Remote meter reading method for concentrator

Abstract

The invention relates to the field of data processing, in particular to a remote meter reading method for a concentrator, which comprises the following steps: collecting historical sample data, each historical sample data comprising metering characteristic data, state characteristic data and communication characteristic data; according to the optimal K value obtained through calculation, K-means clustering is utilized to obtain a plurality of clustering clusters, a label is distributed to each clustering cluster, and the labels comprise normal labels or abnormal labels; real-time sample data after preprocessing is obtained, the Euclidean distance between the sample data and each cluster is calculated, a label of the sample data is obtained according to the Euclidean distance, and if the label of the sample data is abnormal, it is judged that the collected sample data is wrong. According to the method, the K value is dynamically adjusted and the stability of the cluster is evaluated, so that the model can better adapt to new data, when data distribution changes, the model can automatically adjust the K value to adapt to a new data structure, and the accuracy of anomaly detection is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of data processing. In particular, it relates to a remote meter reading method for a concentrator. BACKGROUND

[0002] In a power system, the concentrator remote meter reading technology is a key link to realize the automatic collection and management of power consumption information. Through the remote reading of the data of smart meters or collection terminals by the concentrator, the collection and analysis of power consumption data can be efficiently completed, providing an important basis for the operation management of the power system, user billing, power dispatching, and fault diagnosis. However, in the actual concentrator remote meter reading process, abnormal situations may occur in the meter reading data, and the abnormal meter reading data may mask the actual operating state of the power equipment, leading to the failure to timely discover and handle equipment faults, or causing user electricity fee calculation errors, etc. Therefore, it is particularly important to detect abnormal meter reading data.

[0003] Chinese patent application No. CN112819299A discloses a differential K-means load clustering method based on center optimization, which includes: collecting load data of user power consumption; obtaining clustering initial centers according to the local density of the load data curve; calculating the differential Euclidean distance of the load data, and combining the differential Euclidean distance and the Euclidean distance of the load data as the similarity measure of the load data; randomly selecting K clustering initial centers as the clustering centers of the clusters for the load data, clustering and updating the clustering centers according to the similarity measure, and outputting the divided K clusters and the clustering labels of each cluster after the update iteration is completed; according to the clustering labels, the load data curves in each category of clusters are averaged to obtain the typical load curve of each category of users, and the power consumption type of each category of power residential users is determined.

[0004] However, the actual data distribution will change with time, environment, or business needs, and the fixed K value and initial center cannot adapt to the new data distribution, leading to a disconnection between the clustering results and the actual data distribution, affecting the accuracy of anomaly detection. In addition, if the K value does not match, normal data may be misjudged as abnormal (false alarm) or real abnormal data may be missed (missed alarm), further reducing the reliability of anomaly detection. SUMMARY

[0005] To solve the above technical problems of using a fixed K value to cluster meter reading data, which cannot adapt to changes in data, resulting in low anomaly detection accuracy, the present application provides a remote meter reading method for a concentrator.

[0006] A remote meter reading method for a concentrator, comprising: collecting historical sample data, each historical sample data including metering feature data, state feature data, and communication feature data; a plurality of clustering clusters are obtained by using K-means clustering according to the optimal K value, and a label is assigned to each clustering cluster, the label including normal or abnormal; The real-time sample data after preprocessing is obtained, the Euclidean distance between the sample data and each clustering cluster is calculated, the label of the sample data is obtained according to the Euclidean distance, and if the label of the sample data is abnormal, it is determined that the collected sample data is incorrect; The calculation process of the optimal K value includes: The historical sample data is clustered based on a given initial K value, and a plurality of initial clustering clusters are obtained. The stability score of the initial clustering cluster is calculated, when the stability score is greater than or equal to a preset threshold, the corresponding initial clustering cluster is marked as a stable cluster, when the stability score is less than the preset threshold, the corresponding initial clustering cluster is marked as an unstable cluster, when the unstable cluster meets a first condition, the unstable cluster is split to obtain a new clustering cluster, and when the unstable cluster meets a second condition, the unstable cluster is merged with other clusters to obtain a new clustering cluster. The clustering clusters generated after the splitting or merging operation and the original stable clusters are counted, and the stability scores of the clustering clusters are recalculated until the clustering clusters no longer perform the splitting or merging operation, and the number of the clustering clusters at this time is the optimal K value.

