Power outage scope analysis method and system based on power outage correlation clustering clusters

Abstract

The present invention relates to the technical field of distribution network fault analysis, and discloses a power outage range analysis method and system based on a power outage correlation clustering cluster. The method performs clustering analysis on the power consumption data of each user through the K-means clustering algorithm, classifies the power consumption data with the highest similarity into the same cluster group, calculates the power outage correlation coefficient between pairwise power consumption data within the same cluster group, generates a correlation coefficient matrix, and determines corresponding characteristic users. Based on the power outage active alarm of the characteristic users, the polling technology is used at multiple levels to poll the power-on and power-off states of the transformer substation area, the feeder layer to which it is connected, and other users in the cluster group to which it belongs, so as to determine the power outage range, and can effectively improve the accuracy of power outage range positioning in the low-voltage transformer substation area.

Description

Technical Field

[0001] The present invention relates to the technical field of distribution network fault analysis, and in particular to a power outage range analysis method and system based on a power outage correlation clustering cluster. Background Art

[0002] At present, the informatization level on the user side of the distribution network substation area is low and the management is chaotic. There are still deficiencies in the power user service level. Problems such as inaccurate user power outage information, inability to achieve active warning, and long power supply restoration time are common. When a fault occurs on the low-voltage side, due to the lack or inaccuracy of the low-voltage wiring information in the substation area, power outage faults in the low-voltage substation area mainly rely on users reporting faults by phone and repair personnel checking household by household, making it difficult to achieve active repair on the low-voltage side and affecting the user's power consumption service experience. At the same time, the current technical method of installing a large number of monitoring devices on the low-voltage side to achieve power outage warning has problems of large upfront investment and high later maintenance costs, making it difficult to promote.

[0003] For a long time, power outage repair and handling have been passive, and information such as the power outage location and scale is not accurately known. There is even a situation where users do not discover a power outage at home for a long time, resulting in the continuation of the power outage fault, causing greater damage to users' property and equipment, and thus triggering problems such as user complaints or requests for compensation from the power supply company.

[0004] In the field of power operation and maintenance, there has been research on power outage analysis by dividing the smallest power outage common analysis unit. It receives terminal and meter events, calls the device power consumption to judge whether the device has false alarms; judges whether it is a substation area-level terminal power outage; after excluding the substation area-level power outage, it judges the smallest power outage common analysis unit based on the power outage events of all terminals or meters under the substation area, and conducts power outage research and judgment on the smallest power outage common unit; if it is not a power outage of the smallest power outage common unit, it judges from the next lower level of the smallest power outage common unit. Finally, it confirms the hierarchical power outage and power outage status related to the power outage event, and finally obtains the smallest power outage common unit where the power outage is confirmed.

[0005] Among them, for the research on user power outage correlation, existing technologies currently analyze the power outage range of the distribution network through connection relationships, and quickly locate the power outage range according to the connection relationships of distribution network equipment through the system, and then query the marketing system database to obtain the affected range of the power outage.

[0006] However, in the analysis and judgment of the power outage range, it is necessary to locate through the connection relationship or topological structure of the distribution network. However, the informatization level on the user side of the distribution network substation area is low and the management is chaotic, and the low-voltage wiring information is missing or inaccurate. Therefore, there are large errors in locating the power outage range through the connection relationship of distribution network equipment. Summary of the Invention

[0007] The present invention provides a method and system for analyzing power outage range based on power outage correlation clustering, which solves the technical problem of large power outage range positioning error in the prior art.

[0008] In view of this, a first aspect of the present invention provides a method for analyzing power outage scope based on power outage correlation clustering, comprising the following steps:

[0009] S1. Obtaining the electricity consumption data of each user in the area within a preset period of time based on the user-side meter;

[0010] S2. Use K-means clustering algorithm to perform cluster analysis on the electricity consumption data of each user, so as to classify the electricity consumption data with the highest similarity into the same cluster group, and obtain several cluster groups;

[0011] S3, calculating the power outage correlation coefficient between two power consumption data in the same cluster group, generating a correlation coefficient matrix according to the power outage correlation coefficient, and selecting users with the highest power outage correlation coefficient as feature users according to the correlation coefficient matrix;

[0012] S4. Based on the active power outage alarm signal sent in advance by the characteristic user, a polling technique is used to poll the power outage status of the substation where the characteristic user is located, the feeder layer to which it is connected, and other users in the cluster group where it is located, so as to determine the scope of the power outage.

[0013] Preferably, the method further comprises:

[0014] The power consumption data of each user in the substation is frozen according to the preset freezing time, and the frozen power consumption data is sent to the metering center.

