Electric power acquisition terminal intelligent management system based on data processing

Abstract

The invention relates to the technical field of power data processing, in particular to a power acquisition terminal intelligent management system based on data processing. The system comprises a data acquisition module used for acquiring a load change rate and a voltage deviation at the same moment to form a data point; the sparseness calculation module is used for acquiring an initial convex hull of the data points and acquiring the sparseness of each data point in the initial convex hull; the sparse region acquisition module is used for constructing a first convex hull by using the data points of which the sparseness is smaller than the sparseness mean value; calculating the density gradient of the data points in the first convex hull; a second convex hull is constructed based on the density gradient, and then a sparse region is obtained; the data identification module is used for obtaining an optimized neighborhood radius and an optimized minimum point number according to the sparse region; and processing the data points by using a DBSCAN algorithm based on the optimized neighborhood radius and the optimized minimum point number. According to the method, the power data can be processed by using the algorithm after parameter optimization, and the purpose of managing the power data is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of power data processing, and in particular to a power collection terminal intelligent management system based on data processing. Background Art

[0002] As the power system develops towards intelligence and automation, power collection terminals, as a key component of the power system, have been widely used in the collection, monitoring and management of power data. These terminal devices play a core role in data collection and abnormal warning in the system, providing important data support for the safe and stable operation of the power system.

[0003] In the power collection terminal management system, abnormal data detection has always been a major problem. For key data such as voltage, load and current, the system needs to identify abnormal changes in a timely manner to ensure the safety and stability of power supply. Traditional algorithms such as DBSCAN are often used to process these types of data because their density-based clustering method can classify data points into high-density normal areas, and then identify sparsely distributed data as potential anomalies. Specifically, DBSCAN defines the density threshold of clustering by neighborhood radius and minimum number of points, thereby effectively distinguishing dense normal data points from sparse abnormal points. However, since power data is greatly affected by periodic fluctuations and load changes, the data distribution in different time periods may have special distribution characteristics such as non-uniform distribution or large density differences, making it difficult for the algorithm to find a suitable parameter (neighborhood radius and minimum point) to meet the clustering requirements in this case. Summary of the invention

[0004] In order to solve the above technical problems, the purpose of the present invention is to provide a smart management system for power collection terminals based on data processing. The technical solutions adopted are as follows: An embodiment of the present invention provides a smart management system for power collection terminals based on data processing, the system includes a data collection module, a sparsity calculation module, a sparse area acquisition module and a data identification module, specifically: The data acquisition module is used to collect the load change rate and voltage deviation and perform preprocessing. The load change rate and voltage deviation at the same time after preprocessing constitute a data point; A sparsity calculation module is used to obtain an initial convex hull of the data point, obtain the minimum distance between a data point and other data points in the initial convex hull, recorded as a first distance; and obtain the sparsity of the data point according to the first distance; A sparse region acquisition module is used to obtain the sparsity mean of all data points, construct a first convex hull using data points whose sparsity is less than the sparsity mean; calculate the density gradient of the data points in the first convex hull according to the first distance of the data points; construct a second convex hull using data points whose density gradient is less than the density gradient mean; and obtain a sparse region based on the initial convex hull and the second convex hull; The data recognition module is used to obtain the optimized neighborhood radius according to the density gradient mean of all data points in the sparse area and the area of ​​the sparse area; obtain the optimized minimum number of points according to the number of data points within the optimized neighborhood radius of each data point in the second convex hull; and process the data points using the DBSCAN algorithm based on the optimized neighborhood radius and the optimized minimum number of points.

[0005] Preferably, collecting the load change rate and voltage deviation and preprocessing them includes: The load change rate and voltage deviation are collected at the same sampling frequency; the collected load change rate and voltage deviation are preprocessed including filling missing data, smoothing abnormal fluctuations, eliminating extreme abnormal points and normalization.

