A feeder block division method and terminal based on complementary clustering of source-load characteristics

Abstract

The present invention discloses a feeder block partitioning method based on complementary clustering of source-load characteristics, including defining the maximum power supply radius, the minimum power supply radius and the center circle set of the substation, and based on this, adopting the initial cluster center setting based on the center circle set, the cluster center correction based on the center arc, and the weighted distance calculation between the source load and the cluster center to improve the K-means clustering algorithm, and then divide the feeder blocks, and then construct a feeder block partitioning evaluation index including the block balance rate and the block peak-valley difference rate based on the source-load characteristic matching analysis to judge the quality of the feeder blocks. The present invention can adapt to the current status of a large number of large-scale distributed power sources connected to the distribution network, and at the same time can fully tap the potential of source-load characteristic matching in the power supply area of ​​the substation to achieve efficient use of resources.

Description

Technical Field

[0001] The present invention relates to the technical field of distribution network planning, and in particular to a feeder block division method and a terminal based on source-load characteristic complementary clustering. Background Art

[0002] In recent years, as the global energy supply has transformed towards a clean and low-carbon direction, large-scale distributed power sources (DG) have been connected to the distribution network, which has had an important impact on the planning of the distribution network. In actual engineering applications, the planning of the distribution network includes many links, such as feeder block division, trunk line wiring, branch line wiring, and tie line wiring. As the primary link in the grid planning, feeder block division refers to dividing the power supply range of the substation into several power supply blocks. The quality of the division will affect the subsequent links of the grid planning and is of great significance.

[0003] Most traditional feeder block division methods are proposed by planners based on experience, which can easily lead to problems such as poor adaptability of the division scheme and low fault tolerance. The division of distribution network feeder blocks has achieved certain results so far. Some results have proposed a global optimization model of power supply partitioning and corresponding heuristic methods for medium-voltage target grid planning, so that planners of different levels can obtain basically consistent partition optimization division schemes. However, the application premise of such models is that there must be a candidate trunk channel layout, so the model is not very applicable and generalizable. Other results, based on the construction of a comprehensive evaluation index system for grid block division, proposed a grid optimization division mathematical model based on spatial clustering algorithm and its solution method, which laid the foundation for further promoting the engineering application of grid planning. However, the optimization objects of such models are the distribution transformers and cluster center position coordinates within the grid. The optimization objects are relatively special and the model is not very applicable. In addition, some results have proposed a feeder block division method that considers the complementarity of load timing characteristics, and constructed evaluation indicators for the block division scheme. The basic idea is to cluster loads that are conducive to improving the timing characteristics of feeder blocks. However, this method does not consider the impact of DG in the timing characteristics analysis, and there is still room for improvement. Summary of the invention

[0004] The technical problem to be solved by the present invention is to provide a feeder block division method and terminal based on source-load characteristic complementary clustering, which can adapt to the current status of distribution networks with a large number of large-scale distributed power sources connected and realize feeder block division.

[0005] In order to solve the above technical problems, the technical solution adopted by the present invention is:

[0006] A feeder block division method based on source-load characteristic complementary clustering comprises the steps of:

[0007] S1. Define the maximum power supply radius, minimum power supply radius and center circle set of the substation;

[0008] S2. Based on the maximum power supply radius, the minimum power supply radius and the center circle set, the K-means clustering algorithm is improved by using the initial cluster center setting based on the center circle set, the cluster center correction based on the center arc, and the weighted distance calculation between the source load and the cluster center, and the feeder block is obtained by using the improved K-means clustering algorithm;

[0009] S3. Based on the source-load characteristic matching analysis, a feeder block division evaluation index including a block balancing rate and a block peak-to-valley difference rate is constructed, and the quality of the feeder block is judged according to the feeder block division evaluation index.

[0010] In order to solve the above technical problems, another technical solution adopted by the present invention is:

[0011] A feeder block division terminal based on complementary clustering of source-load characteristics comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the following steps when executing the computer program:

[0012] S1. Define the maximum power supply radius, minimum power supply radius and center circle set of the substation;

[0013] S2. Based on the maximum power supply radius, the minimum power supply radius and the center circle set, the K-means clustering algorithm is improved by using the initial cluster center setting based on the center circle set, the cluster center correction based on the center arc, and the weighted distance calculation between the source load and the cluster center, and the feeder block is obtained by using the improved K-means clustering algorithm;

[0014] S3. Based on the source-load characteristic matching analysis, a feeder block division evaluation index including a block balancing rate and a block peak-to-valley difference rate is constructed, and the quality of the feeder block is judged according to the feeder block division evaluation index.

[0015] The beneficial effects of the present invention are as follows: the present invention provides a feeder block division method and terminal based on complementary clustering of source-load characteristics, by defining the maximum and minimum power supply radius and the center circle set, and on this basis, three improvement measures are proposed for the traditional K-means clustering algorithm, namely, initial cluster center setting based on the center circle set, cluster center correction based on the center arc, and weighted distance calculation between the source load and the cluster center, so as to divide the feeder blocks, and finally, based on the source-load characteristic matching analysis, a comprehensive evaluation index of the division effect including the block balance rate and the peak-valley difference rate is constructed to judge the advantages and disadvantages of the divided feeder blocks, so as to adapt to the current situation of a large number of large-scale distributed power sources connected to the distribution network, and at the same time, the potential of source-load characteristic matching in the power supply area of ​​the substation can be fully explored to achieve efficient utilization of resources. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 It is an overall flow chart of a feeder block division method based on source-load characteristic complementary clustering according to an embodiment of the present invention;

[0017] Figure 2 is a schematic diagram of the center circle set;

[0018] Figure 3 Schematic diagram for setting the initial cluster centers based on the center circle set;

[0019] Figure 4 This is a schematic diagram of cluster center correction;

[0020] Figure 5 Schematic diagram of improving K-means feeder block partitioning algorithm;

[0021] Figure 6 A geographic information map of an embodiment;

[0022] Figure 7 This is the block division result of the rotation center line distance weighted alternating positioning algorithm

[0023] Figure 8 The block partitioning result diagram for improving the K-means feeder block partitioning algorithm;

[0024] Fig. 9 The x-number DG partitioning result diagram is obtained by applying the rotation centerline distance weighted alternating positioning algorithm;

[0025] Fig.10 This is the result of DG partitioning number x using the improved K-means feeder block partitioning algorithm;

[0026] Fig.11 It is a structural schematic diagram of a feeder block division terminal based on source-load characteristic complementary clustering according to an embodiment of the present invention;

[0027] Description of labels:

[0028] 1. A feeder block division terminal based on complementary clustering of source-load characteristics; 2. A memory; 3. A processor. DETAILED DESCRIPTION

[0029] In order to explain the technical content, achieved objectives and effects of the present invention in detail, the following is an explanation in combination with the implementation modes and the accompanying drawings.

