Data processing method of electric power laying device

Abstract

The invention relates to the technical field of data processing, in particular to a data processing method of a power distribution device, which comprises the following steps of: constructing a data point set for reflecting a power load in a current area based on collected user power consumption data in the current area, and clustering data points in the data point set for multiple times by utilizing a K-means clustering algorithm; obtaining a plurality of circular rings of each category in each clustering result, and determining the data density of each circular ring; sequentially determining intra-class density change rationality, inter-class distribution rationality and optimization degree of each clustering result; and determining an optimal clustering result of a K-means clustering algorithm according to the size of the optimization degree, and realizing data processing of the power layout device. According to the method, distribution characteristics of data points in clustering categories are deeply analyzed, the intra-category density change rationality and inter-category distribution rationality are considered, it is ensured that the clustering result better fits the space distribution and density difference of actual power consumption loads, and power consumption behaviors of different areas are more truly reflected.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and particularly to a data processing method for a power distribution device. Background Art

[0002] The development of the data processing method for power distribution devices stems from the needs of smart grid construction and energy digital transformation, and its core goal is to improve the reliability, efficiency, and intelligence level of the power grid. Currently, the popularization of smart meters, sensors, and Internet of Things technologies has enabled high-frequency data collection, but the industry still faces challenges such as data silos, insufficient real-time performance, and security risks.

[0003] The data processing of power distribution devices aims to clean, extract features, and analyze patterns of the user electricity consumption data collected by devices such as smart meters, so as to achieve accurate classification of electricity consumption behaviors and abnormal monitoring. It can not only optimize the operation of the power grid, assist in power grid dispatching and energy allocation by identifying peak loads, balanced or intermittent electricity consumption patterns, ensure electricity consumption safety, detect abnormal behaviors such as electricity theft and equipment failures, but also support demand-side response, predict regional electricity consumption trends, and promote the consumption of renewable energy. Ultimately, it provides data-driven decision-making support for power companies and realizes the safe, economic, and sustainable operation of the power system. The K-means Clustering Algorithm is an unsupervised learning algorithm commonly used for classifying electricity consumption behavior patterns of power users and extracting features of electricity consumption data. By clustering data points with similar electricity consumption patterns into one category, it helps to achieve data processing goals such as identifying peak loads, detecting abnormal electricity consumption behaviors, and optimizing power grid dispatching.

[0004] However, when using the K-means clustering algorithm to cluster user electricity consumption data, the number of clustering categories K needs to be set artificially in advance. However, for different urban areas, the artificially set clustering number cannot accurately reflect the various electricity consumption situations in the current area. For example, when the number of clustering categories is too small, the clustering result is rough, and the electricity consumption data of different categories of users may be assigned to the same cluster, which cannot accurately reflect the electricity consumption situation and reduces the accuracy of electricity consumption behavior classification and abnormal monitoring; when the number of clustering categories is too large, the clustering result may be too detailed, easily assign the same type of samples to different clusters, and the clustering result is unstable and sensitive to minor changes in user electricity consumption data, resulting in the clustering result being unable to effectively guide power grid dispatching and energy allocation decisions. Summary of the Invention

[0005] When using the K-means clustering algorithm to cluster user power consumption data, the artificially set number of clusters cannot accurately reflect the various power consumption situations in each region. When the number of cluster categories is too small, it cannot accurately reflect the power consumption situation, reducing the accuracy of power consumption behavior classification and abnormal monitoring. When the number of cluster categories is too large, the clustering results cannot effectively guide power grid dispatching and energy allocation decisions. To solve this problem, the present invention proposes a data processing method for a power distribution device, and the method includes the following steps: Based on the collected user power consumption data of the current region, construct a data point set for the current region to reflect the power load, and use the K-means clustering algorithm to perform multiple clusterings on the data points in the data point set; Denote any clustering result as the target result, denote any category in the target result as the current category, construct multiple concentric circles for the current category to obtain multiple rings for the current category, and determine the data density of each ring of the current category according to the number of data points in each ring of the current category and the area of each ring; Determine the rationality of the within-class density change of the target result according to the data density of each ring of each category in the target result; Obtain the regular clusters and sparse clusters of the current category, and the sparse region of the current category. The regular clusters, sparse clusters and sparse regions all contain several rectangles. Determine the rationality of the between-class distribution of the target result according to the width and length of the sparse regions of each category in the target result, the number of rectangles in the sparse clusters, the number of rectangles in the sparse regions, the number of data points in the sparse regions, and the number of data points in each regular cluster; Determine the preference degree of the target result according to the distance from the data points in each category in the target result to the within-class center points of each category, the distance between the within-class center points between categories, and the rationality of the within-class density change and the rationality of the between-class distribution; Determine the best clustering result of the K-means clustering algorithm according to the magnitude of the preference degree, and realize the data processing of the power distribution device.