[0007] Preferably, the first condition is obtained as follows: The intra-cluster dispersion of all initial clustering clusters is calculated, the splitting threshold is determined based on the percentage method, and the intra-cluster dispersion of the unstable cluster greater than the splitting threshold is taken as the first condition.

[0008] If the intra-cluster dispersion of the unstable cluster is greater than the splitting threshold, it indicates that the intra-cluster difference of the unstable cluster is large, and the unstable cluster needs to be split into smaller clusters to improve the stability and accuracy of the final clustering result.

[0009] Preferably, the second condition is obtained as follows: The inter-cluster distance of all initial clustering clusters is calculated, the merging threshold is determined based on the percentage method, and the inter-cluster distance between the unstable cluster and a certain clustering cluster less than or equal to the merging threshold is taken as the second condition.

[0010] If the inter-cluster distance between the unstable cluster and other clusters is less than or equal to the merging threshold, it indicates that the two clusters are highly similar and can be merged to form a new clustering cluster, thereby improving the stability and accuracy of the final clustering result.

[0011] Preferably, the calculation process of the optimal K value further includes: When the unstable cluster does not satisfy the first condition and does not satisfy the second condition, it is judged whether the unstable cluster satisfies a third condition, and when satisfying, the unstable cluster is merged into a cluster with the minimum inter-cluster distance to obtain a new cluster; when not satisfying, the unstable cluster is taken as a stable cluster to participate in subsequent steps.

[0012] Preferably, the third condition is obtained as follows: The ratio of the number of historical sample data contained in the unstable cluster to the total number of historical sample data is calculated, and the ratio less than a preset value is taken as the third condition.

[0013] Preferably, the stability score is calculated as follows: For each initial cluster, a plurality of perturbation clusters are generated, the Jaccard index value between the initial cluster and each corresponding perturbation cluster is calculated, and the average of all corresponding Jaccard index values is taken as the stability score of the initial cluster.

[0014] Preferably, the process of generating a plurality of perturbation clusters is as follows: Each initial cluster contains a plurality of historical sample data, for each initial cluster, Bootstrap sampling method is used to extract historical sample data, and a plurality of perturbation subsets are obtained by repeating the extraction, for each perturbation subset, a preset proportion of Gaussian noise is added, and clustering is performed to obtain a perturbation cluster.

[0015] By adding a small amount of Gaussian noise to the data, the uncertainty of the data in the real world can be simulated, and the perturbation can test the robustness of the clustering result in the face of small random fluctuations.

[0016] Preferably, the historical sample data includes a plurality of feature dimensions, and the process of adding Gaussian noise is as follows: Noise is added to the feature data of each feature dimension of each historical sample data in the perturbation subset, and the noise is 5% of the standard deviation of the corresponding feature dimension in the initial cluster.

[0017] In the real world, data always has noise and uncertainty, and by adding Gaussian noise with a standard deviation of 5%, natural fluctuations can be simulated, so that the clustering result is more meaningful.

[0018] Preferably, the intra-cluster dispersion is the average Euclidean distance from all historical sample data in the initial cluster to the cluster center.

[0019] The present application has the following effects: 1、The application classifies the clustering cluster by the stability score of the clustering cluster, splits or merges the unstable clustering cluster, realizes dynamic adjustment of K value, automatically finds the optimal cluster number according to the internal structure of data, ensures that the final clustering result has high stability and accuracy, avoids the deviation of the clustering result caused by the preset K value, and makes the clustering result more stable and reliable.