[0030] Please refer to Figures 1 to 10 , a feeder block partitioning method based on complementary clustering of source-load characteristics, comprising the steps of:

[0031] S1. Define the maximum power supply radius, minimum power supply radius and center circle set of the substation;

[0032] S2. Based on the maximum power supply radius, the minimum power supply radius and the center circle set, the K-means clustering algorithm is improved by using the initial cluster center setting based on the center circle set, the cluster center correction based on the center arc, and the weighted distance calculation between the source load and the cluster center, and the feeder block is obtained by using the improved K-means clustering algorithm;

[0033] S3. Based on the source-load characteristic matching analysis, a feeder block division evaluation index including a block balancing rate and a block peak-to-valley difference rate is constructed, and the quality of the feeder block is judged according to the feeder block division evaluation index.

[0034] From the above description, it can be seen that the beneficial effects of the present invention are: by defining the maximum and minimum power supply radius and the center circle set, and based on this, three improvement measures are proposed for the traditional K-means clustering algorithm, namely, the initial cluster center setting based on the center circle set, the cluster center correction based on the center arc, and the weighted distance calculation between the source load and the cluster center, so as to divide the feeder blocks, and finally, based on the source-load characteristic matching analysis, a comprehensive evaluation index of the division effect including the block balance rate and the peak-to-valley difference rate is constructed to judge the advantages and disadvantages of the divided feeder blocks, so as to adapt to the current situation of a large number of large-scale distributed power sources connected to the distribution network, and at the same time, it can fully tap the potential of source-load characteristic matching in the power supply area of ​​the substation to achieve efficient utilization of resources.

[0035] Furthermore, the step S1 is specifically as follows:

[0036] S11. Define the maximum power supply radius R as the Euclidean distance from the farthest load to the substation on the substation power supply boundary. max ;

[0037] S12. Define the minimum power supply radius R as the Euclidean distance from the nearest load to the substation on the substation power supply boundary. min ;

[0038] S13, define the center circle set Ω as the set of all center circles, each of which has the substation as the center, and the positions of all the center circles are constrained in a strip area, the bandwidth of the strip area is the maximum power supply radius R max and the minimum power supply radius R min The difference between

[0039] A polar coordinate system is established with the location of the substation as the pole, and the radius of the central ring of the strip area is the average polar diameter of all loads and large-scale distributed power sources within the power supply range of the substation.

[0040] From the above description, it can be seen that the work of feeder block division is established after the substation site selection and capacity determination. Therefore, the object to be solved is the power supply area of ​​the substation. Therefore, it is necessary to divide the power supply area of ​​the substation, that is, define the maximum power supply radius, minimum power supply radius and center circle set with the substation as the center, so as to subsequently improve the K-means clustering algorithm to optimize the feeder block division.

[0041] Furthermore, in step S2, the K-means clustering algorithm is improved by using the initial cluster center setting based on the central circle set, the cluster center correction based on the central arc, and the weighted distance calculation between the source load and the cluster center, specifically:

[0042] S21. Initial cluster center setting based on the central circle set:

[0043] At equal angles The points distributed on the central circle are used as the initial clustering centers of each feeder block, and are rotated clockwise around the substation center in turn. Obtaining the initial cluster centers of different groups;

[0044] S22. Cluster center correction based on central arc:

[0045] The initial cluster center is corrected along the direction of the central arc tangent of the central circle, and the correction formula is:

[0046] (1);

[0047] in, x 0 , y 0 are the substation coordinates, x i , y i are the coordinates of the initial cluster centers before correction, x i ’ , y i’ is the geometric center coordinate of all sources and loads in the feeder block after clustering, then the position of the final cluster center after correction should be the position of the initial cluster center rotated clockwise along the central arc trajectory of the central circle by an angle θ get;

[0048] S23. Calculation of weighted distance between source load and cluster center:

[0049] The weighted distance from the load and large-scale distributed power source to the final cluster center is the actual Euclidean distance multiplied by two weight factors, and the formula is as follows:

[0050] (2);

[0051] (3);

[0052] (4);

[0053] in, l ij and l ij ’ They represent the weighted distance and actual Euclidean distance between the load and the final cluster center, respectively. ω 1j , ω 2ij Indicates load i The two weight factors of P max ( j ) indicates feeder block j The peak load, q represents the magnification factor, ξ ij Indicates load i Add feeder block j The peak-to-valley difference of the subsequent blocks;

[0054] (5);

[0055] (6);

[0056] (7);

[0057] in, l jk and l jk ’ They represent the weighted distance and actual Euclidean distance between the large-scale distributed power source and the final cluster center, respectively. ω 1j DG , ω 2kj DG Large-scale distributed power generation k The two weight factors of ξ kj Large-scale distributed power generation k Add feeder block j The peak-to-valley difference of the subsequent blocks.

[0058] From the above description, it can be seen that the essence of the power supply range division of the substation is the spatial division with several points as the core, and the resulting area is quasi-circular. The feeder block division is to further divide the power supply range of the substation, which is essentially a block division with the power supply feeder as the core. Therefore, the feeder block shape is quasi-fan-shaped, and the fan-shaped structure is more in line with the actual planning needs of the project. However, when the K-means clustering algorithm is traditionally used for feeder block division, because the clustering objects of the feeder block division algorithm are loads and large-scale distributed power sources (DG), the substation is not taken into account, and the feeder block area distribution obtained by the division is random, which cannot meet the actual needs of the project with a fan-shaped structure. Therefore, the above three improvements to the K-means clustering algorithm are needed.