[0006] By constructing concentric circles and calculating the data density within each ring, the present invention deeply analyzes the distribution characteristics of data points within the cluster categories, taking into account both the "rationality of within-class density change" and the "rationality of between-class distribution", ensuring that the clustering results are more in line with the spatial distribution and density differences of the actual power load, and more truly reflecting the power consumption behavior in different regions; By evaluating multiple clustering results, the clustering result with the highest preference degree is selected, so that the number of cluster categories is no longer artificially set, but automatically determined by the structural characteristics of the data itself, improving the scientificity and effectiveness of classification and abnormal monitoring; After accurate classification, the sparse regions and regular regions of power consumption behavior can be more effectively identified, providing a more robust basis for abnormal power consumption detection, thereby enhancing the abnormal detection and early warning capabilities of the system; With accurate clustering category division and sparse region positioning, it can provide more targeted regional power load information for the power grid dispatching department, optimize resource allocation, and improve the economy and security of power grid operation.

[0007] Further, constructing the data point set for reflecting the electricity load in the current area includes: using the proportion of electricity consumption during peak hours in the electricity consumption data of users in the current area as the abscissa and the daily average electricity consumption as the ordinate to construct a two-dimensional coordinate system, thereby obtaining the data point set for reflecting the electricity load in the current area.

[0008] Further, obtaining multiple concentric rings for the current category includes: taking the center point within the current category as the center of the circle, constructing the smallest circumscribed circle that contains all the data points within the current category, constructing concentric circles of the multiple smallest circumscribed circles according to a preset scaling ratio, dividing the smallest circumscribed circle into multiple concentric rings, and sorting the concentric rings in order from the inside to the outside to obtain multiple concentric rings for the current category.

[0009] Further, the data density satisfies: ; where is the data density of the concentric ring for the current category, is the number of data points within the concentric ring for the current category, is the preset scaling ratio, is the serial number of the concentric ring for the current category, is the radius of the smallest circumscribed circle for the current category, is the max - min normalization function.

[0010] The present invention realizes the accurate calculation of the data point density per unit area within each concentric ring by excluding the area of the previous concentric ring, avoiding the confusion caused by simple overall density, and reflecting the more real spatial distribution of electricity load; by means of max - min normalization, the data densities between different categories and different concentric rings are scaled to a unified standard interval, facilitating cross - category rationality evaluation and subsequent clustering optimization judgment; the precise density calculation helps to identify the boundaries between conventional clusters and sparse clusters, and judge the clustering quality from the data density change trend, promoting the adaptive adjustment of the clustering results.

[0011] Further, the rationality of the change in the density within the category satisfies: ; where is the rationality of the change in the density within the category of the target result, is the number of categories in the target result, is the number of concentric rings of the categories in the target result, is the data density of the concentric ring of the category in the target result, is the data density of the concentric ring of the category in the target result, is the length of the preset moving unit.

[0012] The present invention utilizes the deformation of the function to smooth the density difference between adjacent rings, avoiding the sharp jumps of simple differences, and effectively reflecting the rationality of the decreasing or increasing density within the category; through the comprehensive evaluation of the density changes of all categories and all ring data, objectively measure the uniformity and continuity of the internal data distribution of the categories in the clustering result, and identify abnormal or unreasonable density mutations.

