A power distribution network planning scenario division method based on information entropy reduction minimum criterion

Abstract

The application discloses a power distribution network planning scene division method based on a minimum information entropy reduction criterion, which comprises the following steps: 1, pre-dividing the annual operation scene of the power distribution network into multiple period scenes; 2, in each period scene, combining a voltage sensitivity index and a new energy access position to determine a key node set of each time period; 3, performing deviation standardization processing on the voltage, load and new energy output of the key node, calculating the information entropy of each time period, and aggregating the global information entropy of the period scene; 4, merging adjacent time periods, selecting an optimal merging scheme based on the minimum information entropy reduction criterion, and combining an elbow rule to determine the final number of time periods; 5, adopting a DBSCAN algorithm to cluster different slice scene clusters, and outputting core points in each slice scene cluster as typical slice scenes. When the power distribution network planning scene is divided, the time sequence characteristics and the spatial distribution characteristics of the scene are comprehensively considered, and the accuracy and practicability of the divided slice scenes are improved.

Description

Technical Field

[0001] This invention relates to the field of power system planning and dispatching, specifically to a method for dividing distribution network planning scenarios based on the criterion of minimizing information entropy. Background Technology

[0002] With the large-scale integration of various resources such as distributed renewable energy, energy storage devices, and responsive loads, the operating characteristics of distribution networks exhibit strong volatility, high uncertainty, and regional differences, posing significant challenges to distribution network planning and scheduling. Current distribution network planning methods typically use typical days as time-series operating scenarios for modeling and analysis, thereby simplifying model size and computational complexity. However, the typical day method presents a contradiction between computational complexity and scenario representativeness: on the one hand, to improve scenario coverage, more typical days need to be introduced, significantly increasing model size and solution difficulty; on the other hand, selecting too few typical days or simplifying scenarios to reduce computational complexity leads to an incomplete description of system operating characteristics, making it difficult to reflect the dynamic changes of source and load resources in time and space. This invention addresses these problems by researching a method for dividing multi-period operating scenarios in distribution network planning. Based on the spatiotemporal characteristics of the distribution network's source and load, it dynamically determines the duration, category, and number of distribution network planning scenarios, providing a representative and computationally efficient operating scenario foundation for new distribution network planning under the background of large-scale uncertain source and load resource integration. Summary of the Invention

[0003] To overcome the shortcomings of existing technologies, this invention proposes a distribution network planning scenario segmentation method based on the information entropy minimization criterion. This method aims to comprehensively consider the time series characteristics and spatial key node differences of the distribution network, scientifically segment the distribution network planning scenarios, and extract typical slice scenarios based on the segmentation results. This enables the extraction of more accurate and representative typical operating scenarios, providing a scenario foundation for subsequent distribution network optimization planning and scheduling.

[0004] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0005] The present invention provides a method for classifying distribution network planning scenarios based on the criterion of minimizing information entropy, characterized by the following steps:

[0006] Step 1: Collect data for the annual operation scenario of the distribution network plan, including: voltage of distribution network nodes, load output of distribution network nodes, and renewable energy output of distribution network nodes. Divide the annual operation scenario into m periodic scenarios according to a fixed period T; and select the j-th time period within the s-th periodic scenario. s The voltage of each distribution network node is denoted as Load output is recorded as New energy output is recorded as , s∈[1,m];

[0007] Step 2: In the s-th cycle scenario, identify the set of voltage-sensitive nodes based on the voltage sensitivity index. And combined with the set of new energy access points Determine the set of key nodes in each time period. ;

[0008] Step 3: Use deviation standardization method to... The voltage, load and new energy output of each key node in the distribution network are normalized, and the information entropy of each time period is calculated using the normalized data to obtain the global information entropy of the distribution network in the s-th cycle scenario.

[0009] Step 4: Merge adjacent time periods within the s-th period scenario to obtain various merging schemes within the s-th period scenario; then, using Step 3, calculate the global information entropy of each merging scheme, and select the optimal merging scheme based on the minimum information entropy reduction criterion. Continue merging the optimal merging schemes until no time periods can be merged, thereby determining the optimal number of time periods within the s-th period scenario using the elbow method. And divide the s-th cycle scenario into A slice of scene;

[0010] Step 5: Following the process of Steps 2 to 4, obtain the slice scenes divided into m periodic scenes respectively, and use the DBSCAN clustering algorithm to cluster all slice scenes to obtain different scene clusters, and identify the typical slice scenes in each scene cluster.