[0015] Preferably, step S2 specifically includes:

[0016] S201, the power consumption data is combined into a data set X = (x 1 , x 2 ,…,x n ), in the data set X=(x 1 , x 2 ,…,x n ) randomly selects K samples as the initial cluster center x = (x 1 , x 2 ,…,x k ), define the set of cluster groups corresponding to each cluster center in the initial cluster center as S = {S 1 ,S 2 ,…,S k};

[0017] S202, the shortest Euclidean distance between each sample in the data set and the initial cluster center is calculated by the following formula 1:

[0018]

[0019] In Equation 1, D(x) represents the shortest Euclidean distance, and x i represents the i-th sample in the dataset, and μ i represents the cluster center. Among them, the cluster center μ i is represented as

[0020]

[0021] S203. Calculate the selection probability P(x) of each sample being selected as the next cluster center through the following Equation 3 based on each sample and the shortest Euclidean distance from the initial cluster center, which is

[0022]

[0023] S204. Select samples as the next cluster center according to the roulette method based on the selection probability of each sample, and repeat this step until k next cluster centers are selected;

[0024] S205. Calculate the shortest Euclidean distance between each sample in the dataset and the k next cluster centers through Equation 1, and assign each sample to the cluster group corresponding to the cluster center with the smallest distance.

[0025] Preferably, step S3 specifically includes:

[0026] S301. Generate an electricity consumption data matrix based on the pairwise electricity consumption data within the same cluster group, denoted as M. M is an m×b-dimensional matrix, where m is the number of types of electricity consumption data, and b is the number of users within the same cluster group;

[0027] S302. Calculate the power outage correlation coefficient between the pairwise electricity consumption data within the same cluster group through the correlation coefficient calculation formula of Equation 4, which is

[0028]

[0029] In Equation 4, ρ represents the power outage correlation coefficient, a t represents the electricity consumption data matrix of meter a at time t, represents the average value of meter a over the entire time period, T represents the total time period, g t represents the electricity consumption data matrix of meter g at time t, represents the average value of meter g over the entire time period, L aa represents the sum of squared mean differences of the electricity consumption data matrix of meter a at time t, L gg represents the sum of squared mean differences of the electricity consumption data matrix of meter g at time t, and cov(a,g) represents the covariance between meter a and meter g. Among them,

[0030]

[0031]

[0032]

[0033] S303, generating a corresponding correlation coefficient matrix according to the power outage correlation coefficients of the same cluster group;

[0034] S304: Select users with the highest power outage correlation coefficient as feature users according to the correlation coefficient matrix.

[0035] Preferably, step S4 specifically includes:

[0036] S401, when a power outage signal is detected by the alarm meter pre-installed by the characteristic user, an active power outage alarm signal is sent to the corresponding concentrator, and the concentrator reports the signal to the master station;

[0037] S402, obtaining the number of active power outage alarm signals received by the master station from the same substation within the same time margin, if the number of active power outage alarm signals is equal to 1, executing step S405; if the number of active power outage alarm signals is greater than 1, executing steps S403 to S405;

[0038] S403, polling the alarm meters at all loads directly connected to the low-voltage side outlet of the transformer in the substation where the characteristic user is located to obtain the power consumption data at all loads, thereby determining the power outage state of the entire substation, wherein the power outage state includes a normal power state and a power outage state;

[0039] S404, determining all feeder nodes to which the characteristic user is connected according to the feeder topology connection relationship, and polling the alarm meter of each feeder node to obtain the power consumption data of all feeder nodes, thereby determining the power outage status of each feeder node;

[0040] S405, polling the remaining other users in the cluster group where the characteristic user is located in order according to the power outage correlation coefficient with the characteristic user to obtain the power consumption data of each remaining other user, thereby determining the power outage status of each user in the cluster group corresponding to the same time, and executing step S406;

[0041] S406: Determine the power outage scope according to the power outage status of the substation corresponding to the characteristic user, the power outage status of each feeder node, and the power outage status of each user in the cluster group.

[0042] In a second aspect, the present invention further provides a power outage range analysis system based on power outage correlation clustering, comprising:

[0043] The power consumption acquisition module is used to acquire the power consumption data of each user in the area within a preset time period based on the user-side meter;

[0044] The clustering module is used to perform clustering analysis on the power consumption data of each user by using the K-means clustering algorithm, so as to classify the power consumption data with the highest similarity into the same cluster group and obtain several cluster groups;

[0045] The correlation calculation module is used to calculate the power outage correlation coefficient between pairwise power consumption data within the same cluster group, generate a correlation coefficient matrix according to the power outage correlation coefficient, and select the user with the highest power outage correlation coefficient as the characteristic user according to the correlation coefficient matrix;

[0046] The power outage analysis module is used to perform polling on the power-on / off status of the area where the characteristic user is located, the feeder layer to which it is connected, and other users in the cluster group to which it belongs based on the power outage active alarm signal pre-sent by the characteristic user by using the polling technology, so as to determine the power outage range.

[0047] Preferably, the system further includes:

[0048] The freezing module is used to freeze the power consumption data of each user in the area according to the preset freezing time and send the frozen power consumption data to the metering center.