[0006] Preferably, obtaining the initial convex hull of the data points includes: The horizontal and vertical coordinates corresponding to the data points in the coordinate system are the load change rate and voltage deviation respectively; the data point with the smallest horizontal and vertical coordinates is selected as the starting point; the polar angles between other data points and the starting point are calculated, and the data points are sorted from small to large according to the polar angles. Based on the sorting results, all data points are visited in a clockwise direction from the starting point, and the convex hull algorithm is used to find the minimum convex polygon of the data point, and the minimum convex polygon is the initial convex hull of the data point.

[0007] Preferably, the calculation formula of sparsity is: , in, Represents the sparsity of the i-th data point in the initial convex hull; represents the first distance of the i-th data point; Represents the mean of the first distances of all data points in the initial convex hull; represents the standard deviation of the first distance of all data points in the initial convex hull; norm() represents the normalization operation; The coefficient representing the fluctuation of the eigenvalue; and They represent the load change rate and voltage deviation of the i-th data point in the initial convex hull respectively; and They represent the mean value of load change rate and voltage deviation of all data points in the initial convex hull respectively.

[0008] Preferably, the calculation formula for the density gradient of a data point is: , in, represents the density gradient of the nth data point in the first convex hull; Represents the first distance of the nth data point in the first convex hull; represents the first distance of the mth data point in the first convex hull; e represents a natural constant; σ represents the standard deviation of the first distances of all data points in the first convex hull; max() represents the maximum value operation; AJ represents the set of data points in the initial convex hull whose sparsity is less than the mean sparsity; m represents the mth data point.

[0009] Preferably, the calculation formula for optimizing the neighborhood radius is: , in, represents the optimized neighborhood radius; represents the initial neighborhood radius; represents the area of ​​the initial convex hull; represents the area of ​​sparse regions; Represents the mean density gradient of all data points in the sparse region; and They represent the adjustment coefficient of the sparse area ratio and the adjustment coefficient of the density gradient respectively; Indicates rounding up.

[0010] Preferably, obtaining the optimized minimum number of points according to the number of data points within the neighborhood radius of each data point in the second convex hull includes: Obtain the area of ​​the region with the optimized neighborhood radius as the radius for each data point in the second convex hull and sum them up to obtain the neighborhood coverage area; compare the sum of the number of data points within the optimized neighborhood radius of each data point in the second convex hull with the neighborhood coverage area to obtain the neighborhood average distribution number; round up the neighborhood average distribution number, and the maximum value of the neighborhood average distribution number after rounding up and the preset value is the optimized minimum number of points.

[0011] Preferably, obtaining the sparse region according to the initial convex hull and the second convex hull includes: The initial convex hull is subtracted from the second convex hull to obtain the sparse region.

[0012] The embodiments of the present invention have at least the following beneficial effects: the data acquisition module of the present invention acquires two-dimensional power data, and uses the load change rate and the voltage deviation for analysis, which can better express the characteristics of the power data compared with one-dimensional data; further, in the sparsity calculation module, an initial convex hull of all data points is constructed, so as to analyze the sparsity of the data points in the initial convex hull, thereby obtaining the sparsity of each data point; according to the sparsity of the data points, screening is performed in the initial convex hull, relatively concentrated data points are screened out to construct a first convex hull, and then further screening is performed in the first convex hull, and a second convex hull is constructed according to the screened data points, which can ensure that the second convex hull is more in line with the aggregation characteristics of normal data, and thus a more accurate sparse area is obtained; then, analysis is performed in the sparse area and the second convex hull, and the neighborhood radius and the minimum number of points are optimized to obtain the optimized neighborhood radius and the optimized minimum number of points, and the data points of the power data are processed using the DBSCAN algorithm based on the optimized neighborhood radius and the optimized minimum number of points, which can more accurately identify abnormal points, improve the ability of the power acquisition terminal intelligent management system to identify abnormal data, and improve the demand for more accurate terminal intelligent management. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] In order to more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0014] Figure 1 A system block diagram of a power collection terminal intelligent management system based on data processing provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0015] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the specific implementation method, structure, features and effects of a smart management system for power collection terminals based on data processing proposed by the present invention are described in detail below in combination with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures, or characteristics in one or more embodiments may be combined in any suitable form.