[0059] Furthermore, in step S2, the improved K-means clustering algorithm is used to divide the feeder blocks, specifically:

[0060] S24, based on the improvement measures of the initial cluster center setting of the central circle set, determine a set of the initial cluster centers, and set the weight factor ω 1j , ω 2ij , ω 1j DG and ω 2kj DG The initial value of is set to 1;

[0061] S25, calculating the weighted distances from all loads and large-scale distributed power sources to the initial cluster center according to formulas (2) and (5), clustering all loads and large-scale distributed power sources according to the proximity principle, and obtaining feeder blocks;

[0062] S26, based on the correction principle of the cluster center correction of the central arc, the coordinates of the initial cluster center of each feeder block are corrected according to formula (1) to obtain the final cluster center, and the weight factors of all loads and large-scale distributed power sources are updated according to formulas (3), (4) and formulas (6), (7);

[0063] S27, determine whether the displacement change of the center point of the final cluster center twice in all feeder blocks satisfies Δd < d bor If not, return to step S25, where Δ d represents the displacement change of the center point of the final cluster center between two adjacent times, d bor Indicates the preset minimum displacement critical value;

[0064] S28, judging whether all the initial cluster centers have been corrected, if so, the correction is completed, otherwise, returning to step S24.

[0065] From the above description, it can be seen that through the three improvement measures of initial cluster center setting based on the central circle set, cluster center correction based on the central arc, and weighted distance calculation between the source load and the cluster center, the traditional K-means clustering algorithm is improved to gradually correct the cluster center position of the feeder block to ensure that the divided feeder block has the optimal evaluation index, thereby ensuring the subsequent grid planning work.

[0066] Furthermore, in step S3, a feeder block division evaluation index including a block balance rate and a block peak-to-valley difference rate is constructed, specifically:

[0067] S31. With the goal of making the maximum net load power between feeder blocks close and the peak-to-valley difference of net load power within a feeder block smaller than the preset value, a block balancing rate index is proposed. And the block peak-to-valley difference rate indicator A fg :

[0068] (8);

[0069] (9);

[0070] in, P max ( j )and P min ( j ) represent feeder blocks respectively j The maximum and minimum values ​​of the timing power, Represents the mean of the sequential maximum power of all feeder blocks:

[0071] (10);

[0072] in, n grid represents the number of feeder blocks. From formula (10), we can see that if the time-series maximum power of each feeder block is more balanced, The larger the value, the smaller the timing peak-to-valley difference of each feeder block.A fg ( j ) is larger;

[0073] S32, the block balance rate index is calculated by a preset weight and the peak-to-valley difference rate index of the block A fg Perform weighted summation to obtain the final evaluation index of feeder block division F feeder :

[0074] (4);

[0075] in, α and β They are the block balance rate indicators and the peak-to-valley difference rate index of the block A fg The weight of α + β =1.

[0076] From the above description, it can be seen that the block balance rate index and the block peak-to-valley difference rate index are evaluated for the divided feeder blocks to ensure the relative uniqueness and optimality of the feeder block division, so as to determine a relatively optimal initial link for the subsequent grid planning and ensure the smooth progress of the subsequent grid planning work.

[0077] Please refer to Fig.11 , a feeder block division terminal based on source-load characteristic complementary clustering, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the following steps when executing the computer program:

[0078] S1. Define the maximum power supply radius, minimum power supply radius and center circle set of the substation;

[0079] S2. Based on the maximum power supply radius, the minimum power supply radius and the center circle set, the K-means clustering algorithm is improved by using the initial cluster center setting based on the center circle set, the cluster center correction based on the center arc, and the weighted distance calculation between the source load and the cluster center, and the feeder block is obtained by using the improved K-means clustering algorithm;

[0080] S3. Based on the source-load characteristic matching analysis, a feeder block division evaluation index including a block balancing rate and a block peak-to-valley difference rate is constructed, and the quality of the feeder block is judged according to the feeder block division evaluation index.

[0081] From the above description, it can be seen that the beneficial effects of the present invention are: based on the same technical concept, in conjunction with the above-mentioned feeder block division method based on complementary clustering of source-load characteristics, a feeder block division terminal based on complementary clustering of source-load characteristics is provided, by defining the maximum and minimum power supply radius and the center circle set, and on this basis, the traditional K-means clustering algorithm is proposed to set the initial cluster center based on the center circle set, correct the cluster center based on the center arc, and calculate the weighted distance between the source load and the cluster center. Three improvement measures are used to divide the feeder blocks, and finally, based on the source-load characteristic matching analysis, a comprehensive evaluation index of the division effect including the block balance rate and the peak-valley difference rate is constructed to judge the advantages and disadvantages of the divided feeder blocks, while adapting to the current situation of a large number of large-scale distributed power sources connected to the distribution network, it can fully tap the potential of source-load characteristic matching in the power supply area of ​​the substation and realize efficient utilization of resources.

[0082] Furthermore, the step S1 is specifically as follows:

[0083] S11. Define the maximum power supply radius R as the Euclidean distance from the farthest load to the substation on the substation power supply boundary. max ;

[0084] S12. Define the minimum power supply radius R as the Euclidean distance from the nearest load to the substation on the substation power supply boundary. min ;

[0085] S13, define the center circle set Ω as the set of all center circles, each of which has the substation as the center, and the positions of all the center circles are constrained in a strip area, the bandwidth of the strip area is the maximum power supply radius R max and the minimum power supply radius R min The difference between

[0086] A polar coordinate system is established with the location of the substation as the pole, and the radius of the central ring of the strip area is the average polar diameter of all loads and large-scale distributed power sources within the power supply range of the substation.

[0087] From the above description, it can be seen that the work of feeder block division is established after the substation site selection and capacity determination. Therefore, the object to be solved is the power supply area of ​​the substation. Therefore, it is necessary to divide the power supply area of ​​the substation, that is, define the maximum power supply radius, minimum power supply radius and center circle set with the substation as the center, so as to subsequently improve the K-means clustering algorithm to optimize the feeder block division.