[0013] Further, the obtaining of the regular cluster and the sparse cluster of the current category includes: obtaining the distance between the center points within the current category and the other categories in the target result, taking the center points within the category of the category closest to the current category as the midpoints of any two opposite sides of the square, taking the distance between the center points within the category of the category closest to the current category as the side length of the square, starting from any side of the square, dividing the square into multiple rectangles with the same area in sequence to obtain multiple rectangles of the current category; based on the number of data points within the rectangle, using the K-means clustering algorithm to cluster the multiple rectangles of the current category to obtain two clusters, recording the cluster with the largest number of data points within the cluster as the regular cluster, and the other as the sparse cluster, so as to obtain the regular cluster and the sparse cluster of the current category.

[0014] Further, the obtaining method of the sparse region of the current category is: merging all adjacent rectangles that belong to the sparse cluster, and recording the region with the largest area after merging as the sparse region in the sparse cluster.

[0015] Further, the rationality of the inter-class distribution satisfies: ; where is the rationality of the inter-class distribution of the target result, is the number of categories in the target result, and are respectively the width and length of the sparse region of category in the target result, and are respectively the number of the sparse region and the rectangles within the sparse cluster of category in the target result, is the number of data points within the sparse region of category in the target result, is the average value of the number of data points within all rectangles of the regular cluster of category in the target result.

[0016] The rationality of the inter-class distribution of the present invention combines the geometric size (width-to-length ratio) of the sparse region, the ratio of the number of sparse regions to the number of sparse cluster rectangles, and the density characteristics of data points within the sparse region and the regular cluster, reflecting the spatial structure and distribution law between categories; expressing the influence of density on the region size and shape in the form of exponential weights makes the distribution of sparse and dense regions in the clustering result more scientific and reasonable, avoiding overly blurred or overlapping category boundaries.

[0017] Further, the preference degree satisfies: ; where is the preference degree of the target result, is the number of categories in the target result, is the category in the target result the number of data points within, is the category in the target result the th data point in to the intra-class center point of category distance, is the category in the target result and category distance between the intra-class center points of, is the rationality of the intra-class density change in the target result, is the rationality of the inter-class distribution in the target result.

[0018] The preference degree of the present invention reflects the compactness and distinguishability of clustering through the ratio of the average distance (compactness) from the intra-class data points to the center to the distance between the category centers (separation), realizing a scientific evaluation of the clustering structure quality; combining the mean values of the rationality of intra-class density change and inter-class distribution rationality ensures that the clustering result is not only tight and distinguishable, but also conforms to the actual characteristics of the electricity load in terms of data density distribution.

[0019] Further, the data processing of the power distribution device includes: taking the clustering result corresponding to the maximum value among the preference degrees of all clustering results as the best clustering result for clustering the current region using the K-means clustering algorithm, and completing the data processing of the power distribution device.

[0020] The present invention has the following beneficial effects: By performing multiple clusterings on the same data point set and calculating the "rationality of intra-class density change" and "rationality of inter-class distribution", the clustering quality under different K values can be objectively evaluated, thus automatically selecting the optimal K value and avoiding the subjective bias caused by the artificial setting of the K value in traditional methods. The concentric circle and ring structures are introduced to evaluate the density change of each cluster of data, distinguish between regular clusters and sparse clusters, and can more precisely characterize the distribution characteristics of the electricity load in each region, improving the accuracy of the classification of electricity consumption behavior types. By identifying sparse clusters and their sparse regions, abnormal points or abnormal regions in the electricity consumption data can be effectively discovered, and the early warning sensitivity for atypical electricity consumption patterns (such as sudden spikes, equipment failures, etc.) can be improved. Finally, the optimal number of clustering categories is obtained, which can avoid the defect of inaccurate clustering caused by the number of clustering categories not conforming to the data distribution characteristics, effectively improving the accuracy of the K-means clustering algorithm and the accuracy of identifying the electricity consumption patterns in the current area. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 is a flowchart of the steps of a data processing method for a power distribution device according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0022] The technical solutions in the embodiments of the present invention will be clearly and completely described below. The described embodiments are some of the embodiments of the present invention. All other embodiments obtained by those skilled in the art without creative efforts based on the embodiments of the present invention belong to the protection scope of the present invention.