[0016] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0017] The following is a detailed description of a specific solution of a smart management system for power collection terminals based on data processing provided by the present invention in conjunction with the accompanying drawings.

[0018] The main application scenarios of the present invention are: In view of the shortcomings of traditional methods, this application determines the sparse area of ​​data distribution by analyzing the distribution of power data points. Then calculate the degree of sparsity and dynamically adjust the parameters of the DBSCAN algorithm. This adaptive density adjustment mechanism enables the DBSCAN algorithm to maintain detection accuracy when the data has special distribution characteristics with non-uniform distribution or large density differences, thereby improving the ability of the power collection terminal intelligent management system to identify abnormal data and the demand for more accurate terminal intelligent management.

[0019] For power data, the DBSCAN algorithm cannot select a suitable density value, resulting in a decrease in recognition accuracy. To address this problem, the present invention analyzes the geometric characteristics of data distribution and adaptively adjusts the density of the DBSCAN algorithm according to the geometric characteristics, thereby improving the accuracy of the algorithm for anomaly detection.

[0020] See also Figure 1 , which shows a system block diagram of a power collection terminal intelligent management system based on data processing provided by an embodiment of the present invention, the system includes the following modules: The data acquisition module is used to collect the load change rate and voltage deviation and perform preprocessing. The load change rate and voltage deviation at the same time after preprocessing constitute a data point; A sparsity calculation module is used to obtain an initial convex hull of the data point, obtain the minimum distance between a data point and other data points in the initial convex hull, recorded as a first distance; and obtain the sparsity of the data point according to the first distance; A sparse region acquisition module is used to obtain the sparsity mean of all data points, construct a first convex hull using data points whose sparsity is less than the sparsity mean; calculate the density gradient of the data points in the first convex hull according to the first distance of the data points; construct a second convex hull using data points whose density gradient is less than the density gradient mean; and obtain a sparse region based on the initial convex hull and the second convex hull; The data recognition module is used to obtain the optimized neighborhood radius according to the density gradient mean of all data points in the sparse area and the area of ​​the sparse area; obtain the optimized minimum number of points according to the number of data points within the optimized neighborhood radius of each data point in the second convex hull; and process the data points using the DBSCAN algorithm based on the optimized neighborhood radius and the optimized minimum number of points.

[0021] Specific: The data acquisition module is used to collect the load change rate and voltage deviation and perform preprocessing. The load change rate and voltage deviation at the same time after preprocessing constitute a data point.

[0022] In order to reduce the error caused by the original data on the abnormality detection effect, the present invention first collects the original data and performs data preprocessing.

[0023] Real-time power data is obtained from the power collection terminal, including the key parameters load change rate and voltage deviation. Abnormal fluctuations in load and voltage are usually accompanied by system failures, equipment abnormalities or operational errors. Therefore, the load change rate and voltage deviation can sensitively capture abnormal conditions in the system and are very reliable features in anomaly detection. These data are usually collected in the form of time series to reflect the real-time operating status of the power system. The load change rate can be calculated by the load difference between adjacent time points, and the voltage deviation can be calculated as the deviation between the current voltage and the reference voltage. The collection frequency of the load change rate and the voltage deviation is the same, and the implementer can set it according to the actual situation.

[0024] The collected power data is preprocessed, including removing noise, missing values ​​and outliers. Specifically, it includes filling missing data, smoothing abnormal fluctuations and eliminating extreme abnormal points, and then normalizing the load change rate and voltage deviation characteristics to ensure that the dimensions of the two are consistent in subsequent calculations. After the preprocessing is completed, the load change rate and voltage deviation at the same time constitute a data point. In this way, the power data can be converted into data points one by one.

