A 5G base station clustering analysis method based on geographical location distribution and spatio-temporal distribution characteristics of energy consumption

Abstract

The present invention discloses a 5G base station clustering analysis method based on geographical location distribution and energy consumption spatio-temporal distribution characteristics, comprising the following steps: S1: Obtain 5G base station information; S2: Cluster according to the principle of proximity in geographical location; S3: Cluster according to the principle of similarity in energy consumption curves; S4: Result analysis. The beneficial effects of the present invention are: It can realize the clustering work of regional 5G base stations, and realize the analysis of the virtual load power consumption characteristics and demand response potential analysis of 5G base stations.

Description

Technical Field

[0001] The present invention belongs to the field related to clustering analysis in new energy and power demand side response, and relates to a 5G base station clustering analysis method based on geographical location distribution and energy consumption spatio-temporal distribution characteristics. Background Art

[0002] With the rapid development of new energy power generation, the power system's demand for flexible regulation resources is becoming increasingly urgent. User-side resources are gradually participating in various operations of power grid companies, and there are more and more interactive regulations between load, power source, and power grid. However, large-scale user-side resources have challenges such as severely differentiated behavioral characteristics, uneven load response situations, and diverse and scattered interest subjects. In the 21st century, 5G communication technology has developed rapidly, and its extremely high energy consumption makes it highly willing to reduce electricity costs.

[0003] Currently, 5G base stations are usually equipped with lithium iron phosphate batteries with long service life, high energy density, and fast charge and discharge, and have advantages such as a single interest subject, similar load characteristics, and convenient centralized management. They have great demand response potential and are good user-side demand response resources. However, 5G base stations have characteristics such as small single-station power and massive distribution. Therefore, seeking an effective big data analysis method for 5G base stations is of great research significance for 5G base stations to participate in power grid demand response. On this basis, deeply exploring their typical electricity consumption characteristics and adjustable potential is an urgent need for the friendly and refined interaction between 5G base stations and the power grid. Research on 5G base stations participating in power grid demand response is still in its initial stage. According to investigations, no research has studied the aggregation method for the massive distribution characteristics of base stations. However, in demand response optimization decisions, if the granularity is refined to each 5G base station, the model scale will be too large to solve. In addition, the typical electricity consumption characteristics of 5G base stations have not been analyzed.

[0004] For example, an "Optimized Scheduling Method for a Large-Scale 5G Base Station to Participate in Distribution Network Demand Response" disclosed in a Chinese patent document, with the publication number: CN114006399A and the application date: September 30, 2021. This invention effectively reduces the energy consumption cost of base stations and improves the stability of the energy storage output of base stations, but it cannot achieve the clustering of regional 5G base stations, nor can it achieve the analysis of the virtual load electricity consumption characteristics and demand response potential of 5G base stations. Summary of the Invention

[0005] Aiming at the deficiencies of the prior art that it cannot achieve the clustering of regional 5G base stations, nor can it achieve the analysis of the virtual load electricity consumption characteristics and demand response potential of 5G base stations, the present invention proposes a 5G base station clustering analysis method based on geographical location distribution and energy consumption spatio-temporal distribution characteristics, which can achieve the clustering of regional 5G base stations and the analysis of the virtual load electricity consumption characteristics and demand response potential of 5G base stations.

[0006] The following is the technical solution of the present invention. A 5G base station clustering analysis method based on geographical location distribution and energy consumption spatio-temporal distribution characteristics includes the following steps:

[0007] S1: Obtain 5G base station information;

[0008] S2: Cluster according to the principle of proximity in geographical location;

[0009] S3: Cluster according to the principle of similarity in energy consumption curves;

[0010] S4: Result analysis. Cluster according to geographical location and energy consumption curves, and analyze the electricity consumption characteristics and demand response potential.

[0011] Preferably, the clustering object in step S2 is the 5G base station, and the clustering result is the initial clustering group; the clustering object in step S3 is the initial clustering group, and the clustering result is the final clustering group. First, cluster according to geographical location, and then cluster according to the energy consumption curve based on the clustering result of geographical location.

[0012] Preferably, in step S2, the method of clustering according to the principle of proximity in geographical location includes the following steps:

[0013] S21: Select and initialize several centroids;

[0014] S22: Allocate each 5G base station according to the principle of the closest Euclidean distance;

[0015] S23: Update the centroid according to the minimum sum of the squares of the distances between all 5G base stations and the clustering center;

[0016] S24: Obtain the initial clustering group. Update the centroid according to the principle of the closest Euclidean distance to obtain the initial clustering group.

