A power transmission section searching method based on variable scale near neighbor propagation clustering

Abstract

The application discloses a power transmission section searching method based on variable scale near neighbor propagation clustering, and steps of the method comprise the following steps: 1, establishing a power network model; 2, searching a power transmission section by variable scale near neighbor propagation clustering. The application considers the demand of the power network, divides the whole power network in different scales by performing multiple variable scale partition operations on a power topology diagram, searches more potential power transmission sections, is beneficial to a dispatching department to timely adjust an operation mode, thereby can avoid large-area power grid outage accidents, and guarantees safe and stable operation and optimal dispatching of the power grid.

Description

TECHNICAL FIELD

[0001] The application belongs to the field of safe and stable operation and optimal scheduling of power grid, and particularly relates to a transmission section searching method based on variable-scale near neighbor propagation clustering. BACKGROUND

[0002] With the increasingly wide application of new energy in the power system, new energy is connected to the power grid on a large scale, and the random and intermittent characteristics of new energy make the power grid structure more complex, and the transmission power fluctuation of the power grid more frequent. Meanwhile, the faults occurring in the power grid are very complex, and are accompanied by planned maintenance, which promotes the continuous change of the topology structure of the power grid, and makes the operation mode of the power grid more complex. Therefore, it is of great significance to search and identify the transmission section of the power grid quickly and accurately to ensure the safety of the power grid.

[0003] The method for searching the transmission section can be divided into two categories. The first category is based on N-1 contingency, which assumes that a line is overloaded and then finds the transmission section. The method for searching the transmission section based on N-1 contingency is based on a preset fault branch, and has poor adaptability to the real-time power grid with variable operation. The second category of method for searching the transmission section is based on the knowledge of graph theory, which simplifies the power network into a topology graph, and then searches the section of the topology graph. The traditional clustering method for searching the transmission section mostly depends on the selection of the clustering center, which leads to the omission of the search, and the division of the partition is too dependent on the expert experience. The number of traditional near neighbor propagation clustering centers is fixed, which directly affects the result of the section search, and there is a situation of missing and omitting the section. SUMMARY

[0004] The application is proposed to solve the above-mentioned problems of the prior art, and provides a transmission section searching method based on variable-scale near neighbor propagation clustering, so as to search more potential transmission sections, which is beneficial to the dispatching department to timely adjust the operation mode, thereby avoiding large-area power failure accidents of the power grid and ensuring the safe and stable operation and optimal scheduling of the power grid.

[0005] In order to achieve the above-mentioned application purposes, the application adopts the following technical scheme:

[0006] The transmission section searching method based on variable-scale near neighbor propagation clustering has the following steps:

[0007] Step 1, an electric power network topology graph is established by using a graph database, and weight information is assigned:

[0008] Step 1.1, an electric power system network topology graph is established according to N network nodes and an adjacent line relationship set R existing in the graph database; if there is an adjacent relationship between the i th node and the j th node, it means that there is a path r between the i th node and the j th node ij .

[0009] Step 1.2, in the power system network topology, the path r between the ith node and the jth node is calculated by formula (1) ij The power flow edge betweenness P B (r ij ) of the node:

[0010]

[0011] In formula (1), σ ij represents the shortest path between the ith node and the jth node, σ ij (r ij ) represents the shortest path between the ith node and the jth node through path r ij ; P r is the line power flow between the ith node and the jth node; P max is the maximum value of the line power flow;

[0012] Step two, the power system network topology is clustered by the reference degree variable scale operation to obtain the clustering partition and its tie line, and the transmission section meeting the conditions is judged according to the transmission section characteristics:

[0013] Step 2.1, define the maximum number of clustering centers as K and the damping coefficient as λ;

[0014] Define the number of loop iterations as t, and initialize t = 0;

[0015] Define the interval update number as m, and initialize m = 0;

[0016] Let the attraction degree r t (i,j) of the jth node suitable for the ith node clustering center in the tth iteration, and let the membership degree a t (i,j) of the ith node selecting the jth node as its clustering center in the tth iteration; The membership matrix A t and the attraction matrix R t of the tth iteration are composed of the membership and attraction between all nodes in the tth iteration; And initialize A t and R t are zero matrix;

[0017] Step 2.2, the similarity matrix S t (i,j) between the ith node and the jth node in the tth iteration is constructed by formula (2), so as to obtain the similarity matrix S t between nodes in the tth iteration;

[0018]

[0019] In formula (2), pt similarity matrix S for the tth iteration t reference degree constituted by the elements located on the diagonal line;