[0025] The sparsity calculation module is used to obtain the initial convex hull of the data point, obtain the minimum distance between a data point and other data points in the initial convex hull, recorded as the first distance; and obtain the sparsity of the data point according to the first distance.

[0026] In the anomaly detection of power collection terminal data, in order to accurately characterize the distribution boundary of the data and identify the sparse area, the convex hull algorithm is used to draw the minimum convex polygon on the preprocessed data. The convex hull is the minimum closed boundary containing all data points, which can effectively wrap the overall distribution of the data. By constructing the minimum convex hull of the data points, the outer boundary of the data distribution can be identified for subsequent sparsity calculation and density adaptive adjustment. In this way, the distribution information of all power data points can be obtained, and an initial convex polygon (i.e., convex hull) containing all the data can be constructed based on these points. In this initial convex hull area, its sparsity is calculated.

[0027] DBSCAN defines the neighborhood by two parameters: the neighborhood radius ε and the minimum number of points minpts. When a point contains at least minpts points in a neighborhood with a radius of ε, the point is called a core point. The area where the core point is located is a data-intensive area and can constitute the core of clustering. Starting from the core point, DBSCAN aggregates all density-connected points into clusters through neighborhood expansion. If there are other points in the neighborhood of the core point, these points are also included in the cluster. If there are other core points, the neighborhood continues to expand until no new points can be found. This process aggregates all density-connected points into a cluster. If a point is neither a core point nor in the neighborhood of any core point, DBSCAN marks it as a noise point or an outlier. These points cannot be classified into any cluster due to their low density and are suitable for anomaly detection and outlier identification.

[0028] Furthermore, the convex hull algorithm is used to draw the minimum convex polygon according to the data points to obtain the initial convex hull of the data points. Specifically, the set of acquired data points is set to , where each data point It represents a two-dimensional feature, and its corresponding horizontal and vertical coordinates in the coordinate system are load change rate and voltage deviation respectively. The implementer can determine the selection of horizontal and vertical coordinates according to actual conditions.

[0029] Select the data point with the smallest horizontal and vertical coordinates from the set as the starting point, denoted as , calculate except Other data points besides the data point The polar angle formed , the calculation formula of the polar angle is: , Among them, the coordinates They are The corresponding coordinates are sorted from small to large polar angles to ensure that Start from, visit all data points in clockwise order according to the sorting results, and use the convex hull algorithm to find the minimum convex polygon of the data points. The minimum convex polygon is the initial convex hull D of the data points.

[0030] To analyze the data sparsity in the initial convex hull and find sparse areas, it is first necessary to evaluate the distribution density of the data inside the polygon. Generally speaking, dense areas have more data points and are closely distributed, while sparse areas have relatively fewer points and are widely spaced. Therefore, the present invention measures the distance between data points to count the point density of each local area to determine the sparsity. First, the distance between each data point and the nearest data point is calculated, so that the density of each data point in the initial convex hull and the overall density can be calculated. The areas with low density correspond to locations with higher sparsity, and these areas may be abnormal areas in the data.

[0031] Furthermore, the minimum distance between a data point and other data points in the initial convex hull is obtained, recorded as the first distance, and then the sparsity of each data point in the initial convex hull is calculated according to the first distance. Specifically, the sparsity calculation formula of the i-th data point can be expressed as: , in, Represents the sparsity of the i-th data point in the initial convex hull; represents the first distance of the i-th data point; Represents the mean of the first distances of all data points in the initial convex hull; represents the standard deviation of the first distance of all data points in the initial convex hull; norm() represents the normalization operation; The coefficient representing the fluctuation of the eigenvalue; and They represent the load change rate and voltage deviation of the i-th data point in the initial convex hull respectively; and They represent the mean value of load change rate and voltage deviation of all data points in the initial convex hull respectively.