[0011] The characteristic of the distribution network planning scenario division method based on the information entropy minimization criterion described in this invention is that step two is performed as follows:

[0012] Step 2.1: Use equation (1) to calculate the j-th time period in the ith time period within the s-th cycle scenario. s Voltage sensitivity index of each distribution network node ;

[0013] (1)

[0014] In formula (1): For the i-th time period within the s-th cycle scenario, excluding the j-th time period s In addition to the load of each distribution network node, the j-th node obtained when calculating the power flow of the distribution network. s Voltage of each distribution network node;

[0015] Step 2.2: Within the s-th cycle scenario, select the distribution network nodes whose voltage sensitivity index is closest to "1" for each time period. Each distribution network node is assigned a voltage-sensitive node for each time period, thus forming a set of voltage-sensitive nodes for each time period within the s-th cycle scenario. ;

[0016] Step 2.3: Connect the data accessed in each time period within the s-th cycle scenario. Each new energy node and its corresponding time period By merging the voltage-sensitive nodes, we obtain the set of key nodes for each time period within the s-th cycle scenario. .

[0017] Furthermore, step three is performed as follows:

[0018] Step 3.1: Use the deviation normalization method to analyze the voltage of the j-th critical node in the i-th time period within the s-th cycle scenario. Load output New energy power output Normalization is performed to obtain the normalized voltage of the j-th key node in the i-th time period within the s-th cycle scenario. Load output New energy power output ;

[0019] Step 3.2: Calculate the voltage probability of the j-th critical node in the i-th time period within the s-th cycle scenario using equations (2), (3), and (4). Load probability and the probability of new energy output ;

[0020] (2)

[0021] (3)

[0022] (4)

[0023] In equations (2)-(4): Let n be the duration of the i-th time period. s The number of time periods in the s-th cycle scenario;

[0024] Step 3.3: Calculate the voltage information entropy of the j-th key node in the i-th time period within the s-th cycle scenario using equations (5), (6), and (7). Load information entropy Information entropy of new energy output ;

[0025] (5)

[0026] (6)

[0027] (7)

[0028] Step 3.4: Calculate the comprehensive information entropy of the j-th key node in the i-th time period within the s-th cycle scenario using equation (8). ;

[0029] (8)

[0030] In equation (8): Let be the voltage information entropy weight of the j-th key node in the i-th time period within the s-th cycle scenario. Let be the load information entropy weight of the j-th key node in the i-th time period within the s-th cycle scenario. The entropy weight of the new energy output information of the j-th key node in the i-th time period within the s-th cycle scenario;

[0031] Step 3.5: Calculate the global information entropy H of the s-th cycle scene using equation (9). s ;

[0032] (9).

[0033] Furthermore, step four is performed as follows:

[0034] Step 4.1: Merge adjacent time periods within the s-th period scene to obtain... Various merger schemes, and The number of time periods for each scheme is ,and Calculate the global information entropy for each merging scheme using steps 3.2-3.5, denoted as . The scheme with the smallest reduction in information entropy is selected as the first optimal merging scheme, and the information entropy of the optimal merging scheme is denoted as . ,in, Indicates the first The global information entropy of the merging schemes;

[0035] Step 4.2: Following the process in Step 4.1, analyze the first optimal merging scheme. Adjacent time periods within the same period are merged and calculated to obtain the number of time periods in the merging scheme. and the second optimal merging scheme ,and ; and thus obtain The optimal merging scheme and the number of time periods for each of the schemes ;

[0036] Step 4.3, based on the number of time periods The horizontal axis represents the information entropy of the corresponding merging scheme. Using the vertical axis as the coordinate system, construct the coordinates corresponding to all solutions. Thus, using equation (10), a system is constructed with the starting coordinates (n) as the basis. s H s ) and endpoint coordinates ( , Baseline between:

[0037] (10)

[0038] Step 4.3: Calculate the coordinates of the y-th scheme using equation (11). , Distance to baseline Thus, the starting coordinates (n) are obtained. s H s ) and endpoint coordinates ( , The distance from the baseline to the coordinates of each scheme between ) The number of time periods corresponding to the scheme with the furthest distance is selected as the optimal number of time periods in the s-th cycle scenario, denoted as . :

[0039] (11).