[0088] Furthermore, in step S2, the K-means clustering algorithm is improved by using the initial cluster center setting based on the central circle set, the cluster center correction based on the central arc, and the weighted distance calculation between the source load and the cluster center, specifically:

[0089] S21. Initial cluster center setting based on the central circle set:

[0090] At equal angles The points distributed on the central circle are used as the initial clustering centers of each feeder block, and are rotated clockwise around the substation center in turn. Obtaining the initial cluster centers of different groups;

[0091] S22. Cluster center correction based on central arc:

[0092] The initial cluster center is corrected along the direction of the central arc tangent of the central circle, and the correction formula is:

[0093] (1);

[0094] in, x 0 , y 0 are the substation coordinates, x i , y i are the coordinates of the initial cluster centers before correction, x i ’ , y i ’ is the geometric center coordinate of all sources and loads in the feeder block after clustering, then the position of the final cluster center after correction should be the position of the initial cluster center rotated clockwise along the central arc trajectory of the central circle by an angle θ get;

[0095] S23. Calculation of weighted distance between source load and cluster center:

[0096] The weighted distance from the load and large-scale distributed power source to the final cluster center is the actual Euclidean distance multiplied by two weight factors, and the formula is as follows:

[0097] (2);

[0098] (3);

[0099] (4);

[0100] in, l ij and l ij ’ They represent the weighted distance and actual Euclidean distance between the load and the final cluster center, respectively. ω1j , ω 2ij Indicates load i The two weight factors of P max ( j ) indicates feeder block j The peak load, q represents the magnification factor, ξ ij Indicates load i Add feeder block j The peak-to-valley difference of the subsequent blocks;

[0101] (5);

[0102] (6);

[0103] (7);

[0104] in, l jk and l jk ’ They represent the weighted distance and actual Euclidean distance between the large-scale distributed power source and the final cluster center, respectively. ω 1j DG , ω 2kj DG Large-scale distributed power generation k The two weight factors of ξ kj Large-scale distributed power generation k Add feeder block j The peak-to-valley difference of the subsequent blocks.

[0105] From the above description, it can be seen that the essence of the power supply range division of the substation is the spatial division with several points as the core, and the resulting area is quasi-circular. The feeder block division is to further divide the power supply range of the substation, which is essentially a block division with the power supply feeder as the core. Therefore, the feeder block shape is quasi-fan-shaped, and the fan-shaped structure is more in line with the actual planning needs of the project. However, when the K-means clustering algorithm is traditionally used for feeder block division, because the clustering objects of the feeder block division algorithm are loads and large-scale distributed power sources (DG), the substation is not taken into account, and the feeder block area distribution obtained by the division is random, which cannot meet the actual needs of the project with a fan-shaped structure. Therefore, the above three improvements to the K-means clustering algorithm are needed.

[0106] Furthermore, in step S2, the improved K-means clustering algorithm is used to divide the feeder blocks, specifically:

[0107] S24, based on the improvement measures of the initial cluster center setting of the central circle set, determine a set of the initial cluster centers, and set the weight factor ω 1j , ω 2ij , ω 1j DG and ω 2kj DG The initial value of is set to 1;

[0108] S25, calculating the weighted distances from all loads and large-scale distributed power sources to the initial cluster center according to formulas (2) and (5), clustering all loads and large-scale distributed power sources according to the proximity principle, and obtaining feeder blocks;

[0109] S26, based on the correction principle of the cluster center correction of the central arc, the coordinates of the initial cluster center of each feeder block are corrected according to formula (1) to obtain the final cluster center, and the weight factors of all loads and large-scale distributed power sources are updated according to formulas (3), (4) and formulas (6), (7);

[0110] S27, determine whether the displacement change of the center point of the final cluster center twice in all feeder blocks satisfies Δ d < d bor If not, return to step S25, where Δ d represents the displacement change of the center point of the final cluster center between two adjacent times, d bor Indicates the preset minimum displacement critical value;

[0111] S28, judging whether all the initial cluster centers have been corrected, if so, the correction is completed, otherwise, returning to step S24.

[0112] From the above description, it can be seen that through the three improvement measures of initial cluster center setting based on the central circle set, cluster center correction based on the central arc, and weighted distance calculation between the source load and the cluster center, the traditional K-means clustering algorithm is improved to gradually correct the cluster center position of the feeder block to ensure that the divided feeder block has the optimal evaluation index, thereby ensuring the subsequent grid planning work.

[0113] Furthermore, in step S3, a feeder block division evaluation index including a block balance rate and a block peak-to-valley difference rate is constructed, specifically:

[0114] S31. With the goal of making the maximum net load power between feeder blocks close and the peak-to-valley difference of net load power within a feeder block smaller than the preset value, a block balancing rate index is proposed. And the block peak-to-valley difference rate indicator A fg :

[0115] (8);

[0116] (9);

[0117] in, P max ( j )and P min ( j ) represent feeder blocks respectively j The maximum and minimum values ​​of the timing power, Represents the mean of the sequential maximum power of all feeder blocks:

[0118] (10);

[0119] in, n grid represents the number of feeder blocks. From formula (10), we can see that if the time-series maximum power of each feeder block is more balanced, The larger the value, the smaller the timing peak-to-valley difference of each feeder block. A fg ( j ) is larger;

[0120] S32, the block balance rate index is calculated by a preset weight and the peak-to-valley difference rate index of the block A fg Perform weighted summation to obtain the final evaluation index of feeder block division F feeder :

[0121] (4);

[0122] in, α and β They are the block balance rate indicators and the peak-to-valley difference rate index of the block A fg The weight of α + β =1.

[0123] From the above description, it can be seen that the block balance rate index and the block peak-to-valley difference rate index are evaluated for the divided feeder blocks to ensure the relative uniqueness and optimality of the feeder block division, so as to determine a relatively optimal initial link for the subsequent grid planning and ensure the smooth progress of the subsequent grid planning work.

[0124] The present invention provides a feeder block division method and terminal based on source-load characteristic complementary clustering, which is used to cope with the current situation of large-scale distributed power sources connected to the distribution network. The method is described in detail below in conjunction with an embodiment.

