that meet the conditions with the power supply node; for the nodes in the adjacent nodes that are not clustered, secondary clustering is performed, and finally the same number of clusters as the number of power supply nodes is obtained, with each power supply node as the center.

[0039] The scenario division module calculates the intra-cluster electrical distance between each node in each cluster and the power supply node in the cluster, and divides the cluster according to the intra-cluster electrical distance.

[0040] The application has at least the following advantages compared with the prior art:

[0041] The application proposes a set of scientific and reasonable scene division indexes according to the characteristics of the direct current power distribution network and the scene characteristics, combined with part of the direct current system power quality indexes, and proposes an equivalent electrical distance calculation method based on the division indexes and the system reference impedance, calculates the dynamic electrical distance between nodes, realizes the reasonable quantification of the closeness of multiple nodes in clustering, and finally completes the scene division based on the size comparison of the dynamic node equivalent electrical distance and the threshold value, clearly defines the source and load characteristics under different operation scenes, realizes the targeted operation control and optimization configuration, solves the problems of lack of dynamic adaptability in scene division, difficulty in reflecting the power quality difference and unclear characteristics under different operation modes in the traditional method. The method proposed in the application provides effective support for new energy consumption, data center power supply reliability, off-grid hydrogen production system stability and rail transit power supply safety, etc. application scenes, thereby improving the operation efficiency and safety of the direct current power distribution network.

[0042] The method proposed in the application can comprehensively consider the actual operation of the direct current system, the operation characteristics of each scene, calculate the equivalent electrical distance through a scientific method, and thus realize more reasonable scene division.

[0043] The application first defines the operation scene of the direct current power distribution network, proposes scene division indexes combined with the actual operation of the direct current system, then couples the indexes to obtain the electrical distance between nodes, then proposes a multi-node equivalent electrical distance calculation method and clusters the nodes in the system, and finally calculates the equivalent electrical distance according to the clustering result to realize scientific and reasonable scene division. BRIEF DESCRIPTION OF DRAWINGS

[0044] Figure 1 is a flow chart of a direct current power distribution network scene division method and system based on multi-index coupling proposed by the application.

[0045] Figure 2 is a schematic diagram of a direct current power distribution network. DETAILED DESCRIPTION

[0046] In order to make the purpose, technical scheme and advantages of the application clearer, the technical scheme of the application will be described clearly and completely below in combination with the drawings in the embodiments of the application. The embodiments described in the application are only a part of the embodiments of the application, not all the embodiments. Based on the spirit of the application, all other embodiments obtained by those skilled in the art without creative labor fall within the protection scope of the application.

[0047] To address the shortcomings of existing technologies, this invention provides a method and system for classifying DC distribution network scenarios based on multi-index coupling. Based on system operation and load characteristics, scenario classification indicators are proposed. Considering time and operating modes, dynamic weights of responses are calculated to obtain the equivalent electrical distances of multiple nodes. All load nodes are clustered, and the equivalent electrical distances are calculated based on the clustering results. Different scenario classifications are completed by comparing the magnitudes of the equivalent electrical distances. This helps improve the scientific rigor and accuracy of the decision-making process, effectively enhances system flexibility and adaptability, promotes more efficient DC distribution network design and operation, and lays the foundation for advancing source-load interaction in DC virtual power plants.

[0048] This invention proposes a method for classifying DC distribution network scenarios based on multi-index coupling, such as... Figure 1 As shown, it includes:

[0049] Step 1 specifically includes the following steps:

[0050] Step 101, as follows Figure 2 As shown, in a DC distribution network, AC power is rectified to generate DC power to supply energy to the loads, or new energy sources in the system are directly connected to the DC bus to supply power to the system. Simultaneously, an energy management module is equipped to coordinate photovoltaic, energy storage, and loads, achieving economic dispatch and smooth power fluctuations. Typical stable operation scenarios for DC distribution systems mainly include: new energy DC aggregation and transmission, data centers, off-grid hydrogen production, and rail transit.

[0051] Step 102: In the scenario of new energy DC aggregation and transmission, master-slave control can be adopted, with the designated master converter stabilizing the DC voltage, and the remaining slave converters using MPPT control to ensure that the photovoltaic operates at the maximum power point; data centers mainly adopt dual closed-loop control of converter voltage and current to ensure stable output DC voltage; the hydrogen production electrolysis system integrates photovoltaic power generation and energy storage as power sources, and provides power for hydrogen production through a power converter. The electrolysis hydrogen production equipment requires a stable and reliable low-voltage DC power supply to produce hydrogen, and at the same time, it places high demands on the system's power quality; taking the subway as an example, the rail transit system needs to maintain the DC grid voltage within the specified range, and the vehicle's own PWM converter control needs to achieve precise speed regulation and electric braking of the traction motor.

