Electrical simulation bus division and management method

Abstract

The invention relates to a method for dividing and managing an electrical simulation bus. The method comprises the steps of primary division of the simulation bus, secondary division of the simulation bus and peak daily load prediction of the simulation bus. According to the primary division, buses are aggregated into preliminary simulation buses through key section selection, GSDF analysis and GSDF hierarchical clustering. According to the secondary division, a final simulation bus is obtained through difference distance calculation, wheeling bus processing, minimum spanning tree calculation, wheeling bus reduction processing, relative compatibility judgment, absolute compatibility judgment and cardinal number screening. The peak daily load prediction is performed by calculating smooth values of related days and values of days of the same type. According to the invention, the operation safety, stability and prediction accuracy of the electrical system can be improved.

Description

Technical Field

[0001] The present invention relates to the field of electrical systems, and in particular to a method for dividing and managing an electrical simulation bus. Background Art

[0002] In electrical systems, busbar partitioning and management are crucial to system safety, stability, and operational efficiency. Traditional busbar partitioning methods are primarily based on the system's topology and operational experience, lacking scientific quantitative indicators and systematic methods. This results in inaccurate partitioning results that fail to fully reflect the electrical and load characteristics of the busbars, thus impacting system operation and prediction. Summary of the Invention

[0003] The purpose of the present invention is to provide a method for dividing and managing an electrical simulation busbar, so as to solve the problem in the prior art that the busbar division is not accurate enough and cannot fully reflect the busbar characteristics.

[0004] The technical solution of the present invention comprises the following steps:

[0005] Simulate busbar primary division:

[0006] Critical section selection: Select the critical sections in the electrical system that have tight safety margins and have a significant impact on the safety of the electrical system;

[0007] GSDF analysis: Calculate the generation transfer distribution factor (GSDF) of the busbar to the key section and obtain the GSDF matrix of the busbar to the key section;

[0008] GSDF hierarchical clustering: Hierarchical clustering method is used to aggregate buses with similar GSDF into preliminary simulated buses.

[0009] Simulate busbar secondary division:

[0010] Difference distance calculation: Calculate the difference distance between the load curves of the sub-busbars in the simulated busbar obtained by one division.

[0011] Processing of transfer busbar: For the planned transfer busbar, set its difference distance to zero;

[0012] Calculate the minimum spanning tree: use the minimum spanning tree algorithm to generate the minimum spanning tree of each primary partition simulation bus;

[0013] Restoration of transfer buses: restore the edge weights between transfer buses that have been set to zero to the original difference distance, and set the edges between transfer buses to be prohibited from deletion;

[0014] Relative compatibility judgment: perform relative compatibility judgment on the edge lengths of the minimum spanning tree and delete incompatible edges;

[0015] Absolute compatibility judgment: Perform absolute compatibility judgment on the remaining edges and delete the edges whose difference distance is greater than the absolute distance threshold;

[0016] Cardinality screening: Set the cardinality threshold value to classify islands with bus numbers greater than this value as the final simulated bus.

[0017] Simulate bus peak day load forecast;

[0018] For the final simulated busbar, calculate the smoothed values of the relevant days before the peak date of this year and the previous year moment by moment;

[0019] Based on the relationship between the value of the same type of day in the previous year and the smoothed value, the value of the day to be predicted in the current year is predicted.

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

[0021] Through scientific quantitative indicators and systematic methods, the accuracy and rationality of busbar division are improved;

[0022] The electrical characteristics and load characteristics between busbars are fully considered, which helps to improve the safety and stability of the electrical system;

[0023] Through peak day load forecasting, the accuracy of system operation and forecasting is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 This is the main execution process of a method for dividing and managing an electrical simulation bus in the present invention. DETAILED DESCRIPTION

[0025] 1. Simulate busbar primary division:

[0026] The first-stage busbar segmentation process includes the following three steps: selecting key sections from the electrical system; calculating the GSDF of the busbars for the key sections; and using the hierarchical clustering method to aggregate busbars with similar GSDFs into a preliminary busbar simulation. The details are as follows:

[0027] (1) Critical section selection. Select the critical sections K = [k1, k2, …, ku] from the electrical system that have a tight safety margin and have a significant impact on the safety of the electrical system. The critical sections can be obtained through actual experience in electrical system operation or through a critical section search algorithm. Through actual experience in electrical system operation, screen sections with a safety margin lower than a threshold (e.g., 15%); or use a critical section search algorithm (e.g., the N-1 verification method based on sensitivity analysis) to calculate the safety margin of each section and select u sections with the lowest margin.