[0032] In the above sparsity calculation formula, It represents the deviation of the distance from the i-th point in the initial convex hull to its nearest data point and the global distance mean, which enhances the perception of the sparsity of the area where the data points are unevenly distributed. Significantly greater than When , it means that the data point may be in a sparse area, so this factor will increase the sparsity value, otherwise it will reduce the sparsity. represents the weighting factor of the eigenvalue volatility, Indicates the fluctuation of load change rate and voltage deviation. The sparsity calculation formula introduces the distance mean deviation and volatility weighting factor, so that the sparsity calculation not only focuses on the local density of the point, but also adapts to the changes in the global data distribution. As increases, the sparsity value of the point will also increase accordingly, making the abnormal area easier to detect.

[0033] In this way, the sparsity of each data point in the initial convex hull can be obtained for subsequent analysis.

[0034] The sparse area acquisition module is used to obtain the sparsity mean of all data points, construct a first convex hull using data points whose sparsity is less than the sparsity mean; calculate the density gradient of the data points in the first convex hull according to the first distance of the data points; construct a second convex hull using data points whose density gradient is less than the density gradient mean; and obtain a sparse area based on the initial convex hull and the second convex hull.

[0035] Further, the sparsity mean of all data points in the initial convex hull is obtained, and the data points with sparsity less than the sparsity mean are screened out to form a first set AJ. A convex hull is constructed for the area formed by the data points in the first set, which is recorded as the first convex hull A. The first convex hull belongs to the dense area where the data points are distributed. At this time, the sparse area can be expressed as DA. However, there may be problems with the sparse area obtained at this time. If the distribution of abnormal data is relatively concentrated, the density value finally calculated using the sparse area will be too large. In order to solve this problem, the present invention calculates the local gradient for the area (first convex hull) with less than the average sparsity. Specifically, the density gradient of the data point is calculated according to the first distance of the data point in the first convex hull. The calculation formula of the density gradient of the data point is specifically: , in, represents the density gradient of the nth data point in the first convex hull; Represents the first distance of the nth data point in the first convex hull; represents the first distance of the mth data point in the first convex hull; e represents a natural constant; σ represents the standard deviation of the first distances of all data points in the first convex hull; max() represents the maximum value operation; AJ represents the set of data points in the initial convex hull whose sparsity is less than the mean sparsity; m represents the mth data point.

[0036] In the above formula Represents the difference between the Gaussian kernels of the nth data point and the mth data point in the first convex hull, It means to find the maximum value of the Gaussian kernel difference between the nth data point and other data points in the first convex hull, then The value can reflect the uniformity of distribution among the points in the first convex hull.

[0037] Furthermore, the density gradient mean of all data points in the first convex hull is obtained, and the data points whose density gradient is less than the density gradient mean of all data points are marked, and the convex hull is constructed based on the convex hull algorithm using these data points to obtain the second convex hull. Finally, the sparse area DH is obtained based on the initial convex hull and the second convex hull, and the coefficient area is obtained by subtracting the initial convex hull from the second convex hull.

[0038] The data recognition module is used to obtain the optimized neighborhood radius according to the density gradient mean of all data points in the sparse area and the area of ​​the sparse area; obtain the optimized minimum number of points according to the number of data points within the optimized neighborhood radius of each data point in the second convex hull; and process the data points using the DBSCAN algorithm based on the optimized neighborhood radius and the optimized minimum number of points.

[0039] After determining the sparse regions in the sparse region acquisition module, the density is adapted to different sparse regions by adjusting the parameters of the DBSCAN algorithm to improve the accuracy of anomaly detection.