[0040] Furthermore, step five is performed as follows:

[0041] Step 5.1: Assume that the m-period scenes are divided into N slice scenes X1, X2, ..., X... N Calculate the dynamic time-warped distance between all slice scenes, let d(X) a X b ) represents the scene X of the a-th slice. a With the b-th slice scene X b The dynamic time-normalized distance between them;

[0042] Step 5.2: Set the radius parameter Eps and the minimum number of neighbors MinPts for the DBSCAN clustering algorithm;

[0043] Step 5.3, when X a When the number of slice scenes contained within a neighborhood with a radius of Eps is greater than or equal to MinPts, X will be... a Mark it as a core slice scene;

[0044] Within any core slice scene x, if there are other slice scenes in the neighborhood with a radius of Eps, the corresponding slice scene will be merged into the same cluster as the core slice scene x.

[0045] Within any core slice scene x, if there are multiple core slice scenes in the region with a radius of Eps, the range of the cluster corresponding to the core slice scene x is expanded until no new slice scenes are added to the cluster corresponding to x, thus forming a complete scene cluster of x.

[0046] Step 5.4: Extract all core slice scenes from each scene cluster and output them as typical slice scenes.

[0047] The present invention provides an electronic device, comprising a memory and a processor, wherein the memory is used to store a program that supports the processor in executing the power distribution network planning scenario division method, and the processor is configured to execute the program stored in the memory.

[0048] The present invention discloses a computer-readable storage medium on which a computer program is stored, wherein the computer program, when executed by a processor, performs the steps of the power distribution network planning scenario division method.

[0049] Compared with existing technologies, the beneficial effects of this invention are reflected in:

[0050] 1. This invention standardizes the multi-dimensional characteristic information of voltage, load, and renewable energy output of key nodes in the distribution network operation, and introduces the information entropy index to characterize the complexity and uncertainty of the system operation status. This can more accurately reflect the temporal changes of operation characteristics and ensure the information fidelity in the process of distribution network planning scenario slicing and typical scenario extraction.

[0051] 2. This invention uses the minimum reduction of information entropy as the optimality criterion in the time period merging process, guides the dynamic merging of adjacent time periods within a periodic scenario, and achieves adaptive determination of the number of scenario segments based on the elbow rule, thereby minimizing the loss of information while realizing the division of power distribution network scenarios.

[0052] 3. Based on the results of the overall distribution network planning scenario division, this invention uses the DBSCAN density clustering algorithm to cluster all slices into different clusters, and identifies core points in each cluster as representative typical scenarios, providing a scenario basis for the optimization planning and scheduling optimization problem of new distribution networks under the background of high penetration renewable energy access.

[0053] 4. By deeply exploring the spatiotemporal characteristics of distribution network planning scenarios, the scenario division method proposed in this invention can provide a scenario foundation for a new distribution network planning method oriented towards multi-dimensional interaction of "source-grid-load-storage". While improving the representativeness of typical scenarios, it effectively reduces computational complexity and enhances the overall operational economy and scheduling intelligence of the system. Attached Figure Description

[0054] Figure 1 A flowchart of a distribution network planning scenario partitioning method based on the minimum information entropy reduction criterion;

[0055] Figure 2 This is a schematic diagram of the scene slice results for the 288th cycle. Detailed Implementation

[0056] In this example, a distribution network planning scenario partitioning method based on the information entropy minimization criterion is applied to the IEEE 33-node test system. It aims to resolve the contradiction between computational complexity and scenario representativeness in existing planning methods. By comprehensively considering the temporal and spatial distribution characteristics of distribution network planning scenarios, it quantifies system uncertainty through information entropy theory and combines a dynamic identification mechanism for key nodes with an adaptive mechanism for the number of scenario segments. This achieves effective partitioning of distribution network planning scenarios, overcoming the contradiction between scenario representativeness and computational complexity inherent in traditional distribution network planning that uses typical days as temporal operation scenarios. This provides a scenario foundation for the optimization planning and scheduling optimization problems of new distribution networks under the background of high-penetration renewable energy access. Specifically, see [link to relevant documentation]. Figure 1 The method is carried out according to the following steps:

[0057] Step 1: Collect data for the annual operation scenario of the distribution network plan, including: voltage of distribution network nodes, load output of distribution network nodes, and renewable energy output of distribution network nodes. Divide the annual operation scenario into m periodic scenarios according to a fixed period T; and select the j-th time period within the s-th periodic scenario. s The voltage of each distribution network node is denoted as Load output is recorded as New energy output is recorded as , s∈[1,m]; In this example, the planning scenario is 8760 hours per year, the period of the divided periodic scenario is 24 hours, the number of divided periodic scenarios is 365, the load output and new energy output data of the nodes are calculated based on the actual output data of a certain place, and the voltage data of the nodes are calculated by applying power flow calculation in the IEEE 33-node test system.