[0020] 2、The application dynamically adjusts the K value and evaluates the stability of the cluster, so that the model can better adapt to new data, when the data distribution changes, the model can automatically adjust the K value to adapt to the new data structure, and improve the accuracy of anomaly detection.

[0021] 3、By real-time detection of the remote meter reading data of the concentrator, the accuracy and reliability of the meter reading data can be significantly improved, power equipment failure can be prevented and diagnosed, power resource distribution can be optimized, the safety and stability of the power system can be improved, the operation and management of the smart grid can be supported, and the economy of the power system can be improved. BRIEF DESCRIPTION OF DRAWINGS

[0022] Figure 1 It is a method flow chart of steps S1-S3 in a remote meter reading method for a concentrator.

[0023] Figure 2 It is a method flow chart of steps S20-S25 in a remote meter reading method for a concentrator. DETAILED DESCRIPTION

[0024] The technical solutions in the embodiments of the application will be described clearly and completely below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are part of the embodiments of the application, not all the embodiments.

[0025] The specific embodiments of the application will be described in detail below with reference to the drawings.

[0026] K-means clustering algorithm is a classic unsupervised learning method, which is used for dividing n d-dimensional samples into K clusters to minimize the intra-cluster dispersion. The traditional K-means clustering algorithm needs to preset K value, randomly initializes K initial centers, and different K value and initial center may lead to very different clustering results. The application considers that the preset fixed K value cannot adapt to the change of data, which leads to the disconnection of the clustering result and the actual data distribution, and affects the precision of anomaly detection, therefore, the application proposes to split or merge the cluster by the stability value of the data itself, dynamically adjust the K value, and adapt to the change of data distribution.

[0027] REFERENCE Figure 1A remote meter reading method for a concentrator includes steps S1-S3, as follows: S1, collect historical sample data, each including metering feature data, state feature data, and communication feature data.

[0028] Collect historical (e.g., previous 3 months) power consumption behavior data from smart meters or collection terminals connected to the concentrator, including active power (unit: kW, reflecting power consumption load) and daily power consumption (unit: kWh, reflecting total power consumption); collect historical operation logs of the remote meter reading system from the storage records of the concentrator, including user type data (residential / commercial / industrial) and power supply voltage (unit: V / A, reflecting power supply stability); collect historical communication transmission quality data from the communication module of the concentrator and the master station / terminal, including signal strength (unit: dBm, reflecting wireless signal quality) and transmission delay (unit: ms, time spent from terminal to master station). Remove missing values and duplicate values in all data.

[0029] Normalize data of different dimensions and value ranges: Min-Max standardization (normalization) is used for numerical type data, such as active power, daily power consumption, and power supply voltage; one-hot encoding is applied to category type data, converting each category into a binary feature, such as user type data. Standardize all collected data to the range [0, 1].

[0030] Extract key features of each sample based on the normalized data, including three categories of metering features, state features, and communication features, each with its corresponding normalized sub-features. Metering features correspond to active power and daily power consumption, state features correspond to user type and power supply voltage, and communication features correspond to signal strength and transmission delay. Concatenate the three types of features of all samples in order to construct their corresponding feature vectors, ensuring consistent feature vector structure, thereby obtaining all historical sample data. In this embodiment, historical sample data includes 7 feature dimensions: active power, daily power consumption, residential power supply voltage, commercial power supply voltage, industrial power supply voltage, signal strength, and transmission delay.

[0031] S2, use K-means clustering based on the calculated optimal K value to obtain multiple clustering clusters, and assign a label to each clustering cluster, including normal or abnormal.

[0032] Traditional K-means clustering needs to pre-set the value of K, but the cluster structure of actual data is often unknown. Dynamically adjusting the value of K can automatically find the optimal number of clusters according to the internal structure of the data, avoiding the deviation of the clustering result caused by improper selection of the value of K; clustering based on the optimal value of K can ensure that the data points in each clustering cluster have high similarity, while the data points between different clusters have great differences, making the clustering result clearer and improving the accuracy of anomaly detection; by assigning "normal" or "abnormal" labels to each clustering cluster, abnormal behaviors in the data can be quickly identified.