[0125] Please refer to Figure 1 , Embodiment 1 of the present invention is:

[0126] A feeder block partitioning method based on complementary clustering of source-load characteristics, such as Figure 1 As shown, the steps include:

[0127] S1. Define the maximum power supply radius, minimum power supply radius and center circle set of the substation.

[0128] S2. Based on the maximum power supply radius, the minimum power supply radius and the center circle set, the K-means clustering algorithm is improved by setting the initial cluster center based on the center circle set, correcting the cluster center based on the center arc, and calculating the weighted distance between the source load and the cluster center. The feeder blocks are divided using the improved K-means clustering algorithm.

[0129] S3. Based on the source-load characteristic matching analysis, a feeder block division evaluation index including block balancing rate and block peak-to-valley difference rate is constructed, and the advantages and disadvantages of the feeder blocks are judged according to the feeder block division evaluation index.

[0130] That is, in this embodiment, by defining the maximum and minimum power supply radius and the center circle set, and based on this, three improvement measures are proposed for the traditional K-means clustering algorithm, including the initial cluster center setting based on the center circle set, the cluster center correction based on the center arc, and the weighted distance calculation between the source load and the cluster center, so as to divide the feeder blocks. Finally, based on the source-load characteristic matching analysis, a comprehensive evaluation index of the division effect including the block balance rate and the peak-to-valley difference rate is constructed to judge the advantages and disadvantages of the divided feeder blocks. While adapting to the current situation of a large number of large-scale distributed power sources connected to the distribution network, it can also fully tap the potential of source-load characteristic matching in the power supply area of ​​the substation and realize efficient utilization of resources.

[0131] Please refer to Figures 2 to 5 , Embodiment 2 of the present invention is:

[0132] A feeder block division method based on complementary clustering of source-load characteristics. On the basis of the above-mentioned embodiment 1, in this embodiment, since the work of feeder block division is established after the substation site selection and capacity determination, the object to be solved is the substation power supply area. Therefore, it is necessary to divide the power supply area of ​​the substation, that is, define the maximum power supply radius, minimum power supply radius and center circle set with the substation as the center, so as to subsequently improve the K-means clustering algorithm to optimize the feeder block division.

[0133] Therefore, in this embodiment, step S1 specifically includes the following steps:

[0134] S11. Define the maximum power supply radius R as the Euclidean distance from the farthest load to the substation on the substation power supply boundary. max .

[0135] S12. Define the minimum power supply radius R as the Euclidean distance from the nearest load to the substation on the substation power supply boundary. min .

[0136] S13. Define the center circle set Ω as the set of all center circles. Each center circle is centered on the substation, and the positions of all center circles are constrained in a strip area. The bandwidth of the strip area is the maximum power supply radius R max and the minimum power supply radius R min The difference between

[0137] A polar coordinate system is established with the location of the substation as the pole. The radius of the central ring of the strip area is the average polar diameter of all loads and large-scale distributed power sources within the power supply range of the substation. In this embodiment, the schematic diagram of the central circle set can be as follows: Figure 2 shown.

[0138] Since the power supply range division of the substation is essentially a spatial division with several points as the core, the resulting area is quasi-circular. The feeder block division is to further divide the power supply range of the substation, which is essentially a block division with the power supply feeder as the core. Therefore, the feeder block is quasi-sector-shaped, and the quasi-sector-shaped structure is more in line with the actual planning needs of the project.

[0139] However, when the K-means clustering algorithm is traditionally used to divide feeder blocks, the feeder block division algorithm clusters loads and large-scale distributed generation (DG) without taking into account substations. As a result, the distribution of the divided feeder blocks is random and cannot meet the actual needs of fan-shaped projects. Therefore, the following three improvements need to be made to the K-means clustering algorithm:

[0140] S21. Initial cluster center setting based on the central circle set:

[0141] like Figure 3As shown, at equal angles The points distributed on the central circle are used as the initial clustering centers of each feeder block, and rotate clockwise around the substation center in turn. Get different groups of initial cluster centers.

[0142] S22. Cluster center correction based on central arc:

[0143] like Figure 4 As shown, the initial cluster center is corrected along the direction of the central arc tangent of the central circle. The correction formula is:

[0144] (1).

[0145] In the above formula, x 0 , y 0 are the substation coordinates, x i , y i are the coordinates of the initial cluster centers before correction, x i ’ , y i ’ is the geometric center coordinate of all sources and loads in the feeder block after clustering. The corrected final cluster center position should be the initial cluster center position rotated clockwise along the central arc trajectory of the central circle by an angle θ get.

[0146] S23. Calculation of weighted distance between source load and cluster center:

[0147] The weighted distance from the load and large-scale distributed generation to the final cluster center is the actual Euclidean distance multiplied by two weight factors, and the formula is as follows:

[0148] (2);

[0149] (3);

[0150] (4).

[0151] In the above formula, l ij and l ij ’ Respectively represent the weighted distance and actual Euclidean distance between the load and the final cluster center, ω 1j , ω 2ij Indicates loadi The two weight factors of P max ( j ) indicates feeder block j The peak load, q represents the magnification factor, ξ ij Indicates load i Add feeder block j The more balanced the net load of each feeder block and the smaller the peak-to-valley difference within the feeder block, the smaller the weighted distance.

[0152] (5);

[0153] (6);

[0154] (7).

[0155] In the above formula, l jk and l jk ’ They represent the weighted distance and actual Euclidean distance between large-scale distributed power and the final cluster center, ω 1j DG , ω 2kj DG Large-scale distributed power generation k The two weight factors of ξ kj Large-scale distributed power generation k Add feeder block j The more balanced the net load of each feeder block and the smaller the peak-to-valley difference within the feeder block, the smaller the weighted distance.

[0156] Based on the three improvement measures proposed above, in this embodiment, Figure 5 As shown, in step S2, the improved K-means clustering algorithm is used to divide the feeder blocks. The specific process is as follows:

[0157] S24, based on the improvement measures of the initial cluster center setting of the central circle set, determine a set of initial cluster centers and set the weight factor ω 1j , ω 2ij , ω 1j DG and ω 2kj DG The initial value of is set to 1.