[0020] Step 2.3, using non-linear adjustment to the reference degree p for the tth iteration t performing variable scaling and clustering search to obtain the power transmission section:

[0021] Step 2.3.1, let the value range of the mth update be wherein e1 represents the parameter of the left end value of the value range E m , and e2 represents the parameter of the right end value of the value range E m ;

[0022] Step 2.3.2, define a variable b, and initialize b = 0, initialize let the step size be

[0023] Step 2.3.3, p t +bΔ is assigned to the similarity matrix S for the tth iteration t ;

[0024] Step 2.3.4, update the membership matrix A t and the attraction matrix R t for the tth iteration, and perform power transmission section judgment:

[0025] Step 2.3.3.1, calculate the attraction r t+1 (i,j) between the ith node and the jth node for the t+1th iteration by using formula (3):

[0026]

[0027] In formula (3), a t (i,k) represents the membership between the ith node and the kth node for the tth iteration, r t (i,k) represents the attraction between the ith node and the kth node for the tth iteration, and S t (i,k) represents the similarity matrix between the ith node and the kth node for the tth iteration;

[0028] Step 2.3.3.2, calculate the membership a t+1 (i,j) between the ith node and the jth node for the t+1th iteration by using formula (4):

[0029]

[0030] In formula (4), r t(k,j) represents the attraction degree between the kth node and the jth node in the tth iteration, r t (j,j) represents the attraction degree between the jth node and the jth node in the tth iteration;

[0031] Step 2.3.3.3, the attribution degree a t +1 (i,j) and the attraction degree r′ t+1 (i,j) are obtained by using formula (5) and formula (6) respectively;

[0032] a′ t+1 (i,j) = (1-λ)a t+1 (i,j) + λa′ t (i,j) (5)

[0033] r′ t+1 (i,j) = (1-λ)r t+1 (i,j) + λr′ t (i,j) (6)

[0034] In formula (5) and formula (6), λ is a damping coefficient, a′ t (i,j) and r′ t (i,j) are the attribution degree and the attraction degree after attenuation in the tth iteration;

[0035] Step 2.3.3.4, a′ t+1 (i,j) is assigned to a t+1 (i,j), and r t+1 (i,j) is assigned to r t+1 (i,j), so that the attribution matrix A t+1 and the attraction matrix R t+1 of the t+1th iteration are obtained;

[0036] Step 2.3.3.5, the clustering center of the t+1th iteration is obtained according to the attribution matrix A t+1 and the attraction matrix R t+1 of the t+1th iteration; and the clustering result of the t+1th iteration includes a clustering partition and a connection line, wherein the connection line between the ith node and the jth node is denoted as l i-j ; and the connection line satisfying the power transmission section characteristic in the clustering result of the t+1th iteration is taken as a group of power transmission sections; the power transmission section characteristic includes the cut-set characteristic, the power grid link vulnerability and the power flow direction consistency;

[0037] Step 2.3.3.6, the nodes in the power system network topology graph are divided according to the clustering center of the t+1th iteration, so as to obtain the clustering result of the t+1th iteration; ​

[0038] Step 2.3.3.7: Determine whether the number of cluster centers in the (t+1)th iteration exceeds the K value. If it does, it indicates that the reference degree p has been obtained. t The corresponding multiple sets of transmission sections constitute a complete transmission section; otherwise, proceed to step 2.3.3.8.

[0039] Step 2.3.3.8: After assigning b+1 to b, determine... Check if the condition is met. If it is, proceed to step 2.3.4; otherwise, return to step 2.3.3.

[0040] Step 2.3.4: After assigning m+1 to m, return to step 2.3.1 and execute sequentially.

[0041] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0042] 1. This invention takes into account the needs of the power network. By performing multiple variable-scale partitioning operations on the power topology map, the entire power network is divided at different scales. It does not rely on expert experience and has strong adaptability to the real-time power grid with its changing operation. It avoids the problem of missing sections in the search and helps the dispatching department to adjust the operation mode in a timely manner, thereby avoiding large-scale power outages and ensuring the safe and stable operation and optimized dispatch of the power grid.

[0043] 2. In response to the problem of the fixed number of cluster centers in traditional nearest-neighbor propagation, this invention adopts variable-scale partitioning operation, which can obtain a more comprehensive power transmission section while reducing redundancy.