[0033] As shown in the formula (1), the calculation process of the optimal value of K includes steps S20-S25, which are as follows: Figure 2 S20, cluster the historical sample data based on the given initial value of K to obtain multiple initial clustering clusters.

[0034] In this embodiment, the initial value of K is set to 5. Five different sample data are selected from all historical sample data as initial clustering centers, the Euclidean distance of each historical sample data to the five initial clustering centers is calculated, each historical sample data is assigned to the clustering cluster corresponding to the initial clustering center with the nearest Euclidean distance, the average value of all historical sample data in each clustering cluster is recalculated, and the average value is taken as the new clustering center. The steps of assigning historical sample data and updating clustering center are repeated until the clustering center no longer changes, the initial clustering is completed, and five initial clustering clusters are obtained.

[0035] S21, calculate the stability score of the initial clustering cluster, and when the stability score is greater than or equal to a preset threshold, mark the corresponding initial clustering cluster as a stable cluster; when the stability score is less than the preset threshold, mark the corresponding initial clustering cluster as an unstable cluster.

[0036] The calculation process of the stability score of the initial clustering cluster is as follows: for each initial clustering cluster, 80% of the historical sample data in the initial clustering cluster is extracted by the Bootstrap sampling method with replacement to generate a perturbation subset, and the above sampling step is repeated until 10 perturbation subsets are obtained; The standard deviation of each feature dimension in the corresponding initial clustering cluster is calculated, and noise is added to the feature data of each feature dimension of each historical sample data in the perturbation subset, wherein the noise is 5% of the standard deviation of the corresponding feature dimension in the corresponding initial clustering cluster, thereby obtaining 10 perturbation data sets. The calculation formula when adding noise is as follows: In the formula (2), σi represents the standard deviation of the i-th feature dimension in the initial clustering cluster; and In the formula (2), σi represents the standard deviation of the i-th feature dimension in the initial clustering cluster; and​​ This indicates that the mean is 0 and the variance is 0. Gaussian distribution; Indicates the first subset of the perturbation The first historical sample data 3D eigenvalues; This indicates the first element in the perturbation subset after adding noise. The first historical sample data 3D eigenvalues. This allows noise to be added to each historical sample data point in the perturbation subset, more realistically simulating random perturbations during data acquisition; Perform K-means clustering with a K value of 1 on each perturbation dataset to generate 10 perturbation clusters. Calculate the Jaccard index value between the initial cluster and each of its corresponding perturbation clusters, and take the mean of all the corresponding Jaccard index values ​​as the stability score of the initial cluster.

[0037] The calculation of the Jaccard index value follows the formula: In the formula, Indicates the initial cluster; Represents the first cluster corresponding to the initial cluster. A cluster of disturbances; This represents the historical sample dataset in the initial cluster; Indicates the first A sample dataset of perturbation clusters; This represents the number of samples that belong to both the initial cluster and the perturbation cluster; This represents the total number of samples in the initial cluster and the perturbation cluster; Indicates the first The initial cluster and its corresponding perturbation cluster in the i-th initial cluster are the i-th... The Jaccard index value of each perturbation cluster is in the range of [0,1]. The larger the index value, the higher the similarity between the initial cluster and the perturbation cluster, and the less affected the initial cluster is by the perturbation, that is, the more stable the initial cluster is.

[0038] In this embodiment, the preset threshold is taken as an empirical value of 0.7. In other embodiments, it can also be adjusted according to the business background. When the stability score of the initial cluster is greater than or equal to 0.7, it indicates that the initial cluster is not sensitive to disturbances and the sample composition is consistent. The initial cluster can be determined to be a stable cluster and no adjustment is needed. When the stability score is less than 0.7, it indicates that the initial cluster is relatively sensitive to disturbances and the initial cluster is marked as an unstable cluster.