[0158] S25. Calculate the weighted distances from all loads and large-scale distributed generation to the initial cluster center according to formulas (2) and (5), and cluster all loads and large-scale distributed generation according to the proximity principle to obtain feeder blocks.

[0159] S26. Based on the correction principle of cluster center correction of the central arc, the coordinates of the initial cluster center of each feeder block are corrected according to formula (1) to obtain the final cluster center, and the weight factors of all loads and large-scale distributed generation are updated according to formulas (3), (4) and (6), (7).

[0160] S27, determine whether the displacement change of the center point of the final cluster center of all feeder blocks twice adjacent to each other satisfies Δ d < d bor If not, return to step S25, where Δ d represents the displacement change of the center point of the final cluster center between two adjacent times, d bor Indicates the preset minimum displacement critical value.

[0161] S28, judging whether all initial cluster centers have been corrected, if so, the correction is completed, otherwise, returning to step S24.

[0162] That is, through three improvement measures, namely, setting the initial cluster center based on the central circle set, correcting the cluster center based on the central arc, and calculating the weighted distance between the source load and the cluster center, the traditional K-means clustering algorithm is improved to gradually correct the cluster center position of the feeder block to ensure that the divided feeder block has the optimal evaluation index, thereby ensuring the subsequent grid planning work.

[0163] In addition, since evaluation indicators are required to judge the pros and cons of the feeder block division scheme, usually all loads in a feeder block are independently powered by one feeder. Since the feeder models and capacities used are the same, it is hoped that the maximum net load power between blocks is close. At the same time, in order to reduce the power supply pressure at high load times, it is also hoped that the peak-to-valley difference of the net load power within the feeder block is small.

[0164] Based on the above feeder block division purpose, in this embodiment, in step S3, a feeder block division evaluation index including a block balance rate and a block peak-to-valley difference rate is constructed, specifically:

[0165] S31. With the goal of making the maximum net load power between feeder blocks close and the peak-to-valley difference of net load power within a feeder block smaller than the preset value, a block balancing rate index is proposed. And the block peak-to-valley difference rate indicator A fg:

[0166] (8);

[0167] (9).

[0168] In the above formula, P max ( j )and P min ( j ) represent feeder blocks respectively j The maximum and minimum values ​​of the timing power, Represents the mean of the sequential maximum power of all feeder blocks:

[0169] (10).

[0170] In the above formula, n grid represents the number of feeder blocks. From formula (10), we can see that if the time-series maximum power of each feeder block is more balanced, The larger the value, the smaller the timing peak-to-valley difference of each feeder block. A fg ( j ) is larger.

[0171] S32, using the preset weight to balance the block rate index And the block peak-to-valley difference rate indicator A fg Perform weighted summation to obtain the final evaluation index of feeder block division F feeder :

[0172] (4).

[0173] In the above formula, α and β They are block balance rate indicators And the block peak-to-valley difference rate indicator A fg The weight of α + β =1.

[0174] That is, the block balance rate index and block peak-to-valley difference rate index are evaluated for the divided feeder blocks to ensure the relative uniqueness and optimality of the feeder block division, so as to determine a relatively optimal initial link for the subsequent grid planning and ensure the smooth progress of the subsequent grid planning work.

[0175] Please refer to Figures 6 to 10 , Embodiment 3 of the present invention is:

[0176] A feeder block division method based on complementary clustering of source-load characteristics is provided. Based on the above-mentioned second embodiment, in this embodiment, an example is proposed to verify the effectiveness and applicability of the method.

[0177] The specific process is as follows:

[0178] First, assume that this example includes a power grid formed by three 2×63MVA substations, with 160, 156, and 150 loads and DGs in each of the three substations. There are four types of loads: residential, commercial, industrial, and administrative, and the DG is photovoltaic. The geographical distribution of the source and load is as follows: Figure 6 As shown, the star represents DG and the other symbols represent various types of loads.

[0179] (1) Considering the influence of complementary source-load characteristics

[0180] In order to analyze the influence of source-load characteristics complementarity in feeder block division, two strategies are adopted to divide the planning area into blocks. The specific strategies are as follows:

[0181] Strategy 1: Ignoring the complementary characteristics of source and load, and considering the block balance rate without considering the block peak-to-valley difference rate, that is, using the weighting factor in the calculation of the weighted distance ω 1j Instead of using ω 2ij ,The improved K-means clustering algorithm is applied to realize feeder block division.

[0182] Strategy 2: Consider the complementary characteristics of source and load, and comprehensively consider the block balance rate and the block peak-to-valley difference rate, that is, use the weighting factor in the calculation of weighted distance ω 1j and ω 2ij ,The improved K-means clustering algorithm is applied to realize feeder block division.

[0183] Through the block number and block balance rate indicators under the two strategies A av And the block peak-to-valley difference rate indicator A fg The results of the three indicators are compared and the specific results are shown in Table 1 below:

[0184] Table 1 Results of feeder block division indicators for each substation under strategy 1 and strategy 2

[0185]

[0186] Analysis of the data in Table 1 shows that the number of feeder blocks obtained by substation division under the source-load characteristic complementary strategy is small, and the block balancing rate and block peak-to-valley difference rate indicators of each substation are high.

[0187] Therefore, the strategy of considering the complementary characteristics of source and load can reduce the number of divided feeder blocks and reduce the economic cost of planning; on the other hand, it can make the balance rate index of each block A av And the block peak-to-valley difference rate indicator A fg Higher, which is beneficial to improve the utilization rate of feeder equipment.

[0188] (2) Effectiveness analysis of improved K-means clustering algorithm

[0189] In order to verify the effectiveness of the feeder block division method based on the improved K-means clustering algorithm proposed in the present invention, the algorithm is compared with the block division effect of the rotation centerline distance weighted alternating positioning algorithm proposed by Ge Shaoyun in Practical automatic wiring of medium-voltage distribution network considering load characteristic complementarity and power supply unit division (Proceedings of the CSEE, 2020, 40(03):790-803.DOI:10.13334 / j.0258-8013.pcsee.181013.), and strategy 3 is set as follows:

[0190] Strategy 3: Consider the complementary characteristics of source and load, and comprehensively consider the block balance rate and the block peak-to-valley difference rate, that is, use the weighting factor in the calculation of weighted distance ω 1j and ω 2ij , the feeder block division is realized by applying the rotation centerline distance weighted alternating positioning algorithm.