[0052] Step 103: Construct scenario segmentation indicators. Based on the operating scenario and control objectives of the DC distribution system, and considering the characteristics of the DC system itself, such as no reactive power and no frequency, extract the voltage amplitude of each node in the system. Injecting active power Voltage ripple and voltage fluctuations This serves as an indicator for node clustering and scene partitioning. Specifically, the positive direction of power is taken as the direction of injection into the node. To provide a positive illustration of the actual power injection node, For negative actual power flow out of the node.

[0053] Table 1 Index characteristics under each scenario

[0054]

[0055] Step 2 specifically includes the following steps:

[0056] Step 201, considering the four indexes of the node, calculate the comprehensive index of the node at time t.

[0057]

[0058] In the formula, The comprehensive index of node i at time t is positive, indicating that the node is a load node, and negative, indicating that the node is a power node, The power quality normalization coefficient is close to 1, indicating that the node is highly sensitive to power quality at this time, and the constant impedance load node is 0.2~0.4, and the constant power load node is 0.7~1, The voltage power coupling coefficient at time t is:

[0059]

[0060] In the formula, The rate of change of the node voltage amplitude with time at t is The rate of change of the actual value of the node injected power with time at t is The sensitivity of the voltage of different nodes to power fluctuations can be quantified, which plays a role in optimizing node clustering and scenario division, and improves the flexibility of voltage and power fluctuation control, ensuring stable operation and efficient scheduling of the system.

[0061] By calculating the comprehensive index of the node, the power node and the load node can be clearly distinguished, which helps to accurately analyze the role and function of each node in the system. This is crucial for subsequent scenario division and resource optimization.

[0062] Step 202, calculate the similarity of the comprehensive indexes of two nodes, which is:

[0063]

[0064]

[0065] In the formula, sij(t) is the similarity of the comprehensive indexes between node i and node j, and the closer the value is to 1, the more similar the two nodes are, and M(t) is the maximum value of the product of the comprehensive indexes of any two nodes in the system at t.

[0066] Here represents the relevance or intersection of the two nodes comprehensive indicators, reflects the upper limit of the system comprehensive indicator, in this case, can reflect the proximity of the relative maximum possible value of the two nodes. The closer to 1, the closer the product of the comprehensive indicator values of node i and node j to the upper limit of the system theoretical indicator, and their indicator values are very similar, that is, the power quality and control strategy of the two nodes are relatively close; on the contrary, if the ratio is low, it indicates that the indicators between the nodes are quite different, and the nodes belong to different types or different working states.

[0067] Step 203, in order to quantify the closeness between nodes more conveniently, based on the similarity of the comprehensive indicators between nodes, the electrical distance between two nodes is calculated, which is:

[0068]

[0069] In the formula, dij(t) is the electrical distance between node i and node j, when dij(t) is 0, it means that sij(t) is 1, that is, the similarity of the two nodes i and j is the highest.

[0070] The comprehensive indicator combines key factors such as voltage amplitude, power injection, voltage ripple rate and voltage fluctuation, helping the system to accurately distinguish between load nodes and power supply nodes, thereby achieving more accurate node clustering and scenario division. This method not only enhances the ability to judge the similarity between nodes, but also effectively improves the system's scheduling ability in different scenarios, especially in the case of new energy access, it can better regulate power quality and optimize system operation efficiency. By quantifying the electrical distance between nodes, the relationship between nodes can be more intuitively presented, making the subsequent scenario division and control strategy formulation more scientific and accurate.

[0071] The step 3 specifically includes the following steps:

[0072] Step 301, assuming that the adjacent nodes of node p are nodes 1~n, the equivalent electrical distance between node p and its adjacent nodes at time t is calculated, as shown in the following formula.

[0073]

[0074] In the formula, is the equivalent electrical distance between node p and its adjacent nodes, is the electrical distance The corresponding weight is:

[0075]

[0076] In the formula, For time-series weighting functions, The running state weight function specifically includes:

[0077]

[0078]

[0079] In the formula, Let be the rated power of node p at time t. Let be the rated voltage of node p at time t.