[0028] (2) GSDF analysis. GSDF power generation transfer distribution factor, bus sensitivity to key sections, reflects the impact of bus power changes on section currents. GSDF of different bus pairs for key sections K, the GSDF matrix Sk of the bus pair for key sections is obtained:

[0029]

[0030] The element Sbku represents the GSDF of bus b to key line ku, Sbku = dPku / dPb, dPku is the power change of key section ku, and dPb is the power change of bus b.

[0031] (3) GSDF hierarchical clustering. For bus b, its GSDF vectors [Sbk1, Sbk2, …, Sbku] for u key lines define its position in the u-dimensional space. The clustering goal is to aggregate buses that are adjacent in the u-dimensional space to form a simulated bus.

[0032] Using Euclidean distance Dij as the distance measurement method, setting the inter-class distance calculation method to the farthest distance method, and setting the minimum inter-class distance threshold value, the hierarchical clustering method is used to cluster the busbars. After clustering is completed, VB preliminary simulated buses are obtained:

[0033]

[0034] Where 1 <= i <= b, 1 <= j <= b, and i ≠ j. Clustering stopping condition: Minimum inter-cluster distance threshold Dth = 0.5. Merging stops when the inter-cluster distance exceeds this value. Preliminary simulation busbar set VB = {VB1, VB2, …, VBm}.

[0035] 2. Simulate busbar secondary division:

[0036] The secondary partitioning stage includes 7 calculation steps, which are described as follows:

[0037] (1) Difference distance calculation. The goal of the difference distance calculation is to determine the similarity of the load curves between the sub-buses within a simulated bus obtained by a single division. The larger the difference distance, the greater the difference in the load curves of the two buses, and the less likely they should be divided into the same simulated bus.

[0038] The calculation steps are as follows:

[0039] For each primary division simulation bus given in the first stage, scan the topological connection relationship within each primary division simulation bus, and for two buses with branch associations, calculate the difference (distance) in the load curves of the two buses. The distance is defined as:

[0040] dij=1-|rij|.

[0041] rij is the linear correlation coefficient of the load curves of two buses i and j. Assume that the load curve period is T, which is typically 24 hours, and the time is the hourly level. The loads of buses i and j at the tth moment are yit and yjt respectively, then:

[0042]

[0043]

[0044]

[0045] rij measures the similarity between the load curves of buses i and j, while dij measures the difference between the load curves of buses i and j. The larger dij is, the greater the difference between the load curves of buses i and j. When dij is close to 1, the loads of buses i and j are almost completely uncorrelated; when dij is close to 0, the loads of buses i and j are almost completely correlated.

[0046] (2) Treatment of transfer buses. The planned transfer buses should be treated separately. The specific method is to directly set the difference distance between the two transfer buses to zero. Read the buses with transfer relationships from the transfer records of the bus loads, temporarily set the difference distance between the transfer buses to 0, and keep the original difference distance. The purpose of this is to prevent the buses with transfer relationships from being divided into two different simulation buses. In the subsequent minimum spanning tree, the edges corresponding to the transfer buses are marked as "must be retained".

[0047] (3) Calculate the minimum spanning tree. After calculating the difference distance, a topological graph can be obtained for each subdivided simulated bus, with its child buses as nodes and the difference distance as the edge weight between connected buses. At this point, the minimum spanning tree (MST) can be generated using the minimum spanning tree algorithm. As can be seen, since the difference distance between the transfer buses is 0, the minimum spanning tree must include the transfer bus, thus ensuring that the transfer bus is divided into the same simulated bus.

[0048] (4) Restoration of the transfer busbars. In order to avoid the possible impact of setting the difference distance between the transfer busbars to zero, the edge weights between the transfer busbars that were set to zero must first be restored to the original difference distance, and the edges between the transfer busbars must be set to prohibited deletion edges to prevent them from being deleted in the following steps.

[0049] Then, the relative compatibility criterion and the absolute compatibility criterion are used to determine whether the length of each edge in the spanning tree is significantly greater than the length of its surrounding edges or the lowest absolute distance threshold e. As long as one of the two criteria is met, the edge will be deleted.

[0050] (5) Relative compatibility judgment. For the length of the edge of the minimum spanning tree MST, calculate the average length of its adjacent edges ave_length. If

[0051] If length>2*ave_length, the edge is considered incompatible and deleted from the MST.

[0052] (6) Absolute compatibility judgment. After completing the relative compatibility judgment for all edges, perform absolute compatibility judgment on the remaining edges. If there are edges among the remaining edges whose difference distance is greater than the absolute distance threshold value e (meaning that its correlation coefficient is less than 1-e), they are deleted. e can be 0.7.