[0191] Both strategies consider the influence of complementary source-load characteristics. The only difference in controlling the two strategies is the different block division algorithms. By comparing the feeder block division effects under Strategy 2 and Strategy 3, the effectiveness of the algorithm proposed in the present invention is verified.

[0192] Figure 7 and Figure 8 Feeder block division results achieved by using strategies 2 and 3 respectively. In the figure, the loads and DGs belonging to the same feeder block are marked with the same color. Figure 7 It can be seen that the feeder block division result achieved by applying the algorithm proposed by Ge Shaoyun (Proceedings of the CSEE, 2020, 40(03):790-803.DOI:10.13334 / j.0258-8013.pcsee.181013.) presents a fan-shaped structure, which meets the actual needs of the project. Adjacent feeder blocks can be strictly separated by a straight line segment, so the source and load near the substation are all within the angle of the fan-shaped structure. Analysis Figure 8It can be seen that the feeder block division result realized by the algorithm proposed in the present invention still has a fan-shaped structure distribution, but compared with Figure 7 The block shapes are more varied, and the boundaries of two adjacent feeder blocks cannot be separated by a straight line segment.

[0193] Therefore, in order to further demonstrate the superiority of the feeder block partitioning algorithm proposed in the present invention, Fig. 9 Taking the division and ownership of DG No. x in the power supply area of ​​substation S2 as an example, the analysis is carried out. Fig. 9 This is the result diagram of DG partitioning of x using the rotation centerline distance weighted alternating positioning algorithm. Fig.10 This is the result of DG partitioning number x using the improved K-means clustering feeder block partitioning algorithm.

[0194] Fig. 9 and Fig.10 The analysis is carried out around the four feeder blocks A, B, C and D. The dotted line drawn from the substation in the feeder block represents the center line ray proposed by Ge Shaoyun in the practical automatic wiring of medium-voltage distribution network considering load characteristic complementarity and power supply unit division (Proceedings of the CSEE, 2020, 40(03):790-803.DOI:10.13334 / j.0258-8013.pcsee.181013.). DG No. x is a distributed power source bordering the S2 substation.

[0195] analyze Fig. 9 It can be seen that the feeder block division principle of the algorithm proposed by Ge Shaoyun in Practical automatic wiring of medium-voltage distribution network considering load characteristic complementarity and power supply unit division (Proceedings of the CSEE, 2020, 40(03):790-803.DOI:10.13334 / j.0258-8013.pcsee.181013.) is to use the N rays derived from the substation as the center line, and to achieve clustering by calculating the weighted distance from the source load to each center line. The center line is a radial ray cluster derived from the substation. Therefore, DG No. x, which borders S2, is located in an area with dense center lines and almost coincides with the trajectory of the center line of feeder block C. In this case, when calculating the weighted distance from the DG to the center line, the marginal effect of the weighting factor will be severely weakened, resulting in DG No. x being unreasonably divided into feeder block C. In fact, the source load division in this case is not conducive to improving the comprehensive indicators of the feeder block.

[0196] Fig.10 This is the feeder block division result achieved by the algorithm of the present invention. The algorithm takes the source load cluster center as the benchmark, and clusters the source load according to the weighted distance to the cluster center to achieve feeder block division. Because the division algorithm does not have a dense cluster center area near the substation, it solves the problem of unreasonable source load division in this area. Fig.10 In the figure, DG x is almost coincident with the centerline trajectory of feeder block A, but this DG is divided into feeder block B. This verifies the conclusion that the algorithm proposed in the present invention is only aimed at improving the feeder block index and is more flexible for the source-load division near the substation.

[0197] This example takes a DG in the substation S2 area as an example to analyze the difference in the division effect of the two algorithms in the area near the substation. In order to more intuitively compare the advantages and disadvantages of the two algorithms, the results of the indicators of each substation under the two algorithms are listed in Table 2 as follows:

[0198] Table 2 Results of feeder block division indicators for each substation under strategy 2 and measurement 3

[0199]

[0200] Table 2 A av and A fg They represent the block balance rate index and the block peak-to-valley difference rate index respectively. The larger the value, the better the block division effect. Analysis of the data in Table 2 shows that when the feeder block division is performed on the power supply area of ​​the substation, strategy 2 can obtain a better feeder block division scheme than strategy 3, which is specifically reflected in the two indicators of each substation being better. Therefore, the feeder block division algorithm based on improved K-means clustering has a better block division effect than the rotation centerline distance weighted alternating positioning algorithm, thereby verifying the effectiveness and advancement of the algorithm proposed in the present invention.

[0201] That is, the example analysis of this embodiment shows that the feeder block division method considering the complementary clustering of source-load characteristics proposed in the present invention, on the one hand, reduces the number of divided feeder blocks and reduces the economic cost of planning; on the other hand, it makes the balance rate index of each feeder block A av And the block peak-to-valley difference rate indicator A fg At the same time, this algorithm and the rotation centerline weighted distance positioning algorithm are used to realize feeder block division in the same power supply area. The feeder block index results show that this algorithm has a better division effect, thus verifying the effectiveness and progress of the improved K-means feeder block division algorithm proposed in this invention.

[0202] Please refer to Fig.11 , Embodiment 4 of the present invention is:

[0203] A feeder block division terminal 1 based on complementary clustering of source-load characteristics includes a memory 2, a processor 3, and a computer program stored in the memory 2 and executable on the processor 3. In this embodiment, when the processor 3 executes the computer program, the steps of any one of the above-mentioned embodiments one to three are implemented.

[0204] In summary, the present invention provides a feeder block division method and terminal based on complementary clustering of source-load characteristics, the purpose of which is to cope with the current situation of large-scale distributed power sources connected to the distribution network, and to establish a feeder block division method that considers the complementary source-load characteristics. The work of feeder block division is established after the site selection and capacity determination of the substation, so the solution object is the power supply area of ​​the substation, and the output of the distributed power source is comprehensively considered. At the same time, the complementary clustering of the source-load timing characteristics is considered, and the feeder block division is performed with the optimal feeder block division evaluation index as the goal. Because the next step of feeder block division is grid planning, the relative uniqueness and optimality of feeder block division make it the key problem to be solved first in medium-voltage target grid planning. Therefore, the quality of feeder block division directly determines the development of grid planning work, so the present invention can determine a relatively optimal initial link for the development of grid planning.

