Transformer area box meter identification method and system

Abstract

The invention discloses a transformer area box meter identification method and system, and the method comprises the steps: obtaining wireless RSSI data of a current electric meter and surrounding neighbor electric meters detected in an HRF channel, and determining a first matrix according to the detection result of the wireless RSSI data; preprocessing the first matrix to obtain a second matrix, and converting the second matrix into a scalar matrix; performing clustering operation on the scalar matrix by using a hierarchical clustering algorithm to determine a third matrix; a point with the cluster number being 1 in the third matrix is used as a first point cluster, the third matrix is circulated, whether a target cluster can be matched or not is judged, whether cluster combination is conducted on the first point cluster and the target cluster or not is determined according to the matching result, then the recognition result of the transformer area box meters is determined according to the cluster combination result, and each electricity meter corresponds to one cluster. The method is short in identification time, high in accuracy, low in chip capability requirement and low in memory occupancy rate.

Description

Technical Field

[0001] The present invention relates to the field of power system data processing, and in particular to a method and system for identifying box-mounted meters in a transformer substation area. Background Art

[0002] The low-voltage distribution network is located at the end of the power distribution system, with a wide distribution, complex environment, and great difficulty in operation and maintenance. For traditional management means, it is very time-consuming to check the relationship between each meter and box one by one, and the workload of manual operation is large, prone to negligence or errors, and it does not have the ability of real-time monitoring. Therefore, the accurate identification of its topological relationship is of great significance for the safe and stable operation of the distribution network.

[0003] Existing methods mainly collect a large amount of measurement data, such as measurement data of voltage, current, power, etc., and utilize the correlation of voltage data and the relationship between currents. However, this kind of data is easily affected by the environment. Through the clustering of a large amount of data, it is easy to fall into the problem of low solution efficiency due to huge amounts of data, and has relatively high requirements for the chip processing ability. Summary of the Invention

[0004] (I) Object of the Invention

[0005] The object of the present invention is to provide a method and system for identifying box-mounted meters in a transformer substation area. By using the RSSI data of the HRF channel in combination with the distance attenuation law, this method can quickly identify the electricity meters in the same electrical box or different electrical meter boxes, with a short identification time, high accuracy, and low requirements for the chip ability and low memory occupancy rate.

[0006] (II) Technical Solution

[0007] To solve the above problems, a first aspect of the present invention provides a method for identifying box-mounted meters in a transformer substation area. The transformer substation area includes N electricity meters, and the method includes:

[0008] S1, obtaining the wireless RSSI data detected by the current electricity meter and the surrounding neighbor electricity meters in the HRF channel;

[0009] S2, determining a first matrix according to the detection result of the wireless RSSI data, where the elements of the first matrix are the wireless RSSI data measured between each electricity meter;

[0010] S3, preprocessing the first matrix to determine a second matrix and converting the second matrix into a scalar matrix. The elements of the second matrix are the wireless RSSI data between each electricity meter after preprocessing and screening. In the second matrix, each electricity meter corresponds to a cluster;

[0011] S4. Use the hierarchical clustering algorithm to perform clustering and merging operations on the scalar matrix to obtain a third matrix. The elements of the third matrix include the cluster numbers of each electricity meter and the distances between each pair of electricity meters. Each cluster before merging is numbered in the third matrix.

[0012] S5. Take the point with cluster number 1 in the third matrix as the initial cluster. Loop through the third matrix to determine whether a target cluster can be matched. Based on the matching result, determine whether to merge the initial cluster with the target cluster, and then determine the identification result of the distribution box meters in the substation area according to the cluster merging result. In the first loop, the cluster number of the current electricity meter is 1. In subsequent loops, the merged cluster is used as the initial cluster.

[0013] Further, the method further includes: S6. If cluster merging cannot be performed, perform data expansion on the first matrix to obtain the cluster set of the current electricity meter, then determine whether the target cluster belongs to a subset of the cluster set. Based on the determination result, determine whether to perform cluster merging, and then determine the identification result of the distribution box meters in the substation area according to the cluster merging result.

[0014] Further, step S2 includes:

[0015] If no wireless RSSI data is detected, form a separate cluster for the current electricity meter, and return to S1 to continue traversing the next electricity meter.

[0016] If wireless RSSI data is detected, determine whether the surrounding neighbor electricity meters corresponding to the wireless RSSI data include allocated electricity meters. If so, exclude the allocated electricity meters, retain the remaining unallocated electricity meters, and form a first matrix with the current electricity meter.