[0058] Step 2: In the s-th cycle scenario, identify the set of voltage-sensitive nodes based on the voltage sensitivity index. And combined with the set of new energy access points Determine the set of key nodes in each time period. Because distribution networks typically have a large scale and complex topology, comprehensively considering the information entropy of all nodes could lead to a sharp increase in computational load without necessarily improving computational accuracy, and might not accurately reflect the contributions of critical nodes. New energy access nodes are usually the points where renewable energy sources such as wind and solar power are integrated. The voltage and output fluctuation characteristics of these nodes are closely related, and due to the intermittency and volatility of new energy sources, they can have a significant impact on the distribution network at different times. Voltage-weak nodes are usually located at the edge of the system or in areas with high loads. Their information entropy reflects the intensity and uncertainty of voltage fluctuations under external disturbances. Especially under conditions of load fluctuations, equipment failures, or changes in grid topology, the entropy value of voltage-weak nodes will increase significantly, indicating that the system may face a higher risk of voltage instability. Therefore, these two types of nodes are selected as critical nodes in the distribution network.

[0059] Step 2.1: Use equation (1) to calculate the j-th time period in the ith time period within the s-th cycle scenario. s Voltage sensitivity index of each distribution network node ;

[0060] (1)

[0061] In formula (1): For the i-th time period within the s-th cycle scenario, excluding the j-th time period s In addition to the load of each distribution network node, the j-th node obtained when calculating the power flow of the distribution network. s Voltage of each distribution network node;

[0062] Step 2.2: Within the s-th cycle scenario, select the distribution network nodes whose voltage sensitivity index is closest to "1" for each time period. Each distribution network node is assigned a voltage-sensitive node for each time period, thus forming a set of voltage-sensitive nodes for each time period within the s-th cycle scenario. ;

[0063] Step 2.3: Connect the data accessed in each time period within the s-th cycle scenario. Each new energy node and its corresponding time period By merging the voltage-sensitive nodes, we obtain the set of key nodes for each time period within the s-th cycle scenario. In this example, the number of voltage-sensitive nodes selected is the same as the number of new energy access nodes.

[0064] Step 3: Use deviation standardization method to... The voltage, load and new energy output of each key node in the distribution network are normalized, and the information entropy of each time period is calculated using the normalized data to obtain the global information entropy of the distribution network in the s-th cycle scenario.

[0065] Step 3.1: Use the deviation normalization method to analyze the voltage of the j-th critical node in the i-th time period within the s-th cycle scenario. Load output New energy power output Normalization is performed to obtain the normalized voltage of the j-th key node in the i-th time period within the s-th cycle scenario. Load output New energy power output ;

[0066] Step 3.2: Calculate the voltage probability of the j-th critical node in the i-th time period within the s-th cycle scenario using equations (2), (3), and (4). Load probability and the probability of new energy output ;

[0067] (2)

[0068] (3)

[0069] (4)

[0070] In equations (2)-(4): Let n be the duration of the i-th time period. s Let be the number of time periods in the s-th cycle scenario.

[0071] Step 3.3: Calculate the voltage information entropy of the j-th key node in the i-th time period within the s-th cycle scenario using equations (5), (6), and (7). Load information entropy Information entropy of new energy output ;

[0072] (5)

[0073] (6)

[0074] (7)

[0075] Step 3.4: Calculate the comprehensive information entropy of the j-th key node in the i-th time period within the s-th cycle scenario using equation (8). ;

[0076] (8)

[0077] In equation (8): Let be the voltage information entropy weight of the j-th key node in the i-th time period within the s-th cycle scenario. Let be the load information entropy weight of the j-th key node in the i-th time period within the s-th cycle scenario. The entropy weight of the new energy output information of the j-th key node in the i-th time period within the s-th cycle scenario.