[0044] 3. Considering that the number of nearest-neighbor propagation cluster centers does not change linearly, this invention adopts variable-scale partitioning operation, which is more in line with the actual power system and provides more monitoring objects for actual power grid operators, thereby reducing the incidence of cascading tripping accidents. Attached Figure Description

[0045] Figure 1 This is a flowchart of the power transmission section search method based on variable-scale nearest neighbor propagation clustering of the present invention;

[0046] Figure 2 This is a 9-node network topology diagram of the present invention;

[0047] Figure 3 The reference degree p for the m-th interval update and the t-th iteration of this invention t Schematic diagram of variable scale operation;

[0048] Figure 4 This is a schematic diagram of the power transmission section determination in the t-th iteration of the present invention. Detailed Implementation

[0049] In this embodiment, as Figure 1As shown, a transmission section search method based on variable-scale nearest neighbor propagation clustering is performed according to the following steps:

[0050] Step 1: Construct a power network topology graph using a graph database and assign weight information:

[0051] Step 1.1: Based on the N network nodes and their adjacent path relationships R existing in the graph database, establish a power system network topology graph; if there is an adjacency relationship between the i-th node and the j-th node, it means that there is a path r between the i-th node and the j-th node. ij In this embodiment, the Neo4j database is used to construct the network topology of the New England 39-node system. If the 1st node and the 39th node are adjacent, it means that there is a path r between the 1st node and the 39th node. 139 = (1,39);

[0052] Step 1.2: In the power system network topology diagram, use equation (1) to calculate the path r between the i-th node and the j-th node. ij Trend edge betweenness P B (r ij ):

[0053]

[0054] In equation (1), σ ij σ represents the shortest path between the i-th node and the j-th node. ij (r ij The path r between the i-th node and the j-th node is represented by ). ij Shortest path; P r P represents the power flow between the i-th node and the j-th node. max This represents the maximum value of the line power flow; for example... Figure 2 As shown, the significance of the power flow edge betweenness is illustrated using a 9-node system. 9 7 The edge betweenness is 0.50, r 6 7 The edge betweenness is 0.38, but r 9 7 Trend below r 6 7 Obviously, at this time r 6 7 They should be included in the monitoring targets, therefore the power flow edge betweenness is used as the weight information;

[0055] Step two involves using a reference degree variable scaling operation to perform nearest neighbor propagation clustering on the power system network topology map, obtaining cluster partitions and their interconnections, and determining the transmission sections that meet the conditions based on the characteristics of the transmission sections:

[0056] Step 2.1: Define the maximum number of cluster centers as K and the damping coefficient as λ; in this example, the maximum number of cluster centers is defined as 15 and the damping coefficient is 0.7. The damping coefficient is used for algorithm convergence.

[0057] Define the number of loop iterations as t, and initialize t = 0;

[0058] Define the interval update count as m, and initialize m = 0;

[0059] Let the attraction r in the t-th iteration be... t (i,j) represents the attractiveness of the j-th node as the cluster center of the i-th node, and let the membership degree a in the t-th iteration be... t (i,j) represents the suitability of the i-th node in choosing the j-th node as its cluster center; the membership matrix A of the t-th iteration is composed of the membership and attraction of all nodes in the t-th iteration. t and attraction matrix R t ; and initialize A t and R t All are zero matrices;

[0060] Step 2.2: Construct the similarity matrix S between the i-th node and the j-th node in the t-th iteration using equation (2). t (i,j), thus obtaining the similarity matrix S between nodes in the t-th iteration. t ;

[0061]

[0062] In equation (2), p t S is the similarity matrix in the t-th iteration. t The reference degree formed by the elements located on the diagonal;

[0063] Step 2.3: Adjust the reference degree p for the t-th iteration using nonlinear adjustment. t The transmission section is obtained by performing variable-scale analysis and cluster search:

[0064] Step 2.3.1: Let the value range of the m-th update be... Where e1 represents the value interval E m The parameter is the left endpoint value, and e2 represents the value range E. m The parameter is the right endpoint value; in this example, the initial value range is set to Em = (max{S t (i,j)} / 2,max{S t (i,j)};

[0065] Step 2.3.2: Define variable b and initialize b = 0. Let the step length be

[0066] This example demonstrates the reference degree p for the m-th interval update and the t-th iteration. t Variable scale operations such as Figure 3 As shown;

[0067] Step 2.3.3, p t +bΔ is assigned to the similarity matrix S in the t-th iteration. t ;

[0068] Step 2.3.4: Update the membership matrix A for the t-th iteration. t and attraction matrix R t And determine the transmission section:

[0069] Step 2.3.3.1: Calculate the attraction r between the i-th node and the j-th node in the (t+1)-th iteration using equation (3). t+1 (i,j):