[0049] Preferably, the clustering module specifically includes:

[0050] The initial module is used to form the power consumption data into a data set X=(x 1 , x 2 ,…, x n ), randomly select K samples in the data set X=(x 1 , x 2 ,…, x n ) as the initial clustering centers x=(x 1 , x 2 ,…, x k ), and define the set of cluster groups corresponding to each clustering center in the initial clustering centers as S={S 1 , S 2 ,…, S k};

[0051] The distance calculation module is used to calculate the shortest Euclidean distance between each sample in the data set and the initial clustering center through the following formula 1,

[0052]

[0053] In formula 1, D(x) represents the shortest Euclidean distance, x iDenote the i-th sample in the dataset, μ i Denote the cluster center, where the cluster center μ i Is expressed as

[0054]

[0055] A probability calculation module, which is used to calculate the selection probability P(x) of each sample being selected as the next cluster center through the following formula 3 according to each sample and the shortest Euclidean distance from the initial cluster center, as follows

[0056]

[0057] A cluster update module, which is used to select samples as the next cluster center according to the selection probability of each sample by the roulette method, so as to select k next cluster centers;

[0058] A clustering module, which is used to calculate the shortest Euclidean distance between each sample in the dataset and the k next cluster centers through formula 1, and assign each sample to the cluster group corresponding to the cluster center with the smallest distance.

[0059] Preferably, the correlation calculation module specifically includes:

[0060] A matrix generation module, which is used to generate an electricity consumption data matrix based on the pairwise electricity consumption data within the same cluster group, denoted as M. M is an m×b-dimensional matrix, where m is the number of types of electricity consumption data, and b is the number of users within the same cluster group;

[0061] A correlation coefficient calculation module, which is used to calculate the outage correlation coefficient between the pairwise electricity consumption data within the same cluster group through the correlation coefficient calculation formula of formula 4, as follows

[0062]

[0063] In formula 4, ρ represents the outage correlation coefficient, a t Represents the electricity consumption data matrix of meter a at time t, Represents the average value of meter a over the entire time period, T represents the total time period, g t Represents the electricity consumption data matrix of meter g at time t, Represents the average value of meter g over the entire time period, L aa Represents the sum of squared deviations of the electricity consumption data matrix of meter a at time t, L gg Represents the sum of squared deviations of the electricity consumption data matrix of meter g at time t, cov(a,g) represents the covariance between meter a and meter g, where

[0064]

[0065]

[0066]

[0067] A matrix module, configured to generate a corresponding correlation coefficient matrix according to the power outage correlation coefficients of the same cluster group;

[0068] A characteristic user determination module, configured to select the user with the highest power outage correlation coefficient as the characteristic user according to the correlation coefficient matrix.

[0069] Preferably, the power outage analysis module specifically includes:

[0070] An alarm module, configured to send an active power outage alarm signal to the concentrator to which it belongs when a power outage signal is detected by an alarm electric meter pre-installed on the characteristic user, and report it to the master station through the concentrator;

[0071] A judgment module, configured to obtain the number of active power outage alarm signals received by the master station from the same distribution transformer area within the same time margin;

[0072] A distribution transformer area polling module, configured to poll the alarm electric meters at all loads directly connected to the low-voltage side outlet of the transformer in the distribution transformer area where the characteristic user is located, so as to obtain the power consumption data of all loads, thereby determining the power-on and power-off states of the entire distribution transformer area, where the power-on and power-off states include the normal power-on state and the power-off state;

[0073] A feeder polling module, configured to determine all feeder nodes connected by the characteristic user according to the feeder topology connection relationship, poll the alarm electric meters of each feeder node, so as to obtain the power consumption data of all feeder nodes, thereby determining the power-on and power-off states of each feeder node;

[0074] A cluster group polling module, configured to poll the remaining other users in the cluster group where the characteristic user is located in sequence according to the magnitude of the power outage correlation coefficient with the characteristic user, so as to obtain the power consumption data of each remaining other user, thereby determining the power-on and power-off states of each user in the corresponding cluster group at the same moment;

[0075] A power outage range determination module, configured to determine the power outage range according to the power-on and power-off states of the distribution transformer area corresponding to the characteristic user, the power-on and power-off states of each feeder node, and the power-on and power-off states of each user in the cluster group.

[0076] It can be seen from the above technical solutions that the present invention has the following advantages:

[0077] The present invention performs clustering analysis on the power consumption data of each user through the K-means clustering algorithm, classifies the power consumption data with the highest similarity into the same cluster group, calculates the power outage correlation coefficient between pairwise power consumption data within the same cluster group, generates a correlation coefficient matrix, and determines the corresponding characteristic users. Based on the proactive power outage alarm of the characteristic users, the polling technology is used at multiple levels to poll the power outage and power-on states of the low-voltage station area, the feeder layer to which it is connected, and other users in the cluster group to which it belongs, so as to determine the power outage range, and can effectively improve the accuracy of power outage range positioning in the low-voltage station area. BRIEF DESCRIPTION OF THE DRAWINGS

[0078] Figure 1 is a flowchart of a power outage range analysis method based on a power outage correlation clustering cluster provided by an embodiment of the present invention;

[0079] Figure 2 is a schematic structural diagram of a power outage range analysis system based on a power outage correlation clustering cluster provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0080] In order to enable those skilled in the art to better understand the solution of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0081] For ease of understanding, please refer to Figure 1 , a power outage range analysis method based on a power outage correlation clustering cluster provided by the present invention includes the following steps:

[0082] S1. Obtain the power consumption data of each user in the station area within a preset time period based on the user-side meters.