[0040] The optimized neighborhood radius is obtained according to the density gradient mean of all data points in the sparse area and the area of ​​the sparse area. The specific calculation formula for the optimized neighborhood radius is: , in, represents the optimized neighborhood radius; represents the initial neighborhood radius; represents the area of ​​the initial convex hull; represents the area of ​​sparse regions; Represents the mean density gradient of all data points in the sparse region; and They represent the adjustment coefficient of the sparse area ratio and the adjustment coefficient of the density gradient respectively; Indicates rounding up.

[0041] It indicates the influence of the proportion of sparse area on the initial neighborhood radius. The larger the proportion of sparse area, the wider the distribution of sparse data. At this time, it is necessary to appropriately increase the neighborhood radius to make the clustering of normal data more accurate. On the contrary, if the proportion of sparse area is small, it means that the distribution of data is denser, and the increase in the neighborhood radius is relatively small to prevent misclassification due to excessively large neighborhood radius. It indicates the influence of average density on the initial neighborhood radius. The larger the average density of the sparse area, the larger the corresponding neighborhood radius value should be. The reference values ​​of 𝜌 and 𝜏 can be 0.1 (which can be adjusted according to actual conditions), and the calculated 𝜀 needs to be rounded up.

[0042] Next, it is necessary to obtain the optimized minimum number of points according to the number of data points within the optimized neighborhood radius of each data point in the second convex hull. Specifically, the area of ​​the region (circle) with the optimized neighborhood radius as the radius of each data point in the second convex hull is obtained and summed to obtain the neighborhood coverage area; the sum of the number of data points within the optimized neighborhood radius of each data point in the second convex hull is compared with the neighborhood coverage area to obtain the neighborhood average distribution number; the neighborhood average distribution number is rounded up, and the maximum value of the neighborhood average distribution number after rounding up and the preset value is the optimized minimum number of points minpts; the calculation formula for the optimized minimum number of points minpts is specifically as follows: , Among them, minpts represents the minimum number of optimized points; max() represents the maximum value operation; 2 represents the preset value; N represents the total number of data points in the second convex hull (dense area) H; is pi; represents the optimized neighborhood radius; Represents the number of data points within the optimized neighborhood radius of the ath data point in the second convex hull. Represents the local density value within the optimization neighborhood radius.

[0043] Finally, the DBSCAN algorithm is executed on the power data using the obtained optimized neighborhood radius and the optimized minimum number of points, which can achieve accurate detection of power data anomalies. By dynamically adjusting the neighborhood radius and the minimum number of points, DBSCAN can adapt to different data distribution characteristics: accurate clustering is achieved in data-dense areas to reduce misjudgments; isolated anomalies are sensitively identified in sparse areas. The adaptive density mechanism proposed in the present invention enables the DBSCAN algorithm to flexibly process the distribution characteristics of power data, thereby improving the accuracy of anomaly detection and providing smarter and more reliable support for the intelligent management of power collection terminals.

[0044] It should be noted that the sequence of the above embodiments of the present invention is only for description and does not represent the advantages and disadvantages of the embodiments. The above is a description of a specific embodiment of this specification. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0045] The various embodiments in this specification are described in a progressive manner, and the same or similar parts between the various embodiments can be referenced to each other, and each embodiment focuses on the differences from other embodiments.

[0046] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A smart management system for power collection terminals based on data processing, characterized in that: The system includes: The data acquisition module is used to collect the load change rate and voltage deviation and perform preprocessing. The load change rate and voltage deviation at the same time after preprocessing constitute a data point; A sparsity calculation module is used to obtain an initial convex hull of the data point, obtain the minimum distance between a data point and other data points in the initial convex hull, recorded as a first distance; and obtain the sparsity of the data point according to the first distance; A sparse region acquisition module is used to obtain the sparsity mean of all data points, construct a first convex hull using data points whose sparsity is less than the sparsity mean; calculate the density gradient of the data points in the first convex hull according to the first distance of the data points; construct a second convex hull using data points whose density gradient is less than the density gradient mean; and obtain a sparse region based on the initial convex hull and the second convex hull; The data recognition module is used to obtain the optimized neighborhood radius according to the density gradient mean of all data points in the sparse area and the area of ​​the sparse area; obtain the optimized minimum number of points according to the number of data points within the optimized neighborhood radius of each data point in the second convex hull; and process the data points using the DBSCAN algorithm based on the optimized neighborhood radius and the optimized minimum number of points.