[0070]

[0071] In equation (3), a t (i,k) represents the degree of belonging between the i-th node and the k-th node in the t-th iteration, r t (i,k) represents the attraction between the i-th node and the k-th node in the t-th iteration, S t (i,k) represents the similarity matrix between the i-th node and the k-th node in the t-th iteration;

[0072] Step 2.3.3.2: Calculate the degree of belonging, a, between the i-th node and the j-th node in the (t+1)-th iteration using equation (4). t+1 (i,j):

[0073]

[0074] In equation (4), r t (k,j) represents the attraction between the k-th node and the j-th node in the t-th iteration, r t (j,j) represents the attraction between the j-th node and the j-th node in the t-th iteration;

[0075] Step 2.3.3.3: Use equations (5) and (6) to obtain the degree of belonging after the decay in the (t+1)th iteration, respectively. t+1 (i,j) and attraction r′ t+1 (i,j):

[0076] a′ t+1 (i,j)=(1-λ)a t+1 (i,j)+λa′ t(i,j) (5)

[0077] r′ t+1 (i,j)=(1-λ)r t+1 (i,j)+λr′ t (i,j) (6)

[0078] In equations (5) and (6), λ is the damping coefficient, and a′ t (i,j), r′ t (i,j) represents the affiliation and attraction after the decay in the t-th iteration;

[0079] Step 2.3.3.4, change a′ t+1 (i,j) is assigned to a t+1 (i,j), r t+1 (i,j) is assigned to r t+1 (i,j), thus obtaining the membership matrix A for the (t+1)th iteration. t+1 and attraction matrix R t+1 ;

[0080] Step 2.3.3.5: Based on the membership matrix A of the (t+1)th iteration... t+1 and attraction matrix R t+1 The cluster centers of the (t+1)th iteration are obtained; and the clustering result of the (t+1)th iteration includes the cluster partitions and their connecting lines, wherein the connecting line between the i-th node and the j-th node is denoted as l. i-j The clustering results of the (t+1)th iteration that satisfy the characteristics of transmission sections are considered as a group of transmission sections. These characteristics include cut-set characteristics, grid link vulnerability, and power flow direction consistency. The method for determining the transmission sections in the tth iteration is as follows: The power network is divided into multiple zones, and the tie lines and power flow directions of each zone are marked. Each zone is an isolated island to the outside world, therefore the tie lines to the outside world already meet the cut-set characteristics. When there are few external lines, the probability of tripping due to power flow transfer is high. However, if there are many external lines, even if the power flow direction is consistent, the lines can share the load. Therefore, if the number of external lines in a certain zone is less than a certain value, then this group of lines meets the grid link vulnerability requirement. In this example, the number of external lines is set to no more than 4. Therefore, as long as the power flow direction of this group of lines is consistent and the number of external lines does not exceed 4, it can be considered a transmission section. Figure 4 As shown, lines l7, l8, and l9 meet the transmission section requirements.

[0081] Step 2.3.3.6: Divide the nodes in the power system network topology diagram according to the cluster centers of the (t+1)th iteration to obtain the clustering results of the (t+1)th iteration;

[0082] Step 2.3.3.7: Determine whether the number of cluster centers in the (t+1)th iteration exceeds the K value. If it does, it indicates that the reference degree p has been obtained. t The corresponding multiple sets of transmission sections constitute a complete transmission section; otherwise, proceed to step 2.3.3.8.

[0083] Step 2.3.3.8: After assigning b+1 to b, determine... Check if the condition is met. If it is met, proceed to step 2.3.4; otherwise, return to step 2.3.3.

[0084] Step 2.3.4: After assigning m+1 to m, return to step 2.3.1 and execute sequentially.

[0085] To verify the effectiveness of this method, a network topology diagram of the New England 39-node system was constructed using the Neo4j database, and a transmission section search was performed on this topology diagram. In this embodiment, the traditional nearest neighbor propagation clustering transmission section search method was used for performance comparison. The experimental results are shown in Table 1.

[0086] Table 1

[0087]

[0088] Experimental results show that the method of the present invention is more comprehensive in effect than the traditional method, thus proving the feasibility of the method proposed in this invention.