[0205] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent transformations made using the contents of the present invention's specification and drawings, or directly or indirectly applied in related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A feeder block division method based on complementary clustering of source-load characteristics, It is characterized in that Includes steps: S1. Define the maximum power supply radius, minimum power supply radius and center circle set of the substation; S2. Based on the maximum power supply radius, the minimum power supply radius and the center circle set, the K-means clustering algorithm is improved by using the initial cluster center setting based on the center circle set, the cluster center correction based on the center arc, and the weighted distance calculation between the source load and the cluster center. The feeder blocks are divided using the improved K-means clustering algorithm. S3. Based on the source-load characteristic matching analysis, a feeder block division evaluation index including block balancing rate and block peak-to-valley difference rate is constructed, and the quality of the feeder block is judged according to the feeder block division evaluation index; Step S1 is specifically as follows: S11. Define the maximum power supply radius R as the Euclidean distance from the farthest load to the substation on the substation power supply boundary. max ; S12. Define the minimum power supply radius R as the Euclidean distance from the nearest load to the substation on the substation power supply boundary. min ; S13. Define the center circle set Ω as the set of all center circles. Each center circle is centered on the substation, and the positions of all center circles are constrained in a strip area. The bandwidth of the strip area is the maximum power supply radius R max and the minimum power supply radius R min The difference between A polar coordinate system is established with the location of the substation as the pole, and the radius of the central ring of the strip area is the average polar diameter of all loads and large-scale distributed power sources within the power supply range of the substation; In step S2, the K-means clustering algorithm is improved as follows: S21. Initial cluster center setting based on the central circle set: At equal angles The points distributed on the central circle are used as the initial clustering centers of each feeder block, and rotate clockwise around the substation center in turn. Get the initial cluster centers of different groups; S22. Cluster center correction based on central arc: The initial cluster center is corrected along the direction of the central arc tangent of the central circle. The correction formula is: （1）； in, x 0 , y 0 are the substation coordinates, x i , y i are the coordinates of the initial cluster centers before correction, x i ’ , y i ’ is the geometric center coordinate of all sources and loads in the feeder block after clustering. The corrected final cluster center position should be the initial cluster center position rotated clockwise along the central arc trajectory of the central circle by an angle θ get; S23. Calculation of weighted distance between source load and cluster center: The weighted distance from the load and large-scale distributed generation to the final cluster center is the actual Euclidean distance multiplied by two weight factors, and the formula is as follows: （2）； （3）； （4）； in, l ij and l ij ’ Respectively represent the weighted distance and actual Euclidean distance between the load and the final cluster center, ω 1j , ω 2ij Indicates load i The two weight factors of P max ( j ) indicates feeder block j The peak load, q represents the magnification factor, ξ ij Indicates load i Add feeder block j The peak-to-valley difference of the subsequent blocks; （5）； （6）； （7）； in, l jk and l jk ’ They represent the weighted distance and actual Euclidean distance between large-scale distributed power and the final cluster center, ω 1j DG , ω 2kj DG Large-scale distributed power generation k The two weight factors of ξ kj Large-scale distributed power generation k Add feeder block j The peak-to-valley difference of the subsequent blocks.

2. According to claim 1, a feeder block division method based on source-load characteristic complementary clustering, It is characterized in that In step S2, the improved K-means clustering algorithm is used to divide the feeder blocks, specifically: S24, based on the improvement measures of the initial cluster center setting of the central circle set, determine a set of the initial cluster centers, and set the weight factor ω 1j , ω 2ij , ω 1j DG and ω 2kj DG The initial value of is set to 1; S25, calculating the weighted distances from all loads and large-scale distributed power sources to the initial cluster center according to formulas (2) and (5), clustering all loads and large-scale distributed power sources according to the proximity principle, and obtaining feeder blocks; S26, based on the correction principle of the cluster center correction of the central arc, the coordinates of the initial cluster center of each feeder block are corrected according to formula (1) to obtain the final cluster center, and the weight factors of all loads and large-scale distributed power sources are updated according to formulas (3), (4) and formulas (6), (7); S27, determine whether the displacement change of the center point of the final cluster center twice in all feeder blocks satisfies Δ d < d bor If not, return to step S25, where Δ d represents the displacement change of the center point of the final cluster center between two adjacent times, d bor Indicates the preset minimum displacement critical value; S28, judging whether all the initial cluster centers have been corrected, if so, the correction is completed, otherwise, returning to step S24.

3. According to claim 2, a feeder block division method based on source-load characteristic complementary clustering, It is characterized in that In step S3, a feeder block division evaluation index including a block balance rate and a block peak-to-valley difference rate is constructed, specifically: S31. With the goal of making the maximum net load power between feeder blocks close and the peak-to-valley difference of net load power within a feeder block smaller than the preset value, a block balancing rate index is proposed. And the block peak-to-valley difference rate indicator A fg : （8）； （9）； in, P max ( j )and P min ( j ) represent feeder blocks respectively j The maximum and minimum values ​​of the timing power, Represents the mean of the sequential maximum power of all feeder blocks: （10）； in, n grid represents the number of feeder blocks. From formula (10), we can see that if the maximum power of each feeder block is more balanced, The larger the value, the smaller the timing peak-to-valley difference of each feeder block. A fg ( j ) is larger; S32, the block balance rate index is calculated by a preset weight and the peak-to-valley difference rate index of the block A fg Perform weighted summation to obtain the final evaluation index of feeder block division F feeder : ； in, α and β They are the block balance rate indicators and the peak-to-valley difference rate index of the block A fg The weight of α + β =1.

4. A feeder block division terminal based on complementary clustering of source-load characteristics, It is characterized in that It comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps in a feeder block division method based on complementary clustering of source-load characteristics as described in any one of claims 1 to 3 when executing the computer program.

Images