[0023] The specific embodiments of the present invention will be described in detail below with reference to the drawings.

[0024] Please refer to Figure 1 , which shows a flowchart of the steps of a data processing method for a power distribution device provided by an embodiment of the present invention. The method includes the following steps: In order to solve the problem that when using the K-means clustering algorithm to cluster user electricity consumption data, the artificially set number of clusters cannot accurately reflect the various electricity consumption situations in each region. When the number of clustering categories is too small, the electricity consumption situation cannot be accurately reflected, reducing the accuracy of electricity consumption behavior classification and abnormal monitoring. When the number of clustering categories is too large, the clustering result cannot effectively guide the power grid dispatching and energy distribution decision-making, the present invention proposes a data processing method for a power distribution device, and the method includes the following steps: S01: Based on the collected user electricity consumption data of the current area, construct a data point set for the current area to reflect the electricity load, and use the K-means clustering algorithm to perform multiple clusterings on the data points in the data point set.

[0025] Implementers can set the number of clustering categories for each clustering according to the specific implementation situation. For example, the values of the number of clustering categories sequentially traverse 2, 3, 4, ……, ( is the number of data points).

[0026] Specifically, the construction of the data point set for the current area to reflect the electricity load includes: Taking the proportion of electricity consumption during peak hours in the user electricity consumption data of the current area as the abscissa and the daily average electricity consumption as the ordinate, a two-dimensional coordinate system is constructed to obtain the data point set for the current area to reflect the electricity load.

[0027] S02: Obtain multiple rings of each category in each clustering result and determine the data density of each ring.

[0028] It should be noted that the data density can intuitively reflect the degree of compactness of the data distribution in the clustering space. By calculating the data density of the rings, the uniformity and rationality of the data distribution of each category in the clustering result can be effectively evaluated. Therefore, this step lays a foundation for subsequent evaluation of the clustering effect by analyzing the data density of the rings.

[0029] Record any clustering result as the target result, and record any category in the target result as the current category. Construct multiple concentric circles for the current category to obtain multiple rings of the current category. According to the number of data points in each ring of the current category and the area of each ring, determine the data density of each ring of the current category.

[0030] Specifically, the obtaining of multiple rings of the current category includes: Taking the center point within the category of the current category as the center, construct the smallest circumscribed circle that contains all the data points within the current category. According to the preset scaling ratio, construct concentric circles of the smallest circumscribed circle. Divide the smallest circumscribed circle into multiple rings, and sort the rings in order from the inside to the outside to obtain multiple rings of the current category.

[0031] Implementers can set the scaling ratio and the number of concentric circles according to the specific implementation situation. For example, the scaling ratio is 0.1 times the length of the radius of the smallest circumscribed circle for each scaling (that is, multiple concentric circles are made with 0.1, 0.2, 0.3, ……, 0.9 times the radius length of the smallest circumscribed circle); the number of concentric circles is 9.

[0032] Specifically, the data density satisfies: ; In the formula, is the data density of the ring of the current category, is the ring of the current category, and is a preset scaling ratio, , is the serial number of the ring of the current category, is the radius of the minimum circumscribed circle of the current category, is the maximum-minimum normalization function.

[0033] S03: Determine the rationality of the within-class density change of each clustering result.

[0034] It should be noted that the rationality of the within-class density change is used to quantify whether the change trend of the data point density from the category center to the boundary in the clustering result conforms to the ideal distribution. Therefore, in this step, by constructing an index of the rationality of the within-class density change, the abnormal data distribution areas in the clustering result can be effectively identified.

[0035] Determine the rationality of the within-class density change of the target result according to the data density of each ring of each category in the target result.

[0036] Specifically, the rationality of the within-class density change satisfies: ; In the formula, is the rationality of the within-class density change of the target result, is the number of categories in the target result, is the number of rings of the category in the target result, is the category in the target result of the ring data density, is the category in the target result of the ring data density, is the length of the preset moving unit.

[0037] Implementers can set the length of the moving unit according to the specific implementation situation. For example, 5.