2. According to claim 1, a power collection terminal intelligent management system based on data processing is characterized in that: The collecting of load change rate and voltage deviation and preprocessing includes: The load change rate and voltage deviation are collected at the same sampling frequency; the collected load change rate and voltage deviation are preprocessed including filling missing data, smoothing abnormal fluctuations, eliminating extreme abnormal points and normalization.

3. According to the data processing-based intelligent management system for power collection terminals according to claim 1, it is characterized in that: The step of obtaining the initial convex hull of the data points includes: The horizontal and vertical coordinates corresponding to the data points in the coordinate system are the load change rate and voltage deviation respectively; the data point with the smallest horizontal and vertical coordinates is selected as the starting point; the polar angles between other data points and the starting point are calculated, and the data points are sorted from small to large according to the polar angles. Based on the sorting results, all data points are visited in a clockwise direction from the starting point, and the convex hull algorithm is used to find the minimum convex polygon of the data point, and the minimum convex polygon is the initial convex hull of the data point.

4. According to the data processing-based intelligent management system for power collection terminals according to claim 1, it is characterized in that: The calculation formula of the sparsity is: ， in, Represents the sparsity of the i-th data point in the initial convex hull; represents the first distance of the i-th data point; Represents the mean of the first distances of all data points in the initial convex hull; represents the standard deviation of the first distance of all data points in the initial convex hull; norm() represents the normalization operation; The coefficient representing the fluctuation of the eigenvalue; and They represent the load change rate and voltage deviation of the i-th data point in the initial convex hull respectively; and They represent the mean value of load change rate and voltage deviation of all data points in the initial convex hull respectively.

5. According to the data processing-based intelligent management system for power collection terminals according to claim 1, it is characterized in that: The calculation formula of the density gradient of the data point is: ， in, represents the density gradient of the nth data point in the first convex hull; Represents the first distance of the nth data point in the first convex hull; represents the first distance of the mth data point in the first convex hull; e represents a natural constant; σ represents the standard deviation of the first distances of all data points in the first convex hull; max() represents the maximum value operation; AJ represents the set of data points in the initial convex hull whose sparsity is less than the mean sparsity; m represents the mth data point.

6. According to the data processing-based intelligent management system for power collection terminals of claim 1, it is characterized in that: The calculation formula for the optimized neighborhood radius is: ， in, represents the optimized neighborhood radius; represents the initial neighborhood radius; represents the area of ​​the initial convex hull; represents the area of ​​sparse regions; Represents the mean density gradient of all data points in the sparse region; and They represent the adjustment coefficient of the sparse area ratio and the adjustment coefficient of the density gradient respectively; Indicates rounding up.

7. The intelligent management system for power collection terminals based on data processing according to claim 1 is characterized in that: The step of optimizing the number of data points within the neighborhood radius of each data point in the second convex hull to obtain the minimum number of optimized points includes: Obtain the area of ​​the region with the optimized neighborhood radius as the radius for each data point in the second convex hull and sum them up to obtain the neighborhood coverage area; compare the sum of the number of data points within the optimized neighborhood radius of each data point in the second convex hull with the neighborhood coverage area to obtain the neighborhood average distribution number; round up the neighborhood average distribution number, and the maximum value of the neighborhood average distribution number after rounding up and the preset value is the optimized minimum number of points.

8. The power collection terminal intelligent management system based on data processing according to claim 1 is characterized in that: The step of obtaining a sparse region according to the initial convex hull and the second convex hull includes: The initial convex hull is subtracted from the second convex hull to obtain the sparse region.

Images