Claims

1. A method for searching transmission sections based on variable-scale nearest neighbor propagation clustering, characterized by: Follow these steps: Step 1: Build a power network topology using a graph database and assign weight information: Step 1.1: Based on the N network nodes and their adjacent line relationship set R existing in the graph database, establish the power system network topology graph; If the i-th node and the j-th node are adjacent, it means that there is a path between the i-th node and the j-th node. ; Step 1.2: In the power system network topology diagram, use equation (1) to calculate the path between the i-th node and the j-th node. Trend edge intermediary : (1) In equation (1), This represents the shortest path between the i-th node and the j-th node. This represents the path between the i-th node and the j-th node. The shortest path; This represents the power flow between the i-th node and the j-th node. This represents the maximum value of the power flow along the line. Step two involves using a reference degree variable scaling operation to perform nearest neighbor propagation clustering on the power system network topology map, obtaining cluster partitions and their interconnections, and determining the transmission sections that meet the conditions based on the characteristics of the transmission sections: Step 2.1: Define the maximum number of cluster centers as K, and the damping coefficient as... ; Define the number of loop iterations as t, and initialize t=0; Define the number of interval updates as m, and initialize m=0; Let the attraction of the t-th iteration be... Let the attractiveness of the j-th node be the cluster center of the i-th node, and let the membership degree be the degree of the t-th iteration. This indicates the suitability of the i-th node in choosing the j-th node as its cluster center; the membership degree matrix for the t-th iteration is composed of the membership degree and attractiveness among all nodes in the t-th iteration. and attraction matrix and initialize and All are zero matrices; Step 2.2: Construct the similarity matrix between the i-th node and the j-th node in the t-th iteration using equation (2). Thus, the similarity matrix between nodes in the t-th iteration is obtained. ; (2) In equation (2), Let t be the similarity matrix of the t-th iteration. The reference degree formed by the elements located on the diagonal; Step 2.3: Adjust the reference degree p for the t-th iteration using nonlinear adjustment. t The transmission section is obtained by performing variable-scale analysis and cluster search: Step 2.3.1: Let the value range of the m-th update be... ;in, Indicates the range of values ​​E m The parameter of the left endpoint value, Indicates the range of values ​​E m The parameter of the right endpoint value; Step 2.3.2, Define variables and initialize ,initialization Let the step size be ; Step 2.3.3 Assigned to the similarity matrix in the t-th iteration ; Step 2.3.4: Update the membership matrix for the t-th iteration. and attraction matrix And determine the transmission section: Step 2.3.3.1: Calculate the attraction between the i-th node and the j-th node in the (t+1)-th iteration using equation (3). : (3) In equation (3), This represents the degree of belonging between the i-th node and the k-th node in the t-th iteration. This represents the attraction between the i-th node and the k-th node in the t-th iteration. Let represent the similarity matrix between the i-th node and the k-th node in the t-th iteration; Step 2.3.3.2: Calculate the degree of belonging between the i-th node and the j-th node in the (t+1)-th iteration using equation (4). : (4) In equation (4), This represents the attraction between the k-th node and the j-th node in the t-th iteration. This represents the attraction between the j-th node and the j-th node in the t-th iteration; Step 2.3.3.3: Use equations (5) and (6) to obtain the degree of belonging after the decay in the (t+1)th iteration. and attractiveness : (5) (6) In equations (5) and (6), The damping coefficient is... , Let be the affiliation and attraction after the decay in the t-th iteration; Step 2.3.3.4, will Assign to , Assign to Thus, the membership matrix of the (t+1)th iteration is obtained. and attraction matrix ; Step 2.3.3.5: Based on the membership matrix of the (t+1)th iteration and attraction matrix The cluster centers of the (t+1)th iteration are obtained; and the clustering result of the (t+1)th iteration includes the cluster partitions and their connecting lines, wherein the connecting line between the i-th node and the j-th node is denoted as l. i-j The tie lines that satisfy the transmission section characteristics in the clustering results of the (t+1)th iteration are taken as a group of transmission sections; the transmission section characteristics include: cut set characteristics, grid link vulnerability, and power flow direction consistency. Step 2.3.3.6: Divide the nodes in the power system network topology diagram according to the cluster centers of the (t+1)th iteration to obtain the clustering results of the (t+1)th iteration; Step 2.3.3.7: Determine whether the number of cluster centers in the (t+1)th iteration exceeds the K value. If it does, it indicates that the reference score has been obtained. The corresponding multiple sets of transmission sections constitute a complete transmission section; otherwise, proceed to step 2.3.3.

8. Step 2.3.3.8, will Assign to Then, make a judgment Check if the condition is met. If it is, proceed to step 2.3.4; otherwise, return to step 2.3.

3. Step 2.3.4: After assigning m+1 to m, return to step 2.3.1 and execute sequentially.

Images