[0038] Among them, is function deformation, taking as the independent variable, and moving the entire function image 5 unit lengths to the left. Since the ideal clustering density distribution feature is that the closer to the category center, the more data points and the denser the distribution; the closer to the category boundary, the fewer data points and the more discrete the distribution. Such a clustering effect not only makes the within-class center point more representative of the entire category, but also can judge whether the current clustering effect is good. When function it indicates that the data point density in the ring of the category in the target result is greater than the ring The data point density within indicates that from the category of the circular ring extending outward to the circular ring when, the density of data points shows a decreasing trend, which conforms to the ideal within-class density change trend, and the larger, the greater the rationality of the within-class density change of the target result; when it indicates that in the category of the target result of the circular ring the data point density within is less than the data point density within the circular ring indicating that from the category of the circular ring extending outward to the circular ring when, the density of data points shows an increasing trend, violating the ideal within-class density change trend, and the smaller, the smaller the rationality of the within-class density change of the target clustering result. The application and transformation of the function are to make the formula more sensitive to negative values and less responsive to positive values. When there are regions that do not conform to the density change trend, their impact on the overall function value becomes larger, resulting in less rationality of the within-class density change. Thus, the impact of regions with unreasonable density changes on the overall rationality can be highlighted. Here, it is also because in a clustering result, as long as there are individual clustering errors and individual unreasonable clustering situations, it can indicate that the clustering effect of this time is poor, but the degree still needs to be judged according to the numerical value.

[0039] S04: Determine the inter-class distribution rationality of each clustering result.

[0040] It should be noted that the inter-class distribution rationality is used to measure the distribution characteristics of data points between different categories and the boundary clarity, and to judge the scientific nature of the classification of each category in the clustering result. Therefore, in this step, the quantitative analysis of the inter-class distribution rationality of the clustering result is realized by comprehensively evaluating the sparse regions and cluster characteristics of each category.

[0041] Obtain the regular clusters and sparse clusters of the current category, as well as the sparse region of the current category. The regular clusters, sparse clusters, and sparse regions all contain several rectangles. Determine the inter-class distribution rationality of the target result according to the width and length of the sparse regions of each category in the target result, the number of rectangles in the sparse clusters, the number of rectangles in the sparse regions, the number of data points in the sparse regions, and the number of data points in each regular cluster.

[0042] Specifically, the obtaining of the regular clusters and sparse clusters of the current category includes: Obtain the distances between the current category and the within-class centroids of the remaining categories in the target result. Take the within-class centroid of the category that is closest to the current category as the midpoints of any two opposite sides of a square, and take the distance between the current category and the within-class centroid of the category that is closest to the current category as the side length of the square. Starting from any one side of the square, divide the square into multiple rectangles with the same area in sequence to obtain multiple rectangles of the current category; Based on the number of data points within the rectangles, use the K-means clustering algorithm to cluster the multiple rectangles of the current category to obtain two clusters. Denote the cluster with the largest number of data points within the cluster as the regular cluster, and the other as the sparse cluster, to obtain the regular cluster and the sparse cluster of the current category.

[0043] Implementers can set the number of rectangles according to the specific implementation situation. For example, 10.

[0044] Specifically, the method for obtaining the sparse region of the current category is as follows: Merge all adjacent rectangles that belong to the sparse cluster, and denote the region with the largest area after merging as the sparse region in the sparse cluster (if there are multiple regions with the largest area, arbitrarily select one of them).

[0045] Specifically, the inter-class distribution rationality satisfies: ; In the formula, is the inter-class distribution rationality of the target result, is the number of categories in the target result, and are respectively the width and length of the sparse region of category in the target result, and are respectively the number of the sparse region and the rectangles within the sparse cluster of category in the target result, is the number of data points within the sparse region of category in the target result, is the mean value of the number of data points within all rectangles in the regular cluster of category in the target result.