[0017] Further, step S3 includes:

[0018] If the wireless RSSI data except the diagonal in the first matrix is the default value, compare the wireless RSSI data between the electricity meter corresponding to the default value and the current electricity meter, select and retain the electricity meter corresponding to the wireless RSSI data with the smallest attenuation, exclude the other electricity meters corresponding to the default value, and form a second matrix with the remaining electricity meters.

[0019] Further, in step S5, determining whether a target cluster can be matched and determining whether to merge the initial cluster with the target cluster according to the matching result includes:

[0020] If no target cluster is matched, return to S1 to continue traversing to the next electricity meter.

[0021] If a target cluster is matched, but there are more than two electricity meters between the initial cluster and the target cluster that cannot detect wireless RSSI data from each other, return to S1 to continue traversing to the next electricity meter.

[0022] If the target cluster is matched and the electric meters between the first point cluster and the target cluster can detect wireless RSSI data from each other, then merge the first point cluster and the target cluster. After updating the cluster class information of the first point cluster, repeat step S5.

[0023] Further, step S4 further includes: determining a merging threshold or an empirical threshold according to the merging relationship of the current electric meter in the third matrix.

[0024] Further, S5 further includes:

[0025] If the distance between the first point cluster and the target cluster after merging is less than the merging threshold, then merge the first point cluster and the target cluster, update the cluster class information of the first point cluster, and then repeat step S5;

[0026] Or if there are more than two electric meters between the first point cluster and the target cluster whose detected attenuation values are less than the empirical threshold, then merge the first point cluster and the target cluster, update the cluster class information of the first point cluster, and then repeat step S5;

[0027] Further, the data expansion of the first matrix in S6 to obtain the cluster set of the current electric meter includes:

[0028] Add the excluded and assigned electric meters to the first matrix for expansion;

[0029] Repeat steps S3 - S4 on the expanded first matrix, and the obtained third matrix is the cluster set of the current electric meter.

[0030] Further, in S6, determining whether the target cluster belongs to a subset of the cluster set and determining whether to perform cluster merging according to the judgment result includes:

[0031] If the target cluster belongs to a subset of the cluster set, then merge the first point cluster and the target cluster;

[0032] If the target cluster does not belong to a subset of the cluster set, then update the cluster information of the third matrix and return to S1 to continue traversing to the next electric meter.

[0033] In addition, a second aspect of the present invention provides a substation box meter identification system. The substation includes N electric meters, including:

[0034] A detection module for obtaining wireless RSSI data detected by the current electric meter and surrounding neighbor electric meters on the HRF channel;

[0035] The first matrix determination module is configured to determine a first matrix according to the detection result of the wireless RSSI data, where the elements of the first matrix are the wireless RSSI data measured between each electricity meter;

[0036] The conversion module is configured to preprocess the first matrix to obtain a second matrix, and convert the second matrix into a scalar matrix. The elements of the second matrix are the wireless RSSI data between each electricity meter after preprocessing and screening. In the second matrix, each electricity meter corresponds to a cluster respectively;

[0037] The clustering operation module is configured to perform clustering and merging operations on the scalar matrix by using a hierarchical clustering algorithm to determine a third matrix; the elements of the third matrix include the cluster numbers of each electricity meter and the distances between each electricity meter, and each cluster before merging is numbered in the third matrix.

[0038] The merging module is configured to use the point with the cluster number 1 in the third matrix as the first point cluster, loop through the third matrix, determine whether a target cluster can be matched, determine whether to merge the first point cluster with the target cluster according to the matching result, and then determine the identification result of the substation box meter according to the cluster merging result. When looping for the first time, the cluster number of the current electricity meter is 1, and in subsequent loops, the merged cluster is used as the first point cluster.

[0039] Further, the system further includes an expansion module, which is configured to, if cluster merging cannot be performed, perform data expansion on the third matrix to obtain the cluster set of the current electricity meter, then determine whether the target cluster belongs to a subset of the cluster set, determine whether to perform cluster merging according to the determination result, and then determine the identification result of the substation box meter according to the cluster merging result.