[0080] During the day, when photovoltaic power generation is strong, voltage fluctuations and ripples have a significant impact on the system. Therefore, voltage-related indicators must be considered when calculating time-series weights. At night, the system load is low, and power injection changes are small, so considering the fluctuation of injected power is more realistic. During grid-connected operation, node voltage fluctuations and ripples have a significant impact on system stability, so the above two indicators must be considered when calculating state weights. However, during islanded operation, ensuring voltage and power support is paramount, so voltage amplitude and injected power indicators must be effectively considered.

[0081] By calculating the equivalent electrical distance between nodes and introducing weight functions for timing and operating status, the electrical distance between nodes can be dynamically evaluated, further enhancing the control over system stability. The timing weight function considers different photovoltaic power generation conditions during the day and night, thereby optimizing calculations for different time periods and ensuring the system's optimal operating state under different environments. The operating status weight is dynamically adjusted according to the specific operating mode of the system (such as grid-connected or islanded operation) to ensure that voltage and power support are adequately guaranteed under different operating conditions.

[0082] Step 302: Assume there are N nodes in the system, of which m are power nodes. Select one power node. Let the set of its adjacent load nodes be N( ), calculate power nodes Equivalent electrical distance to its adjacent load node .

[0083] Step 303, if node i and power node electrical distance Then connect node i with the power node. Clustering is completed if the clustering is not completed, otherwise no clustering is performed, until N( All nodes within) and Electrical distance calculation complete. Proceed to the next power node. Repeat this step until the power node is reached. Complete clustering.

[0084] Step 304, from the power node Start, respectively, calculate the power supply node to The equivalent electrical distance of the unclustered nodes in the set N( ) to , the minimum value in the equivalent electrical distance to , the power supply node , the unclustered nodes in N( ) are clustered with the power supply node , and then go to the next power supply node .

[0085] Step 305, repeat the above steps to calculate the power supply node to to the equivalent electrical distance of the unclustered nodes in the set N( ) to , the unclustered nodes in N( ) are clustered with the power supply node corresponding to the minimum value of the equivalent electrical distance, until all load nodes are clustered, forming m clusters with the power supply node as the cluster center.

[0086] The application can dynamically adjust the clustering result according to the actual electrical distance and is suitable for power systems of different scales and structures. First, the adjacent load nodes with qualified electrical distance are preferentially clustered by each power supply node, and then the unclustered nodes are assigned to the power supply corresponding to the minimum value by comparing the equivalent electrical distances of multiple power supply nodes. This way not only ensures that the adjacent and closely electrically connected loads are preferentially assigned to the nearest power supply, but also allows the loads in the multi-power supply range to be assigned to the "electrically closest" power supply, avoiding unreasonable allocation and achieving optimal matching of loads and power supplies. According to the order of "initial clustering → secondary allocation of unclustered nodes → iteration to complete full load clustering", the logic is clear and the process is clear, without the need for complex global optimization calculation

[0087] The step 4 specifically includes the following steps:

[0088] Step 401, respectively calculate the intra-cluster electrical distance between the m power supply nodes at this time and the power supply nodes within their respective clusters .

[0089] Step 402, scene division according to the intra-cluster electrical distance. As shown in Table 2, if , the cluster with the cluster center node is divided into the data center scenario, which mostly uses short-distance dense wiring and has high voltage sensitivity; if , if node ​if the injection power of the node is positive and the voltage fluctuation value is greater than the first fluctuation threshold, the cluster center is divided into a node The new energy direct current gathering and sending scene is divided into a node if the injection power of the node is negative and the voltage fluctuation value is less than the second fluctuation threshold, the cluster center is divided into a node The off-grid hydrogen production scene is divided into a node if the injection power of the node is positive and the voltage fluctuation value is greater than the first fluctuation threshold, the cluster center is divided into a node The traction scene is divided into a node, and the catenary impedance of the scene is significant. Since the purpose of the application is to divide the typical operation scene of the direct current power distribution network, in the case of only the scene in which the injection power of the node is positive and the voltage fluctuation value is greater than the first fluctuation threshold, and the scene in which the injection power of the node is negative and the voltage fluctuation value is less than the second fluctuation threshold are considered. The scene in which the injection power of the node is positive and the voltage fluctuation is small indicates that the operation characteristics of the node are closer to a conventional steady-state power supply or load, rather than a typical new energy direct current gathering and sending scene, and do not belong to the application range of the application; the scene in which the injection power of the node is negative and the voltage fluctuation is large indicates that the hydrogen production and other constant power loads cannot operate normally, and this state can only appear for a short time under transient or abnormal conditions, and it is difficult to form a stable working condition, so it is also not within the scope of the discussion of the application.