[0078] Step 3.5: Calculate the global information entropy H of the s-th cycle scene using equation (9). s ;

[0079] (9)

[0080] Step 4: Merge adjacent time periods within the s-th period scenario to obtain various merging schemes within the s-th period scenario; then, using Step 3, calculate the global information entropy of each merging scheme, and select the optimal merging scheme based on the minimum information entropy reduction criterion. Continue merging the optimal merging schemes until no time periods can be merged, thereby determining the optimal number of time periods within the s-th period scenario using the elbow method. And divide the s-th cycle scenario into A slice of scene;

[0081] Step 4.1: Merge adjacent time periods within the s-th period scene to obtain... Various merger schemes, and The number of time periods for each scheme is ,and Calculate the global information entropy for each merging scheme using steps 3.2-3.5, denoted as . The scheme with the smallest reduction in information entropy is selected as the first optimal merging scheme, and the information entropy of the optimal merging scheme is denoted as . ,in, Indicates the first The global information entropy of the merging schemes;

[0082] Step 4.2: Following the process in Step 4.1, analyze the first optimal merging scheme. Adjacent time periods within the same period are merged and calculated to obtain the number of time periods in the merging scheme. and the second optimal merging scheme ,and ; and thus obtain The optimal merging scheme and the number of time periods for each of the schemes .

[0083] Step 4.3, based on the number of time periods The horizontal axis represents the information entropy of the corresponding merging scheme. Using the vertical axis as the coordinate system, construct the coordinates corresponding to all solutions. Thus, using equation (10), a system is constructed with the starting coordinates (n) as the basis. s H s ) and endpoint coordinates ( , Baseline between:

[0084] (10)

[0085] Step 4.3: Calculate the coordinates of the y-th scheme using equation (11). , Distance to baseline Thus, the starting coordinates (n) are obtained. s H s ) and endpoint coordinates ( , The distance from the baseline to the coordinates of each scheme between ) The number of time periods corresponding to the scheme with the furthest distance is selected as the optimal number of time periods in the s-th cycle scenario, denoted as . :

[0086] (11)

[0087] Step 5: Following the process of Steps 2 to 4, obtain the slice scenes divided into m periodic scenes respectively, and use the DBSCAN clustering algorithm to cluster all slice scenes to obtain different scene clusters, and identify the typical slice scenes in each scene cluster.

[0088] Step 5.1: Assume that the m-period scenes are divided into N slice scenes X1, X2, ..., X... N Calculate the dynamic time-warped distance between all slice scenes, let d(X) a X b ) represents the scene X of the a-th slice. a With the b-th slice scene X b The dynamic time-normalized distance between them;

[0089] Step 5.2: Set the radius parameter Eps and the minimum number of neighbors MinPts for the DBSCAN clustering algorithm;

[0090] Step 5.3, when X a When the number of slice scenes contained within a neighborhood with a radius of Eps is greater than or equal to MinPts, X will be... a Mark it as a core slice scene;

[0091] Within any core slice scene x, if there are other slice scenes in the neighborhood with a radius of Eps, the corresponding slice scene will be merged into the same cluster as the core slice scene x.

[0092] Within any core slice scene x, if there are multiple core slice scenes in the region with a radius of Eps, the range of the cluster corresponding to the core slice scene x is expanded until no new slice scenes are added to the cluster corresponding to x, thus forming a complete scene cluster of x.

[0093] Step 5.4: Extract all core slice scenes from each scene cluster and output them as typical slice scenes.

[0094] In this embodiment, an electronic device includes a memory and a processor. The memory stores a program that supports the processor in executing the above-described method, and the processor is configured to execute the program stored in the memory.

[0095] In this embodiment, a computer-readable storage medium stores a computer program, which is executed by a processor to perform the steps of the above method.

[0096] The experiment verifying the present invention "a method for classifying urban power distribution network planning scenarios based on the criterion of minimizing information entropy" mainly includes the following steps:

[0097] 1) The test system is constructed using the IEEE 33-node system.

[0098] The system has four new energy power generation systems located at nodes 6, 17, 21, and 28 respectively. Taking the 288th day of the annual operation scenario as an example, the effectiveness of the scenario division method proposed in this invention is verified.

[0099] This represents the total net load of the distribution network at various times on day 288.

[0100] surface Total net load of the IEEE 33-node system at various times on day 288

[0101]

[0102] Voltage-weak nodes calculated using the voltage sensitivity index, such as... As shown.