[0046] Among them, represents the fusion index that couples the spatial shape feature and the data density difference of category , quantifies the distribution rationality of the sparse region in the classification result. If there is no merging of rectangles within the sparse cluster of a certain category, the fusion index of this category is 0. When is larger, it indicates that the region where data points are sparse between category and the nearest other category is larger, indicating that category The greater the necessity for existence and the fewer the data points at the border with another category, the greater the rationality of the distribution between categories; when is smaller, it indicates that for the category the smaller the area with sparse data points between it and the nearest other category, and the more data points at the border with another category. It is very likely that these two categories should actually be classified as one category, and there may be an error in clustering, so the rationality of the distribution between categories is smaller. When is larger, it indicates that the sparse area occupies most of the sparse cluster, and the greater the possibility that the sparse area is the inherent boundary between the two categories, so the rationality of the distribution between categories is greater; if is larger, it indicates that the rectangles in the sparse category are discretely distributed, and it is difficult for the sparse area to represent the boundary, that is, the smaller the possibility of an obvious boundary between the two categories, so the rationality of the distribution between categories is smaller. When is larger, it indicates that the fewer the data points in the sparse area, that is, the greater the possibility that the sparse area is the inherent boundary between the two categories, so the rationality of the distribution between categories is greater; when is smaller, it indicates that the more data points the sparse area contains, that is, the smaller the possibility of an obvious boundary between the two categories, so the rationality of the distribution between categories is smaller.

[0047] S05: Determine the preference degree of each clustering result.

[0048] It should be noted that the preference degree comprehensively considers the compactness of data within the category, the separation degree between categories, as well as the density and distribution rationality, and comprehensively reflects the quality of the clustering result. Therefore, in this step, by calculating the preference degree, a quantitative basis is provided for screening the best clustering result.

[0049] Determine the preference degree of the target result according to the distances from the data points within each category in the target result to the intra-category center points of each category, the distances between the intra-category center points between categories, and the rationality of the intra-category density change and the inter-category distribution rationality.

[0050] Specifically, the preference degree satisfies: ; In the formula, is the preference degree of the target result, is the number of categories in the target result, is the number of data points in the category in the target result, is the distance from the th data point in the category in the target result to the intra-category center point of the category , is the distance between the intra-category center points of the category and the category in the target result, For the rationality of the intra-class density change of the target result, For the rationality of the inter-class distribution of the target result.

[0051] Among them, reflects the average distance between the intra-class data points and the center point. The smaller this value is, the more compact the intra-class data is; represents the average distance between the inter-class center points. The larger this value is, the higher the inter-class separation degree. The ratio of the two acts together with the rationality of the intra-class density change and the rationality of the inter-class distribution, enabling the preference degree to accurately evaluate the comprehensive quality of the clustering result.

[0052] S06: According to the magnitude of the preference degree, determine the best clustering result of the K-means clustering algorithm to implement the data processing of the power distribution device.

[0053] Specifically, the implementation of the data processing of the power distribution device includes: Take the clustering result corresponding to the maximum value among the preference degrees of all clustering results as the best clustering result of clustering the current area using the K-means clustering algorithm, thereby completing the classification of the user electricity consumption patterns in the current area. Based on this classification, power companies can implement differentiated electricity management, such as optimizing time-of-use electricity prices, recommending energy storage solutions, or detecting abnormal electricity consumption, thus completing the full-process data processing of the power distribution device from data collection to intelligent application, providing effective support for power grid optimization and demand-side management, and completing the data processing of the power distribution device.

[0054] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the principles of the present invention shall be included within the protection scope of the present invention.

Claims

1. A data processing method for a power distribution device, characterized in that Including: Based on the collected user electricity consumption data of the current area, constructing a data point set for the current area to reflect the electricity load, and performing multiple clusterings on the data points in the data point set by using the K-means clustering algorithm; Denoting any clustering result as the target result, denoting any category in the target result as the current category, constructing multiple concentric circles for the current category to obtain multiple rings for the current category, and determining the data density of each ring in the current category according to the number of data points in each ring and the area of each ring in the current category; determining the rationality of the within-class density change of the target result according to the data density of each ring in each category in the target result; obtaining the regular clusters and sparse clusters of the current category, and the sparse area of the current category, where the regular clusters, sparse clusters and sparse areas all contain several rectangles, and determining the rationality of the between-class distribution of the target result according to the width and length of the sparse areas in each category in the target result, the number of rectangles in the sparse clusters, the number of rectangles in the sparse areas, the number of data points in the sparse areas, and the number of data points in each regular cluster; determining the preference degree of the target result according to the distances from the data points in each category in the target result to the within-class center points of each category, the distances between the within-class center points of the categories, and the rationality of the within-class density change and the rationality of the between-class distribution; determining the best clustering result of the K-means clustering algorithm according to the magnitude of the preference degree, and realizing the data processing of the power distribution device.