[0083] S2. Perform clustering analysis on the power consumption data of each user by using the K-means clustering algorithm, so as to classify the power consumption data with the highest similarity into the same cluster group, and obtain several cluster groups.

[0084] S3. Calculate the power outage correlation coefficient between pairwise power consumption data within the same cluster group, generate a correlation coefficient matrix according to the power outage correlation coefficient, and select the user with the highest power outage correlation coefficient as the characteristic user according to the correlation coefficient matrix.

[0085] S4. Based on the proactive power outage alarm signal pre-sent by the characteristic user, use the polling technology to poll the power outage and power-on states of the station area where the characteristic user is located, the feeder layer to which it is connected, and other users in the cluster group to which it belongs, so as to determine the power outage range.

[0086] This embodiment provides a power outage scope analysis method based on power outage correlation clusters. By using the K-means clustering algorithm to perform clustering analysis on the power consumption data of each user, the power consumption data with the highest similarity is classified into the same cluster group. Calculate the power outage correlation coefficient between pairwise power consumption data within the same cluster group, generate a correlation coefficient matrix, and determine the corresponding characteristic users. Based on the power outage active alarm of the characteristic users, the polling technology is used at multiple levels to poll the power-on and power-off states of the substation area, the feeder layer to which it is connected, and other users in the cluster group to which it belongs, so as to determine the power outage scope, which can effectively improve the accuracy of power outage scope positioning in low-voltage substation areas.

[0087] In a specific embodiment, this method further includes:

[0088] Freeze the power consumption data of each user in the substation area according to the preset freezing time, and send the frozen power consumption data to the metering center.

[0089] In an example, the power consumption data can freeze the user power data at fixed points on a daily basis, and at the same time upload all the frozen power data of the day to the metering center. Since some communication, statistical and other errors may occur during the data uploading process, which may cause deviations in the original data. Therefore, when extracting data through the "operation-distribution" business data platform, the original data needs to be preprocessed. The preprocessing mainly includes the cleaning of error data, the repair of abnormal data, and data denoising.

[0090] When comparing the similarity between users, the data selected by the two users needs to be in the same time period. If the power data of one of the users is not successfully collected at a certain event point, the data of this event point of these two users should be deleted, and the correlation coefficient is calculated for the remaining data.

[0091] In a specific embodiment, step S2 specifically includes:

[0092] S201. Compose the power consumption data into a data set X = (x 1 , x 2 , …, x n ). Randomly select K samples in the data set X = (x 1 , x 2 , …, x n ) as the initial clustering centers x = (x 1 , x 2 , …, x k ). Define the set of cluster groups corresponding to each clustering center in the initial clustering centers as S = {S 1 , S 2 , …, S k};

[0093] S202. Calculate the shortest Euclidean distance between each sample in the dataset and the initial cluster center through the following formula 1 as

[0094]

[0095] In formula 1, D(x) represents the shortest Euclidean distance, and x i represents the i-th sample in the dataset, and μ i represents the cluster center. Among them, the cluster center μ i is represented as

[0096]

[0097] S203. Calculate the selection probability P(x) of each sample being selected as the next cluster center through the following formula 3 according to each sample and the shortest Euclidean distance from the initial cluster center as

[0098]

[0099] S204. Select samples as the next cluster center according to the selection probability of each sample by the roulette method, and repeat this step until k next cluster centers are selected;

[0100] S205. Calculate the shortest Euclidean distance between each sample in the dataset and the k next cluster centers through formula 1, and assign each sample to the cluster group corresponding to the cluster center with the smallest distance.

[0101] It can be understood that due to the large number of users in the power distribution area, the power consumption data is constantly updated over time and grows rapidly. It is inconvenient to calculate the power outage correlation coefficient for all users. Dividing users into several clustering clusters and conducting analysis and research within the clustering clusters is more convenient.

[0102] In a specific embodiment, step S3 specifically includes:

[0103] S301. Generate a power consumption data matrix based on the pairwise power consumption data within the same cluster group, denoted as M. M is an m×b-dimensional matrix, where m is the number of types of power consumption data and b is the number of users within the same cluster group;

[0104] S302. Calculate the power outage correlation coefficient between the pairwise power consumption data within the same cluster group through the correlation coefficient calculation formula of the following formula 4 as

[0105]

[0106] In formula 4, ρ represents the power outage correlation coefficient, and a t represents the power consumption data matrix of meter a at time t represents the average value of meter a during the whole period, T represents the total period, g t represents the electricity consumption data matrix of meter g in period t, It represents the average value of the meter g during the whole period, L aa represents the mean square sum of the electricity consumption data matrix of meter a in period t, L gg represents the mean square sum of the electricity consumption data matrix of meter g in period t, cov(a,g) represents the covariance of meter a and meter g, where,

[0107]

[0108]

[0109]

[0110] S303, generating a corresponding correlation coefficient matrix according to the power outage correlation coefficient of the same cluster group;

[0111] S304. Select users with the highest power outage correlation coefficient as feature users according to the correlation coefficient matrix.