[0090] wherein dx and dd are respectively the equivalent electrical distance strong coupling threshold and weak coupling threshold, which need to be determined according to the performance requirements of the system.

[0091] Step 403, realizing non-differential aggregation of power supplies, photovoltaic, energy storage and loads, optimizing virtual power plant resource configuration, and completing system scene division.

[0092] Table 2: Scene division based on equivalent electrical distance

[0093]

[0094] Embodiment two

[0095] The application further provides a direct current power distribution network scene division system applying the method in embodiment one, which comprises an index extraction module, an electrical distance calculation module, a load clustering module and a scene division module.

[0096] The index extraction module extracts the voltage amplitude, injection active power, voltage ripple rate and voltage fluctuation of each node in the direct current system as the index for node clustering and scene division.

[0097] The node electrical distance calculation module calculates the comprehensive index of node i at time t based on four indexes of the node; the node is classified by the positive and negative of the comprehensive index of each node, and the types include load nodes and power nodes; the similarity degree of the comprehensive index of each pair of nodes is calculated, and the electrical distance between each pair of nodes is calculated based on the similarity degree;

[0098] The load clustering module obtains the adjacent nodes of each power node, compares the electrical distance between the power node and each node of the adjacent nodes with the equivalent electrical distance between the power node and all the adjacent nodes, and preferentially clusters the power node with the nodes of the adjacent nodes that meet the condition; the unclustered nodes of the adjacent nodes are secondarily clustered, and finally the same number of clusters as the power nodes are obtained, with each power node as the center of the cluster;

[0099] The scene division module calculates the intra-cluster electrical distance between each node in each cluster and the power node in the cluster, and divides the cluster according to the intra-cluster electrical distance.

[0100] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application and not to limit them. Although the present application has been described in detail with reference to the above embodiments, those skilled in the art should understand that the specific embodiments of the present application can be modified or replaced by equivalents without departing from the spirit and scope of the present application, and any modification or equivalent replacement should be covered within the protection scope of the claims of the present application.

Claims

1. A DC distribution network scenario partitioning method based on multi-index coupling, characterized in that, include: The voltage amplitude, injected active power, voltage ripple rate, and voltage fluctuation of each node in the DC system are extracted as indicators for node clustering and scene segmentation. The comprehensive index of node i at time t is calculated based on four indexes of the node; the nodes are classified according to the positive or negative value of the comprehensive index of each node, including load nodes and power supply nodes. Calculate the similarity of the comprehensive indicators for each pair of nodes, and based on the similarity, calculate the electrical distance between each pair of nodes; For each power node, its neighboring nodes are obtained. The electrical distance between the power node and each of its neighboring nodes is compared with the equivalent electrical distance between the power node and all of its neighboring nodes. The neighboring nodes that meet the conditions are preferentially clustered with the power node. For the neighboring nodes that are not clustered, a second clustering is performed to finally obtain a cluster centered on each power node, which is the same as the number of power nodes. Calculate the intra-cluster electrical distance between each node in the cluster and the power node in the cluster, and divide the cluster into scenarios based on the intra-cluster electrical distance.

2. The DC distribution network scenario partitioning method based on multi-index coupling according to claim 1, characterized in that: The formula for calculating the comprehensive index of node i at time t is as follows: In the formula, This is a comprehensive index of node i at time t. A positive value indicates that the node is a load node, and a negative value indicates that the node is a power source node. The power quality normalization factor. The voltage-power coupling coefficient at time t is related to the rate of change of voltage amplitude with time and the rate of change of the actual value of injected power with time. , , and These represent the voltage amplitude, injected active power, voltage ripple rate, and voltage fluctuation value at node i, respectively.

3. The DC distribution network scenario partitioning method based on multi-index coupling according to claim 1, characterized in that: The formula for calculating the similarity of the comprehensive index between two nodes is: In the formula, This represents the similarity of the comprehensive index between node i and node j. The closer the value is to 1, the more similar the two nodes' indices are. It is the maximum value of the product of the comprehensive indexes of any two nodes in the DC system at time t.

4. The DC distribution network scenario partitioning method based on multi-index coupling according to claim 1, characterized in that: Based on similarity, the electrical distance between each pair of nodes is calculated using the following formula: In the formula, Let i be the electrical distance between node i and node j, when When the value is 0, the similarity between the indicators of nodes i and j is the highest.