[0039] The K-means algorithm is prone to getting trapped in local optima, especially when the initial cluster centers are poorly chosen. Data perturbation can generate multiple different perturbed datasets, increasing the chances of finding the global optimum. Furthermore, in practical applications, data is often noisy; adding small amounts of Gaussian noise can simulate the uncertainty of real-world data, and this perturbation can test the robustness of the clustering results to small random fluctuations.

[0040] S22. When an unstable cluster satisfies the first condition, the unstable cluster is split to obtain a new cluster.

[0041] If an initial cluster is marked as unstable, it is necessary to first determine whether the initial cluster meets the first condition. The process of obtaining the first condition is as follows: Calculate the intra-cluster dispersion of all initial clusters. The intra-cluster dispersion of each initial cluster is the mean of the Euclidean distances from all historical sample data within the cluster to the cluster center. Arrange all initial clusters in ascending order of intra-cluster dispersion to obtain the intra-cluster dispersion distribution of all initial clusters. Determine the splitting threshold using the percentage method. Select the distance value corresponding to a high percentile (e.g., 95%) as the splitting threshold. That is, find a value among all intra-cluster dispersions such that 95% of the intra-cluster dispersions are less than or equal to this value. This value is the distance value corresponding to the 95th percentile. The intra-cluster dispersion of unstable clusters being greater than the splitting threshold is taken as the first condition.

[0042] The formula for calculating the intra-cluster dispersion of the initial cluster is as follows: In the formula, Indicates the first The intra-cluster dispersion of the initial cluster, i.e., the first... The mean Euclidean distance from all historical sample data within an initial cluster to the cluster center; Indicates the first The number of all historical samples in each initial cluster; Represents the initial cluster A historical sample of data within; Indicates the first The cluster centers of the initial clusters.

[0043] If the intra-cluster dispersion of an unstable cluster is greater than the splitting threshold, it indicates that the intra-cluster differences of the unstable cluster are large, and it needs to be split into smaller clusters. The unstable cluster is then split into two new clusters by performing a K-means clustering algorithm with a K value of 2.

[0044] S23. When an unstable cluster satisfies the second condition, the unstable cluster is merged with other clusters to obtain a new cluster.

[0045] If the unstable cluster does not satisfy the first condition and does not need to be split, it is determined whether the unstable cluster satisfies the second condition. The second condition is obtained as follows: the inter-cluster distance between each two initial clustering clusters is calculated, that is, the Euclidean distance between the cluster centers of each two initial clustering clusters is calculated, the initial clustering clusters are arranged in ascending order of the inter-cluster distance between each two initial clustering clusters, and the inter-cluster distance distribution of all initial clustering clusters is obtained; the merging threshold is determined by using the percentile method, and a distance value corresponding to a low percentile (for example, 5%) is selected as the merging threshold, that is, in all inter-cluster distances, a value is found, such that 5% of the inter-cluster distances are less than or equal to the value, and the value is the distance value corresponding to the 5% percentile; the inter-cluster distance between the unstable cluster and other clusters that is less than or equal to the merging threshold is taken as the second condition.

[0046] The formula for calculating the inter-cluster distance between two initial clustering clusters is as follows: In the formula, d (i, j) represents the inter-cluster distance between the initial clustering cluster i and the initial clustering cluster j, that is, the Euclidean distance between the cluster centers of the two initial clustering clusters; represents the cluster center sample of the initial clustering cluster i; represents the cluster center sample of the initial clustering cluster j; represents the jth feature value of the cluster center sample of the initial clustering cluster i; represents the jth feature value of the cluster center sample of the initial clustering cluster j.

[0047] If the inter-cluster distance between the unstable cluster and other clusters is less than or equal to the merging threshold, it indicates that the two clusters are highly similar and can be merged to form a new clustering cluster.

[0048] S24, when the unstable cluster does not satisfy the first condition and does not satisfy the second condition, it is determined whether the unstable cluster satisfies the third condition, and if so, the unstable cluster is merged into the clustering cluster with the smallest inter-cluster distance to obtain a new clustering cluster; if not, it is taken as a stable cluster to participate in subsequent steps.