[0112] In one example, based on the study of the power outage correlation coefficient between users in the same cluster, the correlation level between the corresponding attribute and the power outage can be given. The correlation coefficient is between negative 1 and positive 1. The larger the absolute value of the correlation coefficient, the higher the power outage correlation between the users represented. Therefore, the interval is evenly divided into 8 relationship levels, namely 4 levels of negative correlation and 4 levels of positive correlation.

[0113] In a specific embodiment, step S4 specifically includes:

[0114] S401. When a power outage signal is detected by an alarm meter pre-installed by a characteristic user, an active power outage alarm signal is sent to the corresponding concentrator, and the signal is reported to the master station through the concentrator.

[0115] The alarm meter is a smart meter with a supercapacitor, which has the function of detecting power outages. When a power outage occurs, the meter can send an alarm signal to the concentrator, which reports to the main station through the concentrator, thereby realizing active alarm based on characteristic users.

[0116] S402. Obtain the number of active power outage alarm signals received by the master station from the same substation within the same time margin. If the number of active power outage alarm signals is equal to 1, execute step S405; if the number of active power outage alarm signals is greater than 1, execute steps S403 to S405.

[0117] In this embodiment, the alarm situations are classified into single - user alarms and multi - user alarms. If the number of power - outage active alarm signals is equal to 1, the alarm signal can be classified as a single - user alarm; if multiple alarm signals appear simultaneously, that is, the alarm signal is classified as a multi - user alarm. When it is determined to be a multi - user alarm, the polling priority can be formed by combining the calculated power - outage correlation coefficient and the positional relationship of the clustering distribution, and polling is carried out hierarchically in turn.

[0118] S403. Poll the alarm meters at all loads directly connected to the low - voltage side outlet of the transformer in the substation area where the characteristic user is located to obtain the power consumption data at all loads, so as to determine the power - on / off state of the entire substation area. The power - on / off state includes the normal power - on state and the power - outage state.

[0119] Among them, intelligent meters with supercapacitors are installed at all loads directly connected to the low - voltage side outlet of the transformer. Since such nodes are located at the head end of the substation area, if such nodes alarm, it can be judged whether it is a substation - level power outage, so as to realize the layout of the active power - outage alarm device at the substation area level.

[0120] S404. Determine all feeder nodes connected by the characteristic user according to the feeder topology connection relationship, and poll the alarm meters of each feeder node to obtain the power consumption data of each feeder node, so as to determine the power - on / off state of each feeder node.

[0121] In the process of power - outage scope analysis, it can be considered at the branch level downward. At the branch level, fully consider the front - and - back connection relationship of each branch in each feeder, and select the head - end nodes of each feeder and the head - end nodes of each branch to install intelligent meters with supercapacitors as well, so as to realize the layout of the active power - outage alarm device at the branch level.

[0122] S405. Poll the remaining other users in the cluster group where the characteristic user is located in turn according to the magnitude of the power - outage correlation coefficient with the characteristic user to obtain the power consumption data of each remaining other user, so as to determine the power - on / off state of each user in the corresponding cluster group at the same moment, and execute step S406;

[0123] After considering the substation area and feeder connection nodes, install intelligent meters with supercapacitors for the remaining characteristic users in the cluster group to realize the active alarm of the power - outage state at the local user level. At the same time, poll all the remaining users in the same cluster. The polling order depends on the magnitude of the power - outage correlation coefficient between the user and the characteristic user. Users with a higher correlation coefficient value are polled first to determine the power - on / off state of each user in the cluster at this moment.

[0124] S406. Determine the power - outage scope according to the power - on / off state of the substation area corresponding to the characteristic user, the power - on / off state of each feeder node, and the power - on / off state of each user in the cluster group.

[0125] It should be noted that, except for excluding the possibility of planned power outages, based on polling technology, polling plans for different alarm situations are set, a multi-level polling sampling scheme is adopted and combined with the power-on and power-off states of characteristic users to analyze the power-on and power-off states of users within the alarm cluster, gradually narrowing down and finally determining the power outage scope, thereby improving the accuracy of power outage scope positioning.

[0126] The above is a detailed description of an embodiment of a power outage scope analysis method based on a power outage correlation clustering cluster provided by the present invention. The following is a detailed description of an embodiment of a power outage scope analysis system based on a power outage correlation clustering cluster provided by the present invention.

[0127] For ease of understanding, please refer to Figure 2 , the present invention provides a power outage scope analysis system based on a power outage correlation clustering cluster, including:

[0128] An electricity consumption acquisition module 100, configured to obtain electricity consumption data of each user in a preset time period within a power distribution area based on a user-side meter;

[0129] A clustering module 200, configured to perform clustering analysis on the electricity consumption data of each user by using a K-means clustering algorithm, so as to classify the electricity consumption data with the highest similarity into the same cluster group, and obtain a plurality of cluster groups;

[0130] A correlation calculation module 300, configured to calculate a power outage correlation coefficient between pairwise electricity consumption data within the same cluster group, generate a correlation coefficient matrix according to the power outage correlation coefficient, and select a user with the highest power outage correlation coefficient as a characteristic user according to the correlation coefficient matrix;

[0131] A power outage analysis module 400, configured to perform polling on the power-on and power-off states of the power distribution area where the characteristic user is located, the feeder layer to which it is connected, and other users in the cluster group to which it belongs based on a power outage active alarm signal pre-sent by the characteristic user by using polling technology, so as to determine the power outage scope.