[0053] (7) Cardinality screening. After the absolute compatibility judgment is completed, several islands are formed in the MST, and the number of buses in each island varies. To prevent islands with very few buses from being classified as simulated buses, a certain cardinality threshold value MIN_NUM (for example, 5) is set. Islands with fewer buses than this value are merged into adjacent clusters, and islands with more buses than this cardinality threshold are classified as the final simulated buses.

[0054] 3. Simulate bus peak day load forecast:

[0055] For the final simulated bus, the peak day bus load forecast is performed one by one at each moment. Peak day refers to the day when the load in the electrical system reaches or approaches the historical peak level.

[0056] Assume that time t is calculated. Calculate the smoothed values A1t and A2t at time t of the relevant day before the peak day of this year and the previous year respectively. The relevant day refers to the historical date with similar load characteristics to the peak day to be predicted:

[0057]

[0058]

[0059] Where P1nt is the simulated bus load at time t on the nth day of the current year's peak period, selected from N (e.g., seven) relevant days. P2nt is the simulated bus load at time t on the nth day of the previous year's peak period. a is the smoothing coefficient for the point-by-point simulated bus load, generally taking a∈[0.1,0.9], with a recommended value of 0.3. C1 is the current year's base date, and C2 is the previous year's base date.

[0060] Take the value of the same type of day at time t in the previous year and set it as P2(C+F)t. Set the value of the day to be predicted this year at time t as P1(C+F)t. The same type of day refers to a historical day with the same date attribute (such as a working day) as the predicted day. F is the number of days offset between the predicted day and the base date. The following relationship is obtained:

[0061] P1(C1+F)t / A1t=P2(C2+F)t / A2t.

[0062] Therefore, the value at time t on the day to be predicted is:

[0063] P1(C1+F)t=(A1t / A2t)*P2(C2+F)t.

[0064] This process is repeated for each moment in turn to obtain the forecast curve for that day.

[0065] The present invention relates to a method for dividing and managing an electrical simulation bus, comprising primary simulation bus division, secondary simulation bus division, and simulation bus peak day load forecasting. The primary division aggregates the bus into a preliminary simulation bus through key section selection, GSDF analysis, and GSDF hierarchical clustering. The secondary division obtains the final simulation bus through difference distance calculation, transfer bus processing, minimum spanning tree calculation, transfer bus restoration processing, relative compatibility judgment, absolute compatibility judgment, and cardinality screening. Peak day load forecasting is performed by calculating the smoothed value of the relevant day and the value of the same type of day. The present invention can improve the safety, stability, and prediction accuracy of electrical system operation.

[0066] The above embodiments are merely exemplary embodiments of the present invention and are not intended to limit the scope of the present invention. The scope of protection of the present invention is defined by the claims. Those skilled in the art may make various modifications or equivalent substitutions to the present invention within the spirit and scope of protection of the present invention, and such modifications or equivalent substitutions shall also be deemed to fall within the scope of protection of the present invention.

Claims

1. A method for dividing and managing an electrical simulation bus, characterized in that: The following steps are involved: Simulate busbar primary division, including key section selection, GSDF analysis and GSDF hierarchical clustering; Simulate busbar secondary partitioning, including difference distance calculation, transfer busbar processing, calculation of minimum spanning tree, transfer busbar restoration processing, relative compatibility judgment, absolute compatibility judgment and cardinality screening; Simulate the bus peak day load forecast, including calculating the smoothed value of the relevant day and the value of the same type of day for forecasting.

2. The method according to claim 1, characterized in that The key section selection is obtained through actual operation experience of the electrical system or a key section search algorithm.

3. The method according to claim 1, characterized in that The GSDF analysis obtains the GSDF matrix Sk of the busbar to the key section, wherein the element Sbku represents the GSDF of the busbar b to the key line ku.

4. The method according to claim 1, wherein The GSDF hierarchical clustering adopts a hierarchical clustering method to cluster the busbars, and sets a minimum inter-class distance threshold Dth=0.

5.

5. The method according to claim 1, characterized in that The difference distance calculation is performed by calculating the difference distance dij of the load curves between the sub-busbars in the simulated busbar obtained by one division.

6. The method according to claim 1, characterized in that The minimum spanning tree calculation adopts a minimum spanning tree algorithm to generate a minimum spanning tree MST for each primary partitioned simulation bus.

7. The method according to claim 1, characterized in that The cardinality screening sets a cardinality threshold value MIN_NUM=5, and divides islands with bus numbers greater than this value into final simulated buses.

8. The method according to claim 1, characterized in that The relative compatibility judgment deletes edges with length>2*ave_length.

9. The method according to claim 8, characterized in that The absolute compatibility judgment deletes edges whose difference distance is greater than the absolute distance threshold value e (taken as 0.7).

Images