[0049] When the unstable cluster does not satisfy the first condition and does not satisfy the second condition, it is determined whether the unstable cluster satisfies the third condition, and the third condition is obtained as follows: the ratio of the number of historical sample data contained in the unstable cluster to the total number of historical sample data is calculated, and the ratio less than a preset value is taken as the third condition. In this embodiment, the preset value is an empirical value of 3%, and if the ratio is less than 3%, it indicates that the sample quantity of the unstable cluster is too small, and it needs to be merged into the cluster with the smallest inter-cluster distance to form a new clustering cluster. If the ratio is greater than or equal to 3%, it is marked as a stable cluster.

[0050] ​​​​​​S25, count the cluster groups generated after the splitting or merging operation and the original stable cluster, and recalculate the stability score of the cluster groups until the cluster groups no longer undergo splitting or merging operation, and the number of the cluster groups at this time is the optimal K value.

[0051] The unstable cluster in the above step forms a new cluster group after splitting or merging, and the new cluster group and all stable clusters are counted, and the stability score of the cluster groups is recalculated. The calculation method is the same as that of calculating the stability score of the initial cluster group in the above step. If the stability scores of all cluster groups are greater than or equal to 0.7, or there is only a cluster group with a stability score less than 0.7 but marked as a stable cluster, it indicates that the cluster groups at this time do not need to be split or merged, and the number of the cluster groups at this time is the optimal K value. If there is a cluster group with a stability score less than 0.7 and does not meet the third condition, it is marked as an unstable cluster, and the above steps S22 to S25 are performed again. It can be understood that when the above steps are performed again, the initial cluster group in the steps is the cluster group generated after the splitting or merging operation and all stable clusters counted after each iteration.

[0052] For ease of understanding, the above iteration process is simply described in this embodiment: after forming 5 initial cluster groups and calculating the stability scores of the 5 initial cluster groups, it is assumed that there is 1 unstable cluster and 4 stable clusters.

[0053] First, it is judged whether the 1 unstable cluster meets the first condition, i.e., it is judged whether the intra-cluster dispersion of the unstable cluster is greater than the splitting threshold. If it is greater than the splitting threshold, the 1 unstable cluster is split into two new cluster groups, and the two new cluster groups and the original 4 stable clusters form a new round of cluster groups. The number of the cluster groups in this round is counted and the stability scores of the cluster groups in this round are calculated. If the stability scores of all cluster groups in this round are greater than or equal to 0.7, the number of all cluster groups in this round is taken as the optimal K value, i.e., the K value is 6.

[0054] If the 1 unstable cluster does not meet the first condition, it is judged whether the unstable cluster meets the second condition, i.e., it is judged whether the inter-cluster distance between the unstable cluster and other clusters is less than or equal to the merging threshold. If it is less than or equal to the merging threshold, the unstable cluster is merged with other clusters to form a new cluster group, and the new cluster group and the original 3 stable clusters (one of the stable clusters is used to merge with the unstable cluster, and the merged cluster needs to be removed) form a new round of cluster groups. The number of the cluster groups in this round is counted and the stability scores of the cluster groups in this round are calculated. If the stability scores of all cluster groups in this round are greater than or equal to 0.7, the number of all cluster groups in this round is taken as the optimal K value, i.e., the K value is 4.