[0040] (III) Beneficial effects

[0041] The above technical solution of the present invention has the following beneficial technical effects: The present invention provides a method and system for identifying a substation box meter. Before algorithm identification, wireless RSSI data is first collected. Suppose there are N meters (i.e., N nodes) in the current substation area. By obtaining the current electricity meter M iRegarding the wireless RSSI data detected by the electricity meters of surrounding neighbors on the HRF channel, the attenuation between meters in the same meter box is relatively small, while the attenuation between different meter boxes is relatively large. That is, the farther the distance, the greater the attenuation, and the closer the distance, the smaller the attenuation. According to this rule, the attenuation data is converted into a scalar. The conversion process is as follows: First, based on the detection results of the wireless RSSI data, the first matrix is determined. The first matrix is the matrix Y, which measures the original RSSI data between each electricity meter. Then, the first matrix is preprocessed to determine the second matrix, which is the Y_N matrix. Next, the second matrix is converted into a scalar matrix. Finally, the hierarchical clustering algorithm is used to perform clustering operations on the scalar matrix, calculate the Euclidean distance, and determine the third matrix. Finally, according to the conditions for merging clusters, it is judged whether to merge the starting point cluster and the target cluster, and the identification result of the meter box in the substation area is determined based on the cluster merging result. The specific algorithm implementation is achieved at the CCO (Central Coordinator). In the above algorithm process, only one point is selected each time (not all points are clustered together because a large amount of memory is required for intermediate variables during the algorithm calculation). Multiple conditional judgment validations are performed to obtain the cluster merging information of this point, improving the identification accuracy and ensuring that the chip and memory capabilities of the current module meet the requirements with low cost. This method has a short identification time and high accuracy. Each time only the cluster of one point is analyzed until all points are allocated, with low requirements for chip capabilities and low memory occupancy rate. Brief Description of the Drawings

[0042] Figure 1 It is a flowchart of a method for identifying meter boxes in a substation area according to the present invention;

[0043] Figure 2 It is a schematic diagram of a system for identifying meter boxes in a substation area according to the present invention. Detailed Embodiments

[0044] To make the objectives, technical solutions, and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below in conjunction with specific embodiments and with reference to the accompanying drawings. It should be understood that these descriptions are exemplary only and are not intended to limit the scope of the present invention. In addition, in the following description, the descriptions of well-known structures and technologies are omitted to avoid unnecessarily confusing the concepts of the present invention.

[0045] The present invention utilizes the RSSI data of the HRF channel and the fact that the attenuation between meters in the same meter box is relatively small, while the attenuation between different meter boxes is relatively large, that is, the farther the distance, the greater the attenuation; the closer the distance, the smaller the attenuation. The attenuation data is converted into a scalar, and then the Euclidean distance is calculated. The hierarchical clustering algorithm in machine learning is used to obtain the clustering result, and an adaptive merging threshold is selected to obtain the corresponding cluster result. The specific algorithm implementation is realized on the CCO (Central Coordinator) side. The specific method is as follows:

[0046] As Figure 1 shown, the first aspect of the present invention provides a method for identifying meter boxes in a power distribution area. The power distribution area includes N electric meters, and the N electric meters are N nodes. The method includes:

[0047] S1, obtain the wireless RSSI data detected by the current meter M i and the surrounding neighbor meters in the HRF channel, where i starts from 1 and its value range is 1 to N. According to the detection result of the wireless RSSI data, a first matrix is determined. The row and column elements of the first matrix are all the wireless RSSI data measured between each pair of meters. Before the algorithm identification, first, the wireless RSSI data is collected. Only the RSSI data of the neighbor nodes of each point needs to be collected (the neighbor nodes refer to other nodes directly connected to the current node). After the collection, the algorithm identification is carried out, and the specific algorithm implementation is realized on the CCO side. The number of meters to be identified in the current power distribution area is N in total. Which meters are in the same meter box needs to be identified by this algorithm.

[0048] S2, according to the detection result of the wireless RSSI data, determine a first matrix. This step includes:

[0049] S21, if no wireless RSSI data is detected, form a separate cluster for the current meter, and return to S1 to continue traversing the next meter;

[0050] S22. If wireless RSSI data is detected, determine whether the surrounding neighbor meters corresponding to the wireless RSSI data include the allocated meters. If so, exclude the allocated meters, retain the remaining unallocated meters, and form a first matrix with the current meter. The first matrix is matrix Y, and the detected data is the original wireless RSSI data measured between each meter. The number of rows and columns of the first matrix is the same, both being the number of unallocated meters determined in the detection result + 1. For example, among the meters that have wireless RSSI data with the current meter, after excluding the allocated meters, if the number of remaining unallocated meters is 5, then the first matrix is a 6*6 matrix, and the elements in the matrix are the RSSI values between the meters corresponding to the row and column of the element. The example is as follows: Assume that the points that the first point M1 can detect are M3, M4, M8, M9, M15. Then the matrix Y formed by these 6 points is a 6*6 matrix as shown in Table 1:

[0051] Table 1 Example of the first matrix

[0052]

[0053] In Table 1, the numbers represent the matrix elements, and the values are the wireless data measured between the corresponding nodes. For example, the element -36 in the first row and the second column represents the wireless RSSI data measured between node M1 and node M3. Among them, 127 represents the default value of the wireless RSSI data of the meter itself, which is the wireless RSSI data corresponding to the node itself. The closer the value of the wireless RSSI data is to 0, the smaller the attenuation.