[0017] Preferably, in step S21, the method of selecting the centroid includes the following steps:

[0018] S211: Select several centroids, and use different centroid values for clustering respectively;

[0019] S212: Calculate the corresponding distortion degree values according to the clustering;

[0020] S213: Plot a graph with the centroid as the abscissa and the corresponding distortion degree value as the ordinate;

[0021] S214: The abscissa centroid corresponding to the turning point is the corresponding optimal clustering value. Obtain the optimal clustering value according to the graph of the distortion degree value.

[0022] Preferably, in step S21, the method of initializing the centroid includes the following steps:

[0023] S216: Convert the longitude and latitude coordinates of the 5G base station into three-dimensional space coordinates;

[0024] S217: Randomly select the three-dimensional space coordinates of a 5G base station;

[0025] S218: Select the second coordinate according to the principle of being the farthest from the first coordinate;

[0026] S219: Select the third coordinate according to the principle of being the farthest from the first coordinate and the second coordinate, and so on. Randomly select an initial clustering center θ1, then select θ2 according to the principle of being the farthest from θ1, and then select θ3 according to the principle of being the farthest from θ1 and θ2.

[0027] Preferably, in step S3, the method of clustering according to the similarity principle of the energy consumption curve includes the following steps:

[0028] S31: Initialize the centroid;

[0029] S32: Allocate each initial clustering group according to the principle of the smallest difference in energy consumption curves;

[0030] S33: Update the centroid according to the minimum sum of the squares of the differences in the energy consumption curves between the centroids of all initial clustering groups and the clustering center;

[0031] S34: Obtain the final clustering group. Update the centroid according to the principle of the smallest difference in energy consumption curves to obtain the final clustering group.

[0032] Preferably, the centroid of the final clustering group is marked as the virtual load of the 5G base station for analyzing the electricity consumption characteristics and demand response potential.

[0033] Preferably, in step S213, the calculation formula of the distortion degree value is:

[0034]

[0035] In the formula, is the symbol for calculating the Euclidean geometric distance; x i is the geographical coordinate of the i-th 5G base station in the k-th clustering group; θ k is the geographical location coordinate of the centroid in the k-th clustering group; n k is the number of 5G base stations included in the k-th clustering group, and m1 is the number of centroids.

[0036] Preferably, in step S23, the calculation formula for the minimum sum of squared distances is:

[0037]

[0038] In the formula, is the symbol for calculating the Euclidean geometric distance, x iThe geographical coordinates of the i-th 5G base station in the k-th cluster group, θ k The geographical location coordinates of the centroid in the k-th cluster group, x k The class represented by the clustering center, and i is the current class.

[0039] Preferably, in step S33, the calculation formula for the minimum sum of squared differences of the energy consumption curves is:

[0040]

[0041] Among them, and are respectively the power consumption of the 5G base station θ i and μ k at time t; t max is the total time of one-day energy consumption sampling, m k is the class represented by the clustering center, and i is the current class.

[0042] The beneficial effects of the present invention are:

[0043] 1. Cluster 5G base stations within a certain range to complete the clustering work of regional 5G base stations;

[0044] 2. It is beneficial to solve the electrical connection and communication connection problems of 5G base station aggregation, is easy to manage the interaction with the power grid, and can design appropriate scheduling strategies to achieve the coordinated interaction between the virtual load of regional 5G base stations and the power grid;

[0045] 3. It can be applied to the participation of 5G base stations in power grid demand response, and can realize the analysis of the electricity consumption characteristics of 5G base station virtual loads and the analysis of demand response potential. Description of the Drawings

[0046] Figure 1 The method flow chart of a 5G base station clustering analysis method based on geographical location distribution and energy consumption spatio-temporal distribution characteristics provided by the present invention.

[0047] Figure 2 The first-layer clustering and second-layer clustering flow charts of a 5G base station clustering analysis method based on geographical location distribution and energy consumption spatio-temporal distribution characteristics provided by the present invention.

[0048] Figure 3 The clustering result diagram of a 5G base station clustering analysis method based on geographical location distribution and energy consumption spatio-temporal distribution characteristics provided by the present invention.

[0049] Figure 4 The electricity consumption characteristic analysis diagram of a 5G base station clustering analysis method based on geographical location distribution and energy consumption spatio-temporal distribution characteristics provided by the present invention. Detailed Embodiments

[0050] The technical solution of the present invention will be further specifically described below through embodiments in conjunction with the accompanying drawings. In addition, in order to better illustrate the present invention, numerous specific details are given below. Those skilled in the art should understand that the present invention can also be implemented without some specific details. In some instances, methods and means well-known to those skilled in the art are not described in detail in order to highlight the gist of the present invention.