[0103] surface Voltage weak point numbering for each time period

[0104]

[0105] Power flow calculations are used to calculate the time-series changes in voltage at each node. Global information entropy of the distribution network is calculated using renewable energy output information, load information, and voltage information, thereby enabling the division of cyclical scenarios. The results of this division are as follows: As shown:

[0106] surface Slice results table for the 288th cycle scene

[0107]

[0108] Figure 2 This diagram illustrates the net load curve and its segmentation for the 288th slice period. In the diagram, the first period (0:00-5:00) falls between late night and early morning. During this time, residential and commercial activities are largely suspended, and the overall load level remains at its lowest point of the day. System load fluctuations are small, and operating characteristics are relatively stable, indicating a basic power supply guarantee. The second period (6:00-8:00) is the morning start-up period. As residents' daily activities gradually resume and heating loads are rapidly released, the system load shows a significant increase, forming a morning peak characteristic. The third period (9:00-12:00) enters the daytime growth phase. Although distributed renewable energy sources such as photovoltaics begin to connect to the grid, the total system load continues to rise due to the further release of industrial and commercial loads, and the slope of the load curve remains positive. The fourth period (13:00-16:00) is the afternoon load stabilization period, with a slight decrease in load but still remaining at a high level, and the system operation is relatively stable. The fifth period (17:00-20:00) is the evening peak period, with the system load rising rapidly due to residents returning home from get off work and the combined effects of lighting and heating loads, reaching the peak for the day against the backdrop of a significant decline in photovoltaic output. The sixth period (21:00-24:00) is the nighttime load decline period. As most electricity demand gradually decreases, the load drops rapidly, and the system enters a low-load operating range again, with the operating situation tending to be stable.

Claims

1. A method for dividing distribution network planning scenarios based on the criterion of minimizing information entropy, characterized in that, The procedure is as follows: Step 1: Collect data for the annual operation scenario of the distribution network plan, including: voltage of distribution network nodes, load output of distribution network nodes, and renewable energy output of distribution network nodes. Divide the annual operation scenario into m periodic scenarios according to a fixed period T; and select the j-th time period within the s-th periodic scenario. s The voltage of each distribution network node is denoted as Load output is recorded as New energy output is recorded as , s∈[1,m]; Step 2: In the s-th cycle scenario, identify the set of voltage-sensitive nodes based on the voltage sensitivity index. And combined with the set of new energy access points Determine the set of key nodes in each time period. ; Step 3: Use deviation standardization method to... The voltage, load and new energy output of each key node in the distribution network are normalized, and the information entropy of each time period is calculated using the normalized data to obtain the global information entropy of the distribution network in the s-th cycle scenario. Step 4: Merge adjacent time periods within the s-th period scenario to obtain various merging schemes within the s-th period scenario; then, using Step 3, calculate the global information entropy of each merging scheme, and select the optimal merging scheme based on the minimum information entropy reduction criterion. Continue merging the optimal merging schemes until no time periods can be merged, thereby determining the optimal number of time periods within the s-th period scenario using the elbow method. And divide the s-th cycle scenario into A slice of scene; Step 5: Following the process of Steps 2 to 4, obtain the slice scenes divided into m periodic scenes respectively, and use the DBSCAN clustering algorithm to cluster all slice scenes to obtain different scene clusters, and identify the typical slice scenes in each scene cluster.

2. The method for dividing distribution network planning scenarios based on the criterion of minimizing information entropy as described in claim 1, characterized in that, Step two is to proceed as follows: Step 2.1: Use equation (1) to calculate the j-th time period in the ith time period within the s-th cycle scenario. s Voltage sensitivity index of each distribution network node ; (1) In formula (1): For the i-th time period within the s-th cycle scenario, excluding the j-th time period s In addition to the load of each distribution network node, the j-th node obtained when calculating the power flow of the distribution network. s Voltage of each distribution network node; Step 2.2: Within the s-th cycle scenario, select the distribution network nodes whose voltage sensitivity index is closest to "1" for each time period. Each distribution network node is assigned a voltage-sensitive node for each time period, thus forming a set of voltage-sensitive nodes for each time period within the s-th cycle scenario. ; Step 2.3: Connect the data accessed in each time period within the s-th cycle scenario. Each new energy node and its corresponding time period By merging the voltage-sensitive nodes, we obtain the set of key nodes for each time period within the s-th cycle scenario. .