[0054] S3. Preprocess the first matrix to obtain a second matrix, and convert the second matrix into a scalar matrix. The elements of the second matrix are the wireless RSSI data between each meter after preprocessing and screening. In the second matrix, each meter corresponds to a cluster. The second matrix is Y - N. This step S3 specifically includes:

[0055] If the wireless RSSI data except the diagonal in the first matrix is the default value, compare the wireless RSSI data between the meter corresponding to the default value and the current meter, select and retain the meter corresponding to the wireless RSSI data with the smallest attenuation, exclude the other meters corresponding to the default value, and form a second matrix with the remaining meters, as described in steps 3.1 and 3.2 for example.

[0056] S4. Use the hierarchical clustering algorithm to perform clustering and merging operations on the scalar matrix to determine the third matrix. The third matrix is matrix T, and matrix T stores or represents the binary tree in the form of a matrix. The step S4 further includes: determining a merging threshold or an empirical threshold according to the merging relationship of the current electricity meter in the third matrix. The Euclidean distance is calculated in the hierarchical clustering algorithm of this step. By calculating the distance, according to the distance between points, hierarchical clustering is used for classification. Then, use the machine learning hierarchical clustering algorithm to obtain the clustering tree result and select an adaptive merging threshold. Find out the merging distance and calculate its gap value, analyze the gap value and the average value of all merging distances related to the current point, and select an appropriate threshold as the merging threshold (cluster merging threshold) according to the merging distance distribution. The present invention adopts the hierarchical clustering algorithm and does not require specifying the number of clusters in advance. The row and column elements of the third matrix include the cluster numbers of each electricity meter and the distances between each electricity meter. Each cluster in the third matrix is numbered, and each electricity meter corresponds to a cluster respectively. For example, still assuming 6 electricity meters M1, M3, M4, M8, M9, and M15, before merging, each point corresponds to a single cluster, that is, the clusters corresponding to M1, M3, M4, M8, M9, and M15 are 1, 2, 3, 4, 5, and 6 respectively. The following are examples of the merging threshold and the third matrix respectively;

[0057] (1) The selection of the merging threshold includes three cases, which are exemplified as follows:

[0058] The first case is shown in Table 2:

[0059] Table 2 First Example of Merging Threshold

[0060] Cluster ID Cluster ID Distance New merged cluster ID 2 4 6.7082 7 3 6 13.1909 8 1 5 19.6214 9 7 8 22.2036 10 9 10 45.89 11

[0061] For the above-mentioned merging threshold, the gap value is the interval. First, find out which clusters are merged for the current point. First, 1 and 5 are merged into 9, and 9 and 10 are merged into 11. There are two distances, and the corresponding distance sequences are 19.6214 and 45.89. Calculate the gap value, that is, subtract the previous one from the latter, 45.89 - 19.6214. There is only one gap sequence, so the merging threshold will take the average value. The method of taking the average value is to take the average of the sum of all distances, (6.7082 + 13.1909 + 19.6214 + 22.2036 + 45.89) / 5;

[0062] The second case is shown in Table 3:

[0063] Table 3 Second Example of Merging Threshold

[0064] Cluster ID Cluster ID Distance New merged cluster ID 4 7 8.8318 9 1 2 13.8564 10 3 6 14.2478 11 5 8 14.9332 12 9 12 23.1301 13 10 11 30.6594 14 13 14 72.0625 15

[0065] The description is as follows: The merged distance sequence: 13.8564, 30.6594, 72.0625. There are only 2 values in the Gap value sequence: 16.8034, 1.4031. The merging threshold selects the corresponding threshold 30.6594 with the largest Gap value.

[0066] The third case: If none of the above cases are satisfied, select the threshold corresponding to the largest Gap value.