[0051] As Figure 1 shown, a 5G base station clustering analysis method based on geographical location distribution and spatio-temporal distribution characteristics of energy consumption includes the following steps:

[0052] Step 1: Obtain the geographical location distribution (latitude and longitude coordinates) and daily energy consumption (daily load curve) of n 5G base stations in the area. For the daily load curve, if sampled every 60 minutes, it is 24 points.

[0053] Step 2: As Figure 2 shown, perform the first-layer K-means++ clustering of 5G base stations in the area according to the principle of proximity of geographical locations. The clustering objects are n 5G base stations in the area, the number of clusters is m1, and the distance metric is the Euclidean distance. The input is the latitude and longitude coordinates of n 5G base stations in the area and the number of clusters m1, and the output is the initial clustering clusters of m1 classes of 5G base stations and their centroids.

[0054] Step 21: Initialize m1 centroids θ0 = {θ1, θ2,..., θ m1}. First, convert the latitude and longitude coordinates of the 5G base station into three-dimensional space coordinates, randomly select an initial clustering center θ1, then select θ2 according to the principle of the farthest distance from θ1, then select θ3 according to the principle of the farthest distance from θ1 and θ2, and so on.

[0055] The selection method of m1 centroids is the "elbow method". K-means takes minimizing the squared error between the samples and the centroids as the objective function. The sum of the squared distance errors between the centroid of each clustering cluster and the sample points within the clustering cluster is called the distortion degree. Then, for a clustering cluster, the lower its distortion degree, the closer the members within the clustering cluster are, and the higher the distortion degree, the looser the structure within the clustering cluster. The distortion degree will decrease with the increase in the number of classes, but for data with a certain degree of discrimination, at a certain critical point, the distortion degree will be greatly improved, and this critical point can be considered as a point with better clustering performance. The specific method steps are as follows:

[0056] Step 211: Select m1 = 1 to 500, and perform clustering using different m1 values respectively.

[0057] Step 212: Calculate the corresponding distortion degree values according to the clustering.

[0058] Step 213: Plot a graph with the abscissa \(m1 = 1\) to \(500\) and the ordinate being the corresponding distortion degree value;

[0059] The abscissa \(m1\) corresponding to the turning point is the corresponding optimal clustering value.

[0060] The calculation formula for the distortion degree is:

[0061]

[0062] In the formula, is the symbol for calculating the Euclidean geometric distance; \(x\) i is the geographical coordinate of the \(i\)-th 5G base station in the \(k\)-th clustering group; \(\theta\) k is the geographical location coordinate of the centroid in the \(k\)-th clustering group; \(n\) k is the number of 5G base stations included in the \(k\)-th clustering group.

[0063] Step 22: For the 5G base stations \(x\) i in the area, according to their Euclidean distances from these clustering centers, assign them to the class represented by the clustering center \(k\) closest to them respectively according to the nearest distance criterion, \(x\) k as shown in the following formula:

[0064]

[0065] In the formula, is the symbol for calculating the Euclidean geometric distance, \(x\) i is the geographical coordinate of the \(i\)-th 5G base station in the \(k\)-th clustering group, \(\theta\) k is the geographical location coordinate of the centroid in the \(k\)-th clustering group.

[0066] Step 23: Update the centroid. Calculate according to the sum of the squares of the distances from all 5G base stations in the class \(x\) k to the new clustering center \(\theta\) k being the smallest, as shown in the following formula:

[0067]

[0068] In the formula, is the symbol for calculating the Euclidean geometric distance, \(x\) i is the geographical coordinate of the \(i\)-th 5G base station in the \(k\)-th clustering group, \(\theta\) k is the geographical location coordinate of the centroid in the \(k\)-th clustering group, \(x\) k is the class represented by the clustering center, and \(i\) is the current class.

[0069] Step 24: Determine whether the clustering center has changed. If it has changed, repeat the above steps 22 to 24; if not, go to step 25;

[0070] Step 25: The first layer of iteration ends. Obtain the initial clustering clusters x = {x1, x2,... x m1}, and their centroids are θ = {θ1, θ2,..., θ m1}.

[0071] Step 3: As Figure 2 shown, perform the second-layer clustering of K-means++ for the 5G base stations in the area clustered according to the similarity principle of the energy consumption curve. The clustering objects are m1 initial clustering clusters, the number of clusters is m, and the distance metric is the sum of squared differences dp of the daily electricity load i-j . The calculation formula is shown as follows. The input is m1 initial clustering clusters and their centroids, the number of clusters m, and the output is the final clustering clusters of m classes of 5G base stations and their centroids;

[0072]

[0073] where dp i-j is the sum of squared differences of the daily electricity load between the i-th 5G base station and the j-th 5G base station, and are the electricity powers of the 5G base station x i and x j at time t respectively, t max is the total number of sampling times of the daily energy consumption. If the electricity power of the 5G base station is sampled once every 15 minutes, then t max = 96.