[0132] In a specific embodiment, the system further includes:

[0133] A freezing module, configured to freeze the electricity consumption data of each user in the power distribution area according to a preset freezing time, and send the frozen electricity consumption data to a metering center.

[0134] In a specific embodiment, the clustering module specifically includes:

[0135] An initial module, configured to form a data set X=(x 1 , x 2 ,…, x n ) from the electricity consumption data. In the data set X=(x 1, x 2 , …, x n ) randomly select K samples as the initial clustering centers x = (x 1 , x 2 , …, x k ), and define the set of cluster groups corresponding to each clustering center in the initial clustering centers as S = {S 1 , S 2 , …, S k};

[0136] The distance calculation module is used to calculate the shortest Euclidean distance between each sample in the dataset and the initial clustering centers through the following formula 1 as

[0137]

[0138] In formula 1, D(x) represents the shortest Euclidean distance, x i represents the i-th sample in the dataset, μ i represents the clustering center, where the clustering center μ i is represented as

[0139]

[0140] The probability calculation module is used to calculate the selection probability P(x) of each sample being selected as the next clustering center through the following formula 3 according to each sample and the shortest Euclidean distance from the initial clustering center as

[0141]

[0142] The clustering update module is used to select samples as the next clustering centers according to the selection probability of each sample by the roulette method, so as to select k next clustering centers;

[0143] The clustering module is used to calculate the shortest Euclidean distance between each sample in the dataset and the k next clustering centers through formula 1, and allocate each sample to the cluster group corresponding to the clustering center with the smallest distance.

[0144] In a specific embodiment, the relevant calculation module specifically includes:

[0145] The matrix generation module is used to generate an electricity consumption data matrix based on the pairwise electricity consumption data within the same cluster group, denoted as M, where M is an m×b-dimensional matrix, m is the number of types of electricity consumption data, and b is the number of users within the same cluster group;

[0146] The correlation coefficient calculation module is used to calculate the power outage correlation coefficient between the pairwise electricity consumption data within the same cluster group through the correlation coefficient calculation formula of the following formula 4 as

[0147]

[0148] In formula 4, ρ represents the power outage correlation coefficient, a t represents the electricity consumption data matrix of meter a in period t, represents the average value of meter a during the whole period, T represents the total period, g t represents the electricity consumption data matrix of meter g in period t, It represents the average value of the meter g during the whole period, L aa represents the mean square sum of the electricity consumption data matrix of meter a in period t, L gg represents the mean square sum of the electricity consumption data matrix of meter g in period t, cov(a,g) represents the covariance of meter a and meter g, where,

[0149]

[0150]

[0151]

[0152] A matrix module, used for generating a corresponding correlation coefficient matrix according to the power outage correlation coefficients of the same cluster group;

[0153] The characteristic user determination module is used to select the user with the highest power outage correlation coefficient as the characteristic user according to the correlation coefficient matrix.

[0154] In a specific embodiment, the power outage analysis module specifically includes:

[0155] The alarm module is used to send a power outage active alarm signal to the concentrator when a power outage signal is detected by the alarm meter pre-installed by the characteristic user, and report it to the main station through the concentrator;

[0156] A judgment module is used to obtain the number of active power outage alarm signals received by the master station from the same substation within the same time margin;

[0157] The area polling module is used to poll the alarm meters at all loads directly connected to the low-voltage side outlet of the transformer in the area where the characteristic user is located, so as to obtain the power consumption data at all loads, thereby determining the power outage status of the entire area, which includes the normal power consumption status and the power outage status;

[0158] A feeder polling module is used to determine all feeder nodes connected to the characteristic user according to the feeder topology connection relationship, and poll the alarm meter of each feeder node to obtain the power consumption data of all feeder nodes, so as to determine the power outage status of each feeder node;

[0159] The cluster group polling module is used to poll the remaining other users in the cluster group where the characteristic user is located in sequence according to the magnitude of the power outage correlation coefficient with the characteristic user, so as to obtain the power consumption data of each remaining other user, and thus determine the power outage and power-on states of each user in the corresponding cluster group at the same moment;

[0160] The power outage range determination module is used to determine the power outage range according to the power outage and power-on states of the substation area corresponding to the characteristic user, the power outage and power-on states of each feeder node, and the power outage and power-on states of each user in the cluster group.

[0161] Those skilled in the art can clearly understand that for the convenience and conciseness of description, the specific working processes of the above-described system, device, and unit can refer to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0162] In several embodiments provided by the present invention, it should be understood that the disclosed device and method can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed mutual coupling or direct coupling or communication connection can be through some interfaces, and the indirect coupling or communication connection of the device or unit can be in an electrical, mechanical, or other form.