[0067] (2) The value list of the third matrix is as follows:

[0068] Table 4 Example of the value of the third matrix

[0069] Cluster ID Cluster ID Distance New merged cluster ID 2 4 6.7082 7 3 6 13.1909 8 1 5 19.6214 9 7 8 22.2036 10 9 10 45.89 11

[0070] S5. Take the point with cluster number 1 in the third matrix as the initial cluster. Loop through the third matrix to determine whether a target cluster can be matched. According to the matching result, determine whether to merge the initial cluster with the target cluster, and then determine the recognition result of the distribution box meter according to the cluster merging result. When looping for the first time, the cluster number of the current meter is 1, and in subsequent loops, the merged cluster is used as the initial cluster. In step S5, determining whether a target cluster can be matched and determining whether to merge the initial cluster with the target cluster according to the matching result includes:

[0071] S51. If no target cluster is matched, return to S1 and continue to traverse to the next meter;

[0072] S52. If a target cluster is matched, but there are more than two meters between the initial cluster and the target cluster that cannot detect wireless RSSI data from each other, return to S1 and continue to traverse to the next meter;

[0073] S53. If a target cluster is matched, and the meters between the initial cluster and the target cluster can all detect wireless RSSI data from each other, merge the initial cluster with the target cluster. After updating the cluster class information of the initial cluster, repeat step S5. For example, in Table 5: The target cluster starts from the cluster closest to the current meter. For example, the cluster 5 closest to the current meter (cluster ID is 1, initial cluster). It is necessary to determine whether cluster 1 and cluster 5 meet the matching result. If they meet, merge them (indicating that 5 and 1 are of the same class). After merging, it is cluster 9. Update the cluster class information of the merged cluster, and then take cluster 9 as the initial cluster and continue to find the next target cluster. For example, in Table 5, the target cluster is 7. Loop according to the above steps until all clusters that can be merged with the initial cluster (i.e., all meters that can be merged with the current meter) are found.

[0074] Table 5 Example of cluster merging of matrix T

[0075] Cluster ID Cluster ID Distance New merged cluster ID 2 4 6.7082 7 3 6 13.1909 8 1 5 19.6214 9 6 8 22.2036 10 7 9 45.89 11

[0076] In addition, S5 further includes:

[0077] S54 If the distance between the first point cluster and the target cluster for merging is less than the merging threshold, then merge the first point cluster and the target cluster, update the cluster class information of the first point cluster, and then repeat step S5;

[0078] Or S55, if there are more than two electric meters between the first point cluster and the target cluster whose detected attenuation values are less than the empirical threshold, then merge the first point cluster and the target cluster, update the cluster class information of the first point cluster, and then repeat step S5.

[0079] Furthermore, the method further includes: S6. If cluster merging cannot be performed, perform data expansion on the third matrix to obtain the cluster set of the current electric meters, then determine whether the target cluster belongs to a subset of the cluster set, determine whether to perform cluster merging according to the judgment result, and then determine the identification result of the distribution box meters of the substation area according to the cluster merging result. Performing data expansion on the third matrix in this step to obtain the cluster set of the current electric meters includes: adding the excluded and assigned electric meters to the first matrix for expansion; then repeating steps S3 - S4 on the expanded first matrix, and the obtained third matrix is the cluster set of the current electric meters, and this cluster set is denoted as target. Determining whether the target cluster belongs to a subset of the cluster set in S6 and determining whether to perform cluster merging according to the judgment result includes:

[0080] S61 If the target cluster belongs to a subset of the cluster set, then merge the first point cluster and the target cluster;

[0081] S62 If the target cluster does not belong to a subset of the cluster set, then update the cluster information of the third matrix and return to S1 to continue traversing to the next electric meter.

[0082] In addition, as Figure 2 shown, a second aspect of the present invention provides a distribution box meter identification system for a substation area. The substation area includes N electric meters, including:

[0083] A detection module 21 for obtaining the wireless RSSI data detected by the current electric meter M i and the surrounding neighbor electric meters on the HRF channel. The elements of the first matrix are the wireless RSSI data measured between each pair of electric meters;

[0084] A first matrix determination module 22 for determining a first matrix according to the detection result of the wireless RSSI data. The elements of the first matrix are the wireless RSSI data measured between each pair of electric meters;

[0085] A conversion module 23, configured to preprocess the first matrix to obtain a second matrix, and convert the second matrix into a scalar matrix, wherein the elements of the second matrix are the wireless RSSI data between the electric meters after preprocessing and screening, and each electric meter in the second matrix corresponds to a cluster;