[0055] If the unstable cluster does not satisfy the first condition and the second condition, it is determined whether the unstable cluster satisfies a third condition, i.e., whether the ratio of the number of historical samples in the unstable cluster to the total number of historical samples is less than 3%. If the ratio is less than 3%, it indicates that the sample size is too small, and the unstable cluster is merged into the cluster closest to it in the cluster distance to form a new clustering cluster. The new clustering cluster and its original three stable clusters form a new round of clustering clusters. The number of clustering clusters in this round is counted, and the stability scores of all clustering clusters in this round are recalculated. If the stability scores of all clustering clusters in this round are greater than or equal to 0.7, the number of all clustering clusters in this round is taken as the best K value, i.e., the K value is 4. If the ratio is greater than 3%, the unstable cluster is taken as a stable cluster and is marked. At this time, the five initial clustering clusters are all stable clusters, and the best K value is 5. It can be understood that since the above description is based on the assumption that there is one unstable cluster, the best K value is directly obtained after the unstable cluster is taken as a stable cluster. If there are two unstable clusters in the five initial clustering clusters, after one of the unstable clusters is taken as a stable cluster because it does not satisfy the first condition and the second condition and the sample ratio is greater than 3%, the unstable cluster as a stable cluster normally participates in all calculations.

[0056] If the new round of clustering clusters formed after the splitting or merging of the unstable cluster in the above description process still has a cluster with a stability score less than 0.7, the cluster with a stability score less than 0.7 is determined to be an unstable cluster, and the above process is repeated until all clustering clusters in the new round are stable clusters. The number of all clustering clusters in this round is the best K value.

[0057] Based on the best K value, K-means clustering is performed on the collected historical sample data: K different sample data are randomly selected from all historical sample data as initial clustering centers. The Euclidean distance of each historical sample data to the K initial clustering centers is calculated. Each historical sample data is assigned to the clustering cluster corresponding to the initial clustering center with the closest Euclidean distance. The average value of all historical sample data in each clustering cluster is recalculated as a new clustering center. The steps of assigning historical sample data and updating clustering centers are repeated until the clustering centers no longer change. Clustering is completed, and K clustering clusters are obtained.

[0058] The K clustering clusters obtained after clustering are preset business rules, and whether the clustering clusters conform to the real business mode is verified. The preset rules explicitly involve the business scenarios and key objects of the remote meter reading data of the concentrator, output a business scenario list, determine the normal range of each feature based on the business scenario list, industry standards and equipment parameters, and formulate business rules. For all clustering clusters obtained after clustering by the optimal K value, whether they conform to the business rules is verified one by one, business tags are added to the clustering clusters according to the verification results, the clustering clusters conforming to the business rules are normal, and the clustering clusters not conforming to the business rules are abnormal, and finally the clustering clusters with known business tags are obtained.

[0059] S3, obtaining real-time sample data after preprocessing, calculating the Euclidean distance between the sample data and each clustering cluster, and obtaining the label of the sample data according to the Euclidean distance, if the label of the sample data is abnormal, it is determined that the collected sample data is incorrect.

[0060] The real-time data is preprocessed and aligned with the features, the preprocessed sample is converted into a feature vector with the same feature structure as the historical sample data, and real-time sample data is formed. The Euclidean distance between the real-time sample data and the cluster center of each clustering cluster is calculated, the clustering cluster with the smallest Euclidean distance is selected as the belonging cluster of the sample data, and whether the real-time sample data is normal is judged according to the business label of the belonging cluster, if the business label of the belonging cluster is abnormal, it is determined that the sample data is abnormal.

[0061] By detecting the abnormality in the remote meter reading data of the concentrator in real time, the accuracy and reliability of the meter reading data can be significantly improved, power equipment failures can be prevented and diagnosed, power resource distribution can be optimized, the safety and stability of the power system can be improved, the operation and management of the smart grid can be supported, and the economy of the power system can be improved.

[0062] In each iteration, the unstable clusters are split or merged, the K value is dynamically adjusted, the optimal number of clusters is automatically found according to the internal structure of the data, and the final clustering result has high stability and accuracy. This method avoids the deviation of the clustering result caused by manually presetting the K value, so that the clustering result is more stable and reliable. Moreover, by dynamically adjusting the K value and evaluating the stability of the cluster, the model can better adapt to new data. When the data distribution changes, the model can automatically adjust the K value to adapt to the new data structure.