[0163] The unit described as a separated component may or may not be physically separated, and the component displayed as a unit may or may not be a physical unit, that is, it may be located in one place, or may be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0164] In addition, each functional unit in various embodiments of the present invention can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of a software functional unit.

[0165] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. Power outage scope analysis method based on power outage correlation clustering, It is characterized in that The following steps are involved: S1. Obtaining the electricity consumption data of each user in the area within a preset period of time based on the user-side meter; S2. Use K-means clustering algorithm to perform cluster analysis on the electricity consumption data of each user, so as to classify the electricity consumption data with the highest similarity into the same cluster group, and obtain several cluster groups; S3, calculating the power outage correlation coefficient between two power consumption data in the same cluster group, generating a correlation coefficient matrix according to the power outage correlation coefficient, and selecting users with the highest power outage correlation coefficient as feature users according to the correlation coefficient matrix, specifically including: S301, generating an electricity data matrix based on the two-by-two electricity consumption data in the same cluster group, denoted as M, where M is an m×b-dimensional matrix, m is the number of types of electricity data, and b is the number of users in the same cluster group; S302, the power outage correlation coefficient between two power consumption data in the same cluster group is calculated by the correlation coefficient calculation formula of the following formula 4: In Equation 4, ρ represents the power outage correlation coefficient, and a t represents the power consumption data matrix of meter a in the t period, represents the average value of meter a over the entire period, T represents the total period, and g t represents the power consumption data matrix of meter g in the t period, represents the average value of meter g over the entire period, and L aa represents the sum of squared deviations of the power consumption data matrix of meter a in the t period, and L gg represents the sum of squared deviations of the power consumption data matrix of meter g in the t period, and cov(a,g) represents the covariance between meter a and meter g, where S303, generating a corresponding correlation coefficient matrix according to the power outage correlation coefficients of the same cluster group; S304, selecting a user with the highest power outage correlation coefficient as a characteristic user according to the correlation coefficient matrix; S4, based on the active power outage alarm signal sent in advance by the characteristic user, a polling technique is used to poll the power outage status of the station area where the characteristic user is located, the feeder layer connected to it, and other users in the cluster group where it is located, so as to determine the power outage scope, specifically including: S401, when a power outage signal is detected by the alarm meter pre-installed by the characteristic user, an active power outage alarm signal is sent to the corresponding concentrator, and the concentrator reports the signal to the master station; S402, obtaining the number of active power outage alarm signals received by the master station from the same substation within the same time margin, if the number of active power outage alarm signals is equal to 1, executing step S405; if the number of active power outage alarm signals is greater than 1, executing steps S403 to S405; S403, polling the alarm meters at all loads directly connected to the low-voltage side outlet of the transformer in the substation where the characteristic user is located to obtain the power consumption data at all loads, thereby determining the power outage state of the entire substation, wherein the power outage state includes a normal power state and a power outage state; S404, determining all feeder nodes to which the characteristic user is connected according to the feeder topology connection relationship, and polling the alarm meter of each feeder node to obtain the power consumption data of all feeder nodes, thereby determining the power outage status of each feeder node; S405, polling the remaining other users in the cluster group where the characteristic user is located in order according to the power outage correlation coefficient with the characteristic user to obtain the power consumption data of each remaining other user, thereby determining the power outage status of each user in the cluster group corresponding to the same time, and executing step S406; S406: Determine the power outage scope according to the power outage status of the substation corresponding to the characteristic user, the power outage status of each feeder node, and the power outage status of each user in the cluster group.

2. The method for analyzing power outage scope based on power outage correlation clustering according to claim 1, It is characterized in that Also includes: The power consumption data of each user in the substation is frozen according to the preset freezing time, and the frozen power consumption data is sent to the metering center.

3. The method for analyzing power outage scope based on power outage correlation clustering according to claim 1, It is characterized in that Step S2 specifically includes: S201. Compose the electricity consumption data into a data set \(X=(x 1 , x 2 , \ldots, x n )\). Randomly select \(K\) samples from the data set \(X=(x 1 , x 2 , \ldots, x n )\) as the initial cluster centers \(x=(x 1 , x 2 , \ldots, x k )\). Define the set of cluster groups corresponding to each cluster center in the initial cluster centers as \(S = \{S 1 , S 2 , \ldots, S k \}\); S202, the shortest Euclidean distance between each sample in the data set and the initial cluster center is calculated by the following formula 1: In Equation 1, D(x) represents the shortest Euclidean distance, and x i represents the i-th sample in the dataset, and μ i represents the cluster center. Among them, the cluster center μ i is expressed as S203, according to the shortest Euclidean distance between each sample and the initial cluster center, the selection probability P(x) of each sample being selected as the next cluster center is calculated by the following formula 3: S204, selecting a sample as the next cluster center according to the selection probability of each sample using the roulette wheel method, and repeating this step until k next cluster centers are selected; S205 , calculating the shortest Euclidean distance between each sample in the data set and the k next cluster centers by using Formula 1, and assigning each sample to the cluster group corresponding to the cluster center with the shortest distance.