[0074] Step 31: Initialize the centroid μ0 = {μ1, μ2,... μ m}. Select from θ = {θ1, θ2,..., θ m1} according to the principle of the largest sum of squared differences in energy consumption characteristics;

[0075] Step 32: For x i in the initial clustering cluster x, according to the criterion of the smallest sum of squared differences in the energy consumption curves between its clustering center θ i and these clustering centers μ0, assign it to the class m k represented by the nearest clustering center respectively;

[0076] Step 33: Update the centroid. Calculate according to the smallest sum of squared differences in the energy consumption curves between all points in the class m k and the new clustering center μ k , as shown in the following formula:

[0077]

[0078] where, and are the electricity powers of the 5G base station θ i and μ kPower consumption; t max is the total sampling time of daily energy consumption. For example, if the power consumption of a 5G base station is sampled every 15 minutes, then t max = 96, m k is the class represented by the cluster center, and i is the label of the 5G base station.

[0079] Step 34: Determine whether the cluster center has changed. If it has changed, repeat the above steps 32 to 34; if not, go to step 35;

[0080] Step 35: The iteration ends. As Figure 3 shown, the final cluster group M = {m1, m2,..., m m} is obtained, where m i contains multiple initial cluster groups in x, and the centroids are μ = {μ1, μ2,... μ m}.

[0081] Step 4: Analyze the virtual load power consumption characteristics and demand response potential of the 5G base stations in the clustered area based on the clustering results. Mark the centroid of the final cluster group as the typical virtual load of this type of 5G base station, and analyze its power consumption characteristics and demand response potential.

[0082] Mark the centroid of the final cluster group as the typical virtual load of the aggregated group of this type of 5G base station. According to its daily power load curve, the power consumption characteristics and demand response potential of this aggregated group can be analyzed. The demand response potential is mainly obtained based on the magnitude of the DC power consumption of 5G base stations during peak and high peak electricity price periods. The energy storage of 5G base stations can only supply power to DC loads and cannot supply power to the large power grid. When the power consumption characteristics of the aggregated group show a relatively high DC load power consumption during peak and high peak electricity price periods, the demand response potential of this type of 5G base station is relatively large.

[0083] Example:

[0084] According to the regional 5G base station two - layer K - means++ clustering analysis method that comprehensively considers geographical location distribution and energy consumption spatio - temporal distribution characteristics, cluster 5000 5G base stations with different geographical location distributions and energy consumption characteristics in a certain area. Perform programming operations based on MATLAB R2020a, and finally form 4 cluster groups. The clustering results are as Figure 2 shown.

[0085] To verify the effectiveness of the double-layer K-means++ clustering analysis method for regional 5G base stations that comprehensively considers the geographical location distribution and the spatio-temporal distribution characteristics of energy consumption, a control group 1 is set up: the K-means++ clustering algorithm based only on the principle of the nearest geographical location; and a control group 2: the K-means++ clustering algorithm based only on the principle of similar energy consumption characteristics. The compactness (CP) index and the separation (SP) index are used to evaluate the clustering results. The comparison of the clustering result evaluation indexes of the algorithm and the two control groups is shown in Table 1.

[0086] Table 1 Control experiment table.

[0087]

[0088] As Figure 4 shown, based on the above clustering calculations, the virtual load power consumption characteristics of the regional 5G base stations after clustering can be analyzed. The clustering centers of the 4 final clusters are marked as the typical virtual loads of the 5G base station aggregation group, and their power consumption characteristics are analyzed. The power consumption characteristics of the clustering centers of the 4 clusters obtained by the clustering calculation. It can be seen that the demand response dispatchable capacity potential of the 5G base stations in the first category is relatively large, followed by the second category, and then the third and fourth categories.

[0089] The above content is only the specific implementation manner of this application, but the protection scope of this application is not limited thereto. Any person skilled in the art within the technical scope disclosed by this application can easily think of changes or substitutions, which should all be covered within the protection scope of this application. Therefore, the protection scope of this application should be subject to the protection scope of the claims.