[0063] It should be noted that, for those skilled in the art, without departing from the concept of the present application, a number of modifications and improvements can be made, which all belong to the protection scope of the present application. Therefore, the protection scope of the present application patent should be subject to the appended claims.

Claims

1. A remote meter reading method for a concentrator, characterized in that, include: Collect historical sample data, each of which includes measurement characteristic data, status characteristic data, and communication characteristic data; Based on the calculated optimal K value, K-means clustering is used to obtain multiple clusters, and each cluster is assigned a label, which includes normal or abnormal. Obtain preprocessed real-time sample data, calculate the Euclidean distance between the sample data and each cluster, obtain the label of the sample data based on the Euclidean distance, and determine that the collected sample data is incorrect if the label of the sample data is abnormal. The calculation process for the optimal K value includes: Based on a given initial K value, historical sample data are clustered to obtain multiple initial clusters; Calculate the stability score of the initial clusters. When the stability score is greater than or equal to a preset threshold, the corresponding initial cluster is marked as a stable cluster. When the stability score is less than the preset threshold, the corresponding initial cluster is marked as an unstable cluster. When an unstable cluster meets the first condition, it is split to obtain a new cluster. When an unstable cluster meets the second condition, it is merged with other clusters to obtain a new cluster. The stability scores of the clusters generated after splitting or merging operations, as well as the original stable clusters, are counted. The stability scores of these clusters are recalculated until no further splitting or merging operations are performed. The number of clusters at this point is the optimal K value.

2. The remote meter reading method for a concentrator according to claim 1, characterized in that, The process of obtaining the first condition is as follows: Calculate the intra-cluster dispersion of all initial clusters, determine the splitting threshold based on the percentage method, and take the intra-cluster dispersion of unstable clusters that is greater than the splitting threshold as the first condition.

3. The remote meter reading method for a concentrator according to claim 2, characterized in that, The process of obtaining the second condition is as follows: Calculate the inter-cluster distances of all initial clusters, determine the merging threshold based on the percentage method, and use the condition that the inter-cluster distance between an unstable cluster and a certain cluster is less than or equal to the merging threshold as a second condition.

4. The remote meter reading method for a concentrator according to claim 3, characterized in that, The process of calculating the optimal K value also includes: When an unstable cluster does not meet either the first or the second condition, it is determined whether the unstable cluster meets the third condition. If it does, the unstable cluster is merged into the cluster with the smallest inter-cluster distance to obtain a new cluster; otherwise, it is treated as a stable cluster and participates in subsequent steps.

5. The remote meter reading method for a concentrator according to claim 4, characterized in that, The process of obtaining the third condition is as follows: Calculate the ratio of the number of historical sample data contained in the unstable cluster to the total number of historical sample data, and use a ratio that is less than a preset value as the third condition.

6. The remote meter reading method for a concentrator according to claim 1, characterized in that, The stability score is calculated as follows: For each initial cluster, several perturbation clusters are generated. The Jaccard index values ​​between the initial cluster and each of its corresponding perturbation clusters are calculated, and the mean of all corresponding Jaccard index values ​​is taken as the stability score of the initial cluster.

7. A remote meter reading method for a concentrator according to claim 6, characterized in that, The process of generating several perturbation clusters is as follows: Each initial cluster contains multiple historical sample data. For each initial cluster, the Bootstrap sampling method is used to extract historical sample data. This process is repeated to obtain several perturbation subsets. For each perturbation subset, a preset proportion of Gaussian noise is added, and clustering is performed to obtain perturbation clusters.

8. A remote meter reading method for a concentrator according to claim 7, characterized in that, The historical sample data includes multiple feature dimensions, and the process of adding Gaussian noise is as follows: Noise is added to the feature data of each feature dimension of the historical sample data in the perturbation subset, where the noise is 5% of the standard deviation of the corresponding feature dimension in the corresponding initial cluster.

9. A remote meter reading method for a concentrator according to claim 2, characterized in that, The intra-cluster dispersion is the average Euclidean distance from all historical sample data within the initial cluster to the cluster center.

Images