4. Power outage range analysis system based on power outage correlation clustering, It is characterized in that include: The power consumption acquisition module is used to obtain the power consumption data of each user in the substation area within a preset period of time based on the user-side meter; The clustering module is used to perform cluster analysis on the power consumption data of each user using the K-means clustering algorithm, so as to classify the power consumption data with the highest similarity into the same cluster group, and obtain several cluster groups; A correlation calculation module is used to calculate the power outage correlation coefficient between two power consumption data in the same cluster group, generate a correlation coefficient matrix according to the power outage correlation coefficient, and select the user with the highest power outage correlation coefficient as the characteristic user according to the correlation coefficient matrix; A power outage analysis module is used to poll the power outage status of the substation where the characteristic user is located, the feeder layer connected to the characteristic user, and other users in the cluster group where the characteristic user is located based on the power outage active alarm signal sent in advance by the characteristic user, so as to determine the power outage scope; The relevant calculation modules specifically include: A matrix generation module is used to generate an electricity data matrix based on the pairwise electricity consumption data in the same cluster group, denoted as M, where M is an m×b dimensional matrix, m is the number of types of electricity data, and b is the number of users in the same cluster group; The correlation coefficient calculation module is used to calculate the power outage correlation coefficient between two power consumption data in the same cluster group through the correlation coefficient calculation formula of the following formula 4: In Equation 4, ρ represents the power outage correlation coefficient, and a t represents the power consumption data matrix of meter a at time period t, represents the average value of meter a over the entire time period, T represents the total time period, and g t represents the power consumption data matrix of meter g at time period t, represents the average value of meter g over the entire time period, and L aa represents the sum of squared deviations of the power consumption data matrix of meter a at time period t, and L gg represents the sum of squared deviations of the power consumption data matrix of meter g at time period t, and cov(a,g) represents the covariance between meter a and meter g, where A matrix module, used for generating a corresponding correlation coefficient matrix according to the power outage correlation coefficients of the same cluster group; A characteristic user determination module, used for selecting a user with the highest power outage correlation coefficient as a characteristic user according to the correlation coefficient matrix; The power outage analysis module specifically includes: An alarm module is used to send a power outage active alarm signal to the concentrator when a power outage signal is detected by the alarm meter pre-installed by the characteristic user, and report to the main station through the concentrator; A judgment module, configured to obtain the number of power outage active alarm signals received by the master station from the same distribution transformer area within the same time margin; A distribution transformer area polling module, configured to poll the alarm electric meters at all loads directly connected to the low-voltage side outlet of the transformer in the distribution transformer area where the characteristic user is located, so as to obtain the power consumption data at all loads, thereby determining the power-on and power-off states of the entire distribution transformer area, where the power-on and power-off states include a normal power-on state and a power-off state; A feeder polling module, configured to determine all feeder nodes connected by the characteristic user according to the feeder topology connection relationship, poll the alarm electric meters of each feeder node, so as to obtain the power consumption data of all feeder nodes, thereby determining the power-on and power-off states of each feeder node; A cluster group polling module, configured to poll the remaining other users in the cluster group where the characteristic user is located in sequence according to the magnitude of the power outage correlation coefficient with the characteristic user, so as to obtain the power consumption data of each remaining other user, thereby determining the power-on and power-off states of each user in the corresponding cluster group at the same moment; A power outage range determination module, configured to determine the power outage range according to the power-on and power-off states of the distribution transformer area corresponding to the characteristic user, the power-on and power-off states of each feeder node, and the power-on and power-off states of each user in the cluster group.

5. The power outage range analysis system based on a power outage correlation clustering cluster according to claim 4, wherein, it further includes: A freezing module, configured to freeze the power consumption data of each user in the distribution transformer area according to a preset freezing time, and send the frozen power consumption data to the metering center.

6. The power outage range analysis system based on a power outage correlation clustering cluster according to claim 4, wherein, the clustering module specifically includes: Initial module, used to form a data set X = (x 1 , x 2 , …, x n ) from the electricity consumption data. Randomly select K samples from the data set X = (x 1 , x 2 , …, x n ) as the initial cluster centers x = (x 1 , x 2 , …, x k ). Define the set of cluster groups corresponding to each cluster center in the initial cluster centers as S = {S 1 , S 2 , …, S k}; A distance calculation module, configured to calculate the shortest Euclidean distance between each sample in the dataset and the initial clustering center through the following formula 1 as In Equation 1, D(x) represents the shortest Euclidean distance, and x i represents the i-th sample in the dataset, and μ i represents the cluster center. Among them, the cluster center μ i is expressed as A probability calculation module, configured to calculate the selection probability P(x) of each sample being selected as the next clustering center through the following formula 3 according to each sample and the shortest Euclidean distance between the sample and the initial clustering center as A clustering update module, configured to select samples as the next clustering center according to the selection probability of each sample by the roulette method, so as to select k next clustering centers; A clustering module, configured to calculate the shortest Euclidean distance between each sample in the dataset and the k next clustering centers through formula 1, and assign each sample to the cluster group corresponding to the clustering center with the smallest distance.

Images