[0086] A clustering operation module 24 is used to perform a clustering and merging operation on the scalar matrix using a hierarchical clustering algorithm to obtain a third matrix, wherein the third matrix includes cluster numbers of each electric meter and distances between each electric meter, and each cluster before merging is numbered in the third matrix, and each electric meter corresponds to a cluster;

[0087] The merging module 25 is used to take the point with cluster number 1 in the third matrix as the first point cluster, loop the third matrix, determine whether the target cluster can be matched, determine whether to merge the first point cluster with the target cluster according to the matching result, and then determine the recognition result of the substation box meter according to the cluster merging result, wherein the cluster number of the current meter is 1 in the first loop, and the merged cluster is used as the first point cluster in subsequent loops. In addition, the system also includes an expansion module 26, which is used to perform data expansion on the third matrix if cluster merging cannot be performed, obtain the cluster set of the current meter, and then determine whether the target cluster belongs to a subset of the cluster set, determine whether to merge the clusters according to the judgment result, and then determine the recognition result of the substation box meter according to the cluster merging result.

[0088] The present invention is described in detail below with reference to specific embodiments.

[0089] Assume that there are N tables in the current area, that is, N nodes {M1, M2...M N}, use the wireless RSSI data measured between points for meter box identification. First, wireless RSSI data collection is performed. It only needs to collect RSSI data of neighboring nodes of each point. After the collection, the algorithm calculation is performed. The specific algorithm is implemented in CCO measurement. The specific process is as follows:

[0090] (1) Select the first point M1, find the wireless RSSI data that can be measured by this point, remove the points that have been assigned to clusters, and only keep the points that are not assigned to clusters; the wireless RSSI data that can be measured by point M1 is used to form a matrix Y (first matrix). Assume that the set of points that can measure wireless data by point M1 and are all unassigned points is A = {M11, M12, M13...M1K}, then the wireless data between M1 and A constitutes the matrix Y (first matrix);

[0091] (2) If point M1 does not detect other wireless data, it is considered that the point belongs to a separate cluster, and the cluster information of the current point is updated, and the next corresponding point M is found. i , return to step 1;

[0092] (3) Preprocessing the matrix Y (the first matrix) to obtain the second matrix includes:

[0093] 3.1 If the wireless RSSI data other than the diagonal in the first matrix is ​​the default value, then compare the wireless RSSI data between the meter corresponding to the default value and the current meter, select and retain the meter corresponding to the wireless RSSI data with the smallest attenuation, remove the other meters corresponding to the default value, and form the second matrix with the remaining meters. That is: if there is a meter set with the default value (that is, the corresponding meters cannot measure each other) in the data other than the diagonal in the first matrix, select the meter set with the smallest attenuation with the current meter and retain it, and remove the other meters in the meter set. For example, if M13 and M1K cannot measure each other, then select the value with the smallest attenuation between (M1, M13) and (M1, M1K). Assuming that (M1, M1K) has the smallest attenuation, then remove the M13 point, and form a new matrix Y_N with the wireless data between M1 and {M11, M12, M14...M1K}.

[0094] 3.2 Convert the new matrix Y_N into a scalar matrix. Set the diagonal values ​​of the scalar matrix to 0. The diagonal means measuring itself to itself, and the corresponding measurement is 0.

[0095] (4) Each point is first treated as a separate cluster, and the resulting scalar matrix is ​​classified using a hierarchical clustering algorithm, that is, the matrix T (the third matrix) of cluster information is obtained. The total number of points is the order of the Y_N matrix, and point M1 is ranked at the first point of the point sequence.

[0096] (5) According to the matrix T, we only need to care about the merging relationship of the current point M1 in the matrix T, find out the merging distance and calculate its gap value (the gap value is used to evaluate the effectiveness of the clustering results and determine the optimal number of clusters), analyze the gap value and the average value of all merging distances about the current point M1, and select a suitable threshold as the merging threshold (cluster merging threshold, if the threshold is not exceeded during the cluster merging process, the clusters are merged, or determine the ε empirical value that meets the merging requirements (i.e., the empirical threshold, by collecting a large number of substations and analyzing the attenuation data belonging to the same meter box, an ε empirical value will be obtained);

[0097] (6) For the first time, the point numbered 1 in the T matrix cluster is set as the first point cluster (previous point);

[0098] (7) Loop through the matrix T to find the next cluster to be merged with the previous one (next is the target cluster):

[0099] 7.1 If not found, update the cluster information of the merged point and continue to find the next point M i, go back to step (1), and continue to select the cluster information of the next point; updating the cluster information means that, for example, if it is the first time to find the cluster information of point M1, and when it reaches step 7, the cluster information of 1 has been found. Assuming that 3, 5, 9, and 1 are in one cluster, after finding it, go back to step (1), continue to find the next node, and it is required that the point has not been assigned to a cluster. Then the next point to be found is 2.