Claims

1. A clustering analysis method for 5G base stations based on geographical location distribution and spatio-temporal distribution characteristics of energy consumption, characterized in that, It includes the following steps: S1: Obtain 5G base station information; S2: Conduct clustering according to the principle of proximity in geographical location; Among them, S21: Select and initialize several centroids; S22: Allocate each 5G base station according to the principle of the closest Euclidean distance; S23: Update the centroids according to the principle that the updated centroids satisfy the minimum sum of the squares of the distances to all other 5G base stations; S24: Obtain the initial clustering clusters; S3: According to the obtained initial clustering clusters, conduct clustering according to the principle of similarity of the energy consumption curves of 5G base stations to obtain the final clustering clusters; S4: Result analysis.

2. A 5G base station clustering analysis method based on geographical location distribution and energy consumption spatio-temporal distribution characteristics according to claim 1, characterized in that The clustering object in step S2 is 5G base stations, and the clustering result is the initial clustering clusters; The clustering object in step S3 is the initial clustering clusters, and the clustering result is the final clustering clusters.

3. A 5G base station clustering analysis method based on geographical location distribution and energy consumption spatio-temporal distribution characteristics according to claim 1 or 2, characterized in that In step S2, the method of conducting clustering according to the principle of proximity in geographical location includes the following steps: S21: Select and initialize several centroids; S22: Allocate each 5G base station according to the principle of the closest Euclidean distance; S23: Update the centroids according to the principle that the updated centroids satisfy the minimum sum of the squares of the distances to all other 5G base stations; S24: Obtain the initial clustering clusters.

4. A 5G base station clustering analysis method based on geographical location distribution and energy consumption spatio-temporal distribution characteristics according to claim 3, characterized in that In step S21, the method of selecting centroids includes the following steps: S211: Select several centroids, and conduct clustering respectively with different centroid values; S212: Calculate the corresponding distortion degree values respectively according to the clustering; S213: Plot a graph with the centroid as the abscissa and the corresponding distortion degree value as the ordinate; S214: The abscissa centroid corresponding to the turning point is the corresponding optimal clustering value.

5. A 5G base station clustering analysis method based on geographical location distribution and energy consumption spatio-temporal distribution characteristics according to claim 3, characterized in that In step S21, the method of initializing centroids includes the following steps: S216: Convert the longitude and latitude coordinates of 5G base stations into three-dimensional space coordinates; S217: Randomly select a three-dimensional space coordinate of a 5G base station; S218: Select the second coordinate according to the principle of the farthest distance from the first coordinate; S219: Select the third coordinate according to the principle of the farthest distance from the first coordinate and the second coordinate, and so on.

6. A 5G base station clustering analysis method based on geographical location distribution and energy consumption spatio-temporal distribution characteristics according to claim 5, characterized in that In step S3, the method of conducting clustering according to the principle of similarity of energy consumption curves includes the following steps: S31: Initialize centroids; S32: Allocate each initial clustering cluster according to the principle of the minimum difference in energy consumption curves; S33: Update the centroids according to the principle that the updated centroids satisfy the minimum sum of the squares of the differences in energy consumption curves of the centroids of all other initial clustering clusters; S34: Obtain the final clustering clusters.

7. A 5G base station clustering analysis method based on geographical location distribution and energy consumption spatio-temporal distribution characteristics according to claim 6, characterized in that The centroid of the final clustering cluster is marked as the virtual load of the 5G base station, which is used to analyze the electricity consumption characteristics and demand response potential.

8. A 5G base station clustering analysis method based on geographical location distribution and energy consumption spatio-temporal distribution characteristics according to claim 4, characterized in that In step S213, the calculation formula of the distortion degree value is: In the formula, is the symbol for calculating the Euclidean geometric distance; x i is the geographical coordinate of the i-th 5G base station in the k-th clustering cluster; θ k is the geographical location coordinate of the centroid in the k-th clustering cluster; n k is the number of 5G base stations included in the k-th clustering cluster, and m1 is the number of centroids.

9. A 5G base station clustering analysis method based on geographical location distribution and energy consumption spatio-temporal distribution characteristics according to claim 3, characterized in that In step S23, the calculation formula for the minimum sum of squared distances is: In the formula, is the symbol for calculating the Euclidean geometric distance, x i is the geographical coordinate of the i-th 5G base station in the k-th clustering cluster, θ k is the geographical location coordinate of the centroid in the k-th clustering cluster, x k is the class represented by the clustering center, and i is the current class.

10. A 5G base station clustering analysis method based on geographical location distribution and energy consumption spatio-temporal distribution characteristics according to claim 6, characterized in that In step S33, the calculation formula for the minimum sum of squared differences of the energy consumption curves is: Among them, and are the power consumptions of the 5G base station θ i and μ k at time t respectively; t max is the total sampling time of the daily energy consumption, m k is the class represented by the clustering center, and i is the label of the 5G base station.

Images