3. The method for dividing distribution network planning scenarios based on the criterion of minimizing information entropy as described in claim 2, characterized in that, Step three is performed as follows: Step 3.1: Use the deviation normalization method to analyze the voltage of the j-th critical node in the i-th time period within the s-th cycle scenario. Load output New energy power output Normalization is performed to obtain the normalized voltage of the j-th key node in the i-th time period within the s-th cycle scenario. Load output New energy power output ; Step 3.2: Calculate the voltage probability of the j-th critical node in the i-th time period within the s-th cycle scenario using equations (2), (3), and (4). Load probability and the probability of new energy output ; (2) (3) (4) In equations (2)-(4): Let n be the duration of the i-th time period. s The number of time periods in the s-th cycle scenario; Step 3.3: Calculate the voltage information entropy of the j-th key node in the i-th time period within the s-th cycle scenario using equations (5), (6), and (7). Load information entropy Information entropy of new energy output ; (5) (6) (7) Step 3.4: Calculate the comprehensive information entropy of the j-th key node in the i-th time period within the s-th cycle scenario using equation (8). ; (8) In equation (8): Let be the voltage information entropy weight of the j-th key node in the i-th time period within the s-th cycle scenario. Let be the load information entropy weight of the j-th key node in the i-th time period within the s-th cycle scenario. The entropy weight of the new energy output information of the j-th key node in the i-th time period within the s-th cycle scenario; Step 3.5: Calculate the global information entropy H of the s-th cycle scene using equation (9). s ; (9)。 4. The method for dividing distribution network planning scenarios based on the criterion of minimizing information entropy as described in claim 3, characterized in that, Step four is performed as follows: Step 4.1: Merge adjacent time periods within the s-th period scene to obtain... Various merger schemes, and The number of time periods for each scheme is ,and Calculate the global information entropy for each merging scheme using steps 3.2-3.5, denoted as . The scheme with the smallest reduction in information entropy is selected as the first optimal merging scheme, and the information entropy of the optimal merging scheme is denoted as . ,in, Indicates the first The global information entropy of the merging schemes; Step 4.2: Following the process in Step 4.1, analyze the first optimal merging scheme. Adjacent time periods within the same period are merged and calculated to obtain the number of time periods in the merging scheme. and the second optimal merging scheme ,and ; and thus obtain The optimal merging scheme and the number of time periods for each of the schemes ; Step 4.3, based on the number of time periods The horizontal axis represents the information entropy of the corresponding merging scheme. Using the vertical axis as the coordinate system, construct the coordinates corresponding to all solutions. Thus, using equation (10), a system is constructed with the starting coordinates (n) as the basis. s H s ) and endpoint coordinates ( , Baseline between: (10) Step 4.3: Calculate the coordinates of the y-th scheme using equation (11). , Distance to baseline Thus, the starting coordinates (n) are obtained. s H s ) and endpoint coordinates ( , The distance from the baseline to the coordinates of each scheme between ) The number of time periods corresponding to the scheme with the furthest distance is selected as the optimal number of time periods in the s-th cycle scenario, denoted as . : (11)。 5. The method for dividing distribution network planning scenarios based on the criterion of minimizing information entropy as described in claim 4, characterized in that, Step five is performed as follows: Step 5.1: Assume that the m-period scenes are divided into N slice scenes X1, X2, ..., X... N Calculate the dynamic time-warped distance between all slice scenes, let d(X) a X b ) represents the scene X of the a-th slice. a With the b-th slice scene X b The dynamic time-normalized distance between them; Step 5.2: Set the radius parameter Eps and the minimum number of neighbors MinPts for the DBSCAN clustering algorithm; Step 5.3, when X a When the number of slice scenes contained within a neighborhood with a radius of Eps is greater than or equal to MinPts, X will be... a Mark it as a core slice scene; Within any core slice scene x, if there are other slice scenes in the neighborhood with a radius of Eps, the corresponding slice scene will be merged into the same cluster as the core slice scene x. Within any core slice scene x, if there are multiple core slice scenes in the region with a radius of Eps, the range of the cluster corresponding to the core slice scene x is expanded until no new slice scenes are added to the cluster corresponding to x, thus forming a complete scene cluster of x. Step 5.4: Extract all core slice scenes from each scene cluster and output them as typical slice scenes.

6. An electronic device, comprising a memory and a processor, characterized in that, The memory is used to store a program that supports the processor in executing the distribution network planning scenario division method according to any one of claims 1-5, and the processor is configured to execute the program stored in the memory.

7. A computer-readable storage medium storing a computer program thereon, characterized in that, When the computer program is run by the processor, it executes the steps of the power distribution network planning scenario division method as described in any one of claims 1-5.

Images