[0100] 7.2 If there are at least two points between the two clusters previous and next to be merged where the wireless RSSI cannot be measured from each other, update the cluster information of the merged points in the third matrix, and directly return to step (1) to continue finding the next corresponding point M i ;

[0101] 7.3 If the wireless RSSI data can be detected between the two clusters previous and next to be merged, then directly merge them. After updating the cluster class information of the starting point cluster, go back to step 6, reassign the value to previous, and repeat steps 6 and 7;

[0102] 7.4 If the attenuation value between at least two points between the two clusters previous and next to be merged is less than the ε value, it can be considered that they also belong to one cluster, that is, the condition for cluster merging is met, and then merge them. Then go back to step 6, reassign the value to previous, and repeat steps 6 and 7;

[0103] (8) If the current point M1 is merged during the last merge of matrix T or the previous merge of the cluster that meets the last merge condition contains point M1, then enter the data expansion. The specific expansion process is as follows:

[0104] 8.1 Only care about the points whose wireless RSSI data can be measured by the current point M1, and do not care whether the measured points have been assigned to a cluster. Then enter step 3 - 4;

[0105] 8.2 After steps 3 - 4 are completed, for the matrix T calculated from the expanded data, repeat steps 5 - 7, but do not execute the process of continuing to find the next corresponding point M i process, and only calculate the relevant merged cluster information about point M1. This cluster is denoted as target (the cluster set of the current electricity meter, and the set is larger); judge whether to merge the clusters according to the following conditions:

[0106] 8.21 If next belongs to the target subset, it is considered that the two clusters previous and next to be merged need to be merged;

[0107] 8.22 If next does not belong to the target subset, then do not merge, update the cluster information of the merged points, and continue to find the next corresponding point M i, go back to step 1.

[0108] It should be understood that the above specific embodiments of the present invention are only for illustrative explanation or interpretation of the principles of the present invention, and do not constitute a limitation on the present invention. Therefore, any modifications, equivalent replacements, improvements, etc. made without departing from the spirit and scope of the present invention shall be included within the protection scope of the present invention. In addition, the appended claims of the present invention are intended to cover all changes and modifications that fall within the scope and boundaries of the appended claims, or equivalent forms of such scope and boundaries. Obviously, the above embodiments are merely examples given for clear illustration and are not limitations on the implementation manners. For those of ordinary skill in the art, other different forms of changes or variations can be made based on the above description. It is not necessary and impossible to enumerate all the implementation manners here. And the obvious changes or variations derived therefrom are still within the protection scope of the present invention.

Claims

1. A method for identifying box meters in a transformer substation area, characterized in that, The station area includes N electric meters, and the method includes: S1, obtain the wireless RSSI data detected by the current meter and the neighboring meters in the HRF channel; S2, determining a first matrix according to the detection result of the wireless RSSI data, wherein the elements of the first matrix are the wireless RSSI data measured between the electric meters; S3, preprocessing the first matrix to obtain a second matrix, converting the second matrix into a scalar matrix, wherein the elements of the second matrix are the wireless RSSI data between the electricity meters after preprocessing and screening, and each electricity meter in the second matrix corresponds to a cluster; S4, performing clustering and merging operations on the scalar matrix using a hierarchical clustering algorithm to obtain a third matrix, wherein the elements of the third matrix include cluster numbers of the electric meters and distances between the electric meters, and each cluster before merging is numbered in the third matrix; S5, taking the point with cluster number 1 in the third matrix as the first point cluster, looping the third matrix to determine whether the target cluster can be matched, determining whether to merge the first point cluster with the target cluster according to the matching result, and then determining the recognition result of the substation box meter according to the cluster merging result. In the first loop, the cluster number of the current meter is 1, and the merged cluster is used as the first point cluster in subsequent loops.

2. The method for identifying the box meters in the substation area according to claim 1, wherein The method also includes: S6, if cluster merging cannot be performed, performing data expansion on the first matrix to obtain the cluster set of the current electricity meter, then judging whether the target cluster belongs to a subset of the cluster set, determining whether to perform cluster merging based on the judgment result, and then determining the recognition result of the substation box meter based on the cluster merging result.