5. The DC distribution network scenario partitioning method based on multi-index coupling according to claim 1, characterized in that: The priority clustering method is: Suppose there are N nodes in the system, of which m are power supply nodes. Select one power supply node. Let the set of its adjacent load nodes be N( ), calculate power nodes Equivalent electrical distance to its adjacent load node ; If node i and power node electrical distance Then connect node i with the power node. Clustering is completed if clustering is not completed, otherwise no clustering is performed, until N( All nodes within) and The electrical distance calculation is complete; Proceed to the next power node Repeat the above steps until the power node is reached. Clustering is complete; priority clustering ends here.

6. The DC distribution network scenario partitioning method based on multi-index coupling according to claim 5, characterized in that: Let the neighboring nodes of node p be nodes 1 to n. Calculate the equivalent electrical distance between node p and its neighboring nodes at time t using the following formula: In the formula, Let p be the equivalent electrical distance between node p and its neighboring nodes. electrical distance The corresponding weights Let be the electrical distance between nodes p and i.

7. The DC distribution network scenario partitioning method based on multi-index coupling according to claim 6, characterized in that: Weight The calculation formula is: In the formula, For time-series weighting functions, The operating state weight function has different values ​​for daytime and nighttime. During the day, the time-series weight function is related to the node's voltage ripple rate, rated voltage, and voltage amplitude. At night, the time-series weight function is related to the node's rated power and injected active power. Similarly, the operating state weight function has different values ​​for grid-connected and islanded operation. During grid-connected operation, the operating state weight function is related to the node's rated power, voltage ripple rate, and voltage fluctuation. During islanded operation, the operating state weight function is related to the node's rated power, rated voltage, voltage amplitude, and injected active power.

8. The DC distribution network scenario partitioning method based on multi-index coupling according to claim 1, characterized in that: For the nodes that are not clustered among the adjacent nodes, a secondary clustering is performed, using the following method: For m power nodes in the system, from power node Begin by calculating the power nodes separately. to With power nodes The set of adjacent load nodes N ( Equivalent electrical distance of unclustered nodes in ) arrive In equivalent electrical distance arrive Take the minimum value Corresponding power node , will N( Unclustered nodes and power nodes Clustering, then proceeding to the next power node. ; Repeat the above steps to calculate the power nodes. to To set N ( Equivalent electrical distance of unclustered nodes in ) arrive , will N( The unclustered nodes in the load are clustered with the power supply nodes corresponding to the minimum equivalent electrical distance, until all load nodes are clustered, forming m clusters with the power supply nodes as the cluster centers, and the secondary clustering is completed.

9. The DC distribution network scenario partitioning method based on multi-index coupling according to claim 1, characterized in that: Based on the calculated intra-cluster electrical distances of m clusters, and the strong coupling threshold... and equivalent electrical distance weak coupling threshold Comparison, if the intra-cluster electrical distance of the i-th cluster is... ,in, Then the cluster center is taken as the node. Clusters are allocated to data center scenarios; if If node When the injected power is positive and the voltage fluctuation value is greater than the first fluctuation threshold, the cluster center is taken as the node. Clusters are assigned to new energy DC aggregation and transmission scenarios, if nodes When the injected power is negative and the voltage fluctuation value is less than the second fluctuation threshold, the cluster center is taken as the node. Clusters are assigned to off-grid hydrogen production scenarios; if Then the cluster center is taken as the node. The clusters are assigned to the rail transit scenario.

10. A DC distribution network scenario segmentation system based on the method of any one of claims 1-9, comprising an index extraction module, an electrical distance calculation module, a load clustering module, and a scenario segmentation module, characterized in that: The indicator extraction module extracts the voltage amplitude, injected active power, voltage ripple rate, and voltage fluctuation of each node in the DC system as indicators for node clustering and scenario classification. The node electrical distance calculation module calculates the comprehensive index of node i at time t based on four indicators of the node; the nodes are classified according to the positive or negative value of the comprehensive index of each node, including load nodes and power supply nodes. Calculate the similarity of the comprehensive indicators for each pair of nodes, and based on the similarity, calculate the electrical distance between each pair of nodes; The load clustering module, for each power node, obtains its neighboring nodes, compares the electrical distance between the power node and each of its neighboring nodes with the equivalent electrical distance between the power node and all of its neighboring nodes, and prioritizes clustering the neighboring nodes that meet the conditions with the power node; for the neighboring nodes that are not clustered, a secondary clustering is performed, and finally a cluster centered on each power node is obtained with the same number of power nodes; The scene segmentation module calculates the intra-cluster electrical distance between the nodes in each cluster and the power nodes in the cluster, and segments the clusters into scenes based on the intra-cluster electrical distance.

Images