2. The data processing method of a power distribution device according to claim 1, characterized in that, The construction of the data point set for the current area to reflect the electricity load includes: Taking the peak-hour electricity consumption ratio in the user electricity consumption data of the current area as the abscissa and the daily average electricity consumption as the ordinate to construct a two-dimensional coordinate system, and obtaining a data point set for the current area to reflect the electricity load.

3. The data processing method of a power distribution device according to claim 1, characterized in that, The obtaining of the multiple rings for the current category includes: Taking the within-class center point of the current category as the center of the circle, constructing the smallest circumscribed circle containing all the data points in the current category, constructing multiple concentric circles of the smallest circumscribed circle according to a preset scaling ratio, dividing the smallest circumscribed circle into multiple rings, and sorting the rings in order from the inside to the outside to obtain multiple rings for the current category.

4. The data processing method of a power distribution device according to claim 3, characterized in that, The data density satisfies: ； Wherein, is the data density of the current category of circular ring , is the number of data points within the current category of circular ring , is a preset scaling ratio is the serial number of the current category of circular ring is the radius of the minimum circumscribed circle of the current category is the maximum-minimum normalization function 5. The data processing method of a power distribution device according to claim 1, characterized in that, The rationality of the within-class density change satisfies: ； In the formula, is the rationality of the within-class density change of the target result, is the number of categories in the target result, is the number of circular rings of the categories in the target result, is the category in the target result of the circular ring is the data density of the category in the target result of the circular ring is the length of the preset moving unit.

6. The data processing method of a power distribution device according to claim 1, characterized in that The obtaining of the regular clusters and sparse clusters of the current category includes: Obtaining the distances between the within-class center points of the current category and the other categories in the target result, taking the within-class center points of the current category and the category closest to the current category as the midpoints of any two opposite sides of a square, taking the distance between the within-class center points of the current category and the category closest to the current category as the side length of the square, and starting from any side of the square, dividing the square into multiple rectangles with the same area in sequence to obtain multiple rectangles for the current category; Based on the number of data points in the rectangles, using the K-means clustering algorithm to cluster the multiple rectangles of the current category to obtain two clusters, denoting the cluster with the largest number of data points as the regular cluster, and the other as the sparse cluster, and obtaining the regular cluster and sparse cluster of the current category.

7. The data processing method of a power distribution device according to claim 6, characterized in that The obtaining method of the sparse area of the current category is: Merge all adjacent rectangles that belong to the sparse cluster, and record the region with the largest area after merging as the sparse region in the sparse cluster.

8. The data processing method of a power distribution device according to claim 1, characterized in that, The rationality of the inter-class distribution satisfies: ； In the formula, is the rationality of the inter-class distribution of the target result, is the number of categories in the target result, and are respectively the width and length of the sparse region of category in the target result, and are respectively the number of the sparse region and the rectangles within the sparse cluster of category in the target result, is the number of data points in the sparse region of category in the target result, is the average value of the number of data points within all rectangles in the regular cluster of category in the target result.

9. The data processing method of a power distribution device according to claim 1, characterized in that The preference degree satisfies: ； Wherein, is the preference degree of the target result, is the number of categories in the target result, is the category in the target result is the number of data points within the category, is the category in the target result is the th data point in the category to the within-class center point of the category distance, is the distance between the category and the within-class center point of the category in the target result, is the rationality of the within-class density change of the target result, is the rationality of the between-class distribution of the target result.

10. The data processing method of a power distribution device according to claim 1, characterized in that, The data processing of the power distribution device includes: Take the clustering result corresponding to the maximum value among the preference degrees of all clustering results as the best clustering result for clustering the current region using the K-means clustering algorithm, and complete the data processing of the power distribution device.

Images