3. The method for identifying the box meters in the substation area according to claim 1, wherein, The step S2 comprises: If no wireless RSSI data is detected, the current meter is formed into a separate cluster and returns to S1 to continue traversing the next meter; If wireless RSSI data is detected, it is determined whether the surrounding neighbor meters corresponding to the wireless RSSI data include allocated meters. If included, the allocated meters are removed, and the remaining unallocated meters are retained and combined with the current meters to form a first matrix.

4. The method for identifying the box meters in the substation area according to claim 1, wherein The step S3 comprises: If the wireless RSSI data except the diagonal in the first matrix is ​​the default value, the wireless RSSI data between the electric meter corresponding to the default value and the current electric meter are compared, and the electric meter corresponding to the wireless RSSI data with the smallest attenuation is selected to retain, and the other electric meters corresponding to the default value are eliminated, and the remaining electric meters form the second matrix.

5. The method for identifying the box meters in the substation area according to claim 1, characterized in that, In the step S5, determining whether a target cluster can be matched and determining whether to merge the first point cluster with the target cluster according to the matching result includes: If the target cluster is not matched, return to S1 and continue traversing to the next meter; If the target cluster is matched, but there are two or more electric meters between the first point cluster and the target cluster, and the wireless RSSI data of each other cannot be detected, then return to S1 and continue to traverse to the next electric meter; If the target cluster is matched, and the electric meters between the first point cluster and the target cluster can detect wireless RSSI data from each other, the first point cluster is merged with the target cluster, and the cluster information of the first point cluster is updated, and step S5 is repeated.

6. The method for identifying the box meters in the substation area according to claim 1, wherein The step S4 further includes: determining a merging threshold or an empirical threshold according to the merging relationship of the current electricity meter in the third matrix.

7. The method for identifying the box and meter in the substation area according to claim 6, characterized in that, The step S5 further includes: If the distance at which the first point cluster merges with the target cluster is less than the merging threshold, then merge the first point cluster with the target cluster, update the cluster class information of the first point cluster, and then repeat step S5; Or if there are more than two electricity meters between the first point cluster and the target cluster and the detected attenuation values between them are less than the empirical threshold, then merge the first point cluster with the target cluster, update the cluster class information of the first point cluster, and then repeat step S5.

8. The method for identifying the box and meter in the substation area according to claim 3, wherein In step S6, performing data expansion on the first matrix to obtain the cluster set of the current electricity meter includes: Adding the excluded and assigned electricity meters to the first matrix for expansion; Repeating steps S3 to S4 on the expanded first matrix, and the obtained third matrix is the cluster set of the current electricity meter.

9. The method for identifying the box meters in the substation area according to claim 8, wherein, In step S6, determining whether the target cluster belongs to a subset of the cluster set and determining whether to perform cluster merging according to the judgment result includes: If the target cluster belongs to a subset of the cluster set, then merge the first point cluster with the target cluster; If the target cluster does not belong to a subset of the cluster set, then update the cluster information of the third matrix and return to S1 to continue traversing to the next electricity meter.

10. A transformer substation box and meter identification system, characterized in that, The substation area includes N electricity meters, including: A detection module for obtaining the wireless RSSI data detected by the current electricity meter and surrounding neighbor electricity meters on the HRF channel; A first matrix determination module for determining a first matrix according to the detection result of the wireless RSSI data, and the elements of the first matrix are the wireless RSSI data measured between each electricity meter; A conversion module for preprocessing the first matrix to obtain a second matrix, and converting the second matrix into a scalar matrix. The elements of the second matrix are the wireless RSSI data between each electricity meter after preprocessing and screening. Each electricity meter in the second matrix corresponds to a cluster; A clustering operation module for performing a clustering and merging operation on the scalar matrix by using a hierarchical clustering algorithm to obtain a third matrix. The elements of the third matrix include the cluster numbers of each electricity meter and the distances between each electricity meter. Each cluster before merging in the third matrix is numbered; A merging module for taking the point with the cluster number 1 in the third matrix as the first point cluster, looping through the third matrix, determining whether a target cluster can be matched, determining whether to merge the first point cluster with the target cluster according to the matching result, and then determining the identification result of the substation area box meter according to the cluster merging result. The cluster number of the current electricity meter is 1 in the first loop, and the merged cluster is used as the first point cluster in subsequent loops.

Images