A 5g base station virtual power plant feasible region aggregation method based on coupling degree clustering

Abstract

The application provides a 5G base station virtual power plant feasible region aggregation method based on coupling degree clustering, belongs to the flexible resource aggregation field, and comprises the following steps: a super-dense heterogeneous 5G base station network model containing macro base stations and micro base stations is constructed, and base station positions, user distribution, channel parameters and service quality constraints are obtained; the base station coupling degree concept is proposed; the coupling degree is used as the similarity measurement, the greedy clustering idea is adopted, the k-means++ optimization initial value is used as the initial clustering center, the remaining base stations are sequentially grouped into the cluster with the maximum coupling degree, the clustering is iterated, and the dynamic grouping is realized according to the flow prediction change; for each base station cluster, the base station communication energy consumption and energy storage resources that can be reduced are combined, the multi-objective optimization is used to solve the adjustable power internal distribution feasibility, and the cluster-level feasible region is obtained. The application significantly improves the accuracy and efficiency of the 5G base station virtual power plant feasible region aggregation, and provides key technical support for the large-scale participation of 5G base stations in the power system auxiliary service market.

Description

Technical Field

[0001] This invention belongs to the field of flexible resource aggregation, and specifically relates to a feasible domain aggregation method for 5G base station virtual power plants based on coupling degree clustering. Background Technology

[0002] 5G base stations, serving as carriers of massive communication loads and distributed energy storage, can be aggregated into virtual power plants to participate in electricity market transactions and support peak shaving, valley filling, and frequency and voltage regulation of the power grid. As 5G networks evolve towards Ultra-Dense Heterogeneous Networks (UDHNs), the network not only includes macro base stations with wide coverage and densely deployed, low-power micro base stations, but also integrates distributed energy storage units, forming a complex spatiotemporally coupled network of base stations and energy storage. In this context, accurately aggregating the feasible domain of 5G base station clusters is crucial for their participation in grid-friendly interactions. Two core coupling characteristics exist in the actual aggregation process: 1) Power regulation coupling: adjacent base stations, due to their shared user coverage, have interconnected communication energy consumption reduction and energy storage charging / discharging regulation behaviors; a single base station's regulation can trigger power linkage within the cluster; 2) Complementary feasible domain coupling: the high-power regulation potential of macro base stations and the flexible response characteristics of micro base stations complement each other; the feasible domain of base stations within the cluster can achieve an aggregation effect of "1+1>2" through collaborative optimization. Discovering and utilizing the coupling and correlation characteristics between base stations is the core path to achieve efficient and accurate aggregation of the feasible domain of 5G base station virtual power plants.

[0003] Current feasible domain aggregation methods for 5G base station virtual power plants are mainly divided into three categories: single-unit aggregation, homogeneous aggregation, and whole-network aggregation. Single-unit aggregation only models a single base station, completely ignoring the coupling and synergistic effects between base stations. The aggregation results are fragmented and cannot reflect the overall adjustment capability of the cluster. Homogeneous aggregation (such as aggregating only macro base stations or micro base stations) simplifies the model complexity, but it severs the complementary adjustment characteristics between macro and micro base stations, seriously underestimating the overall feasible domain size of ultra-dense heterogeneous networks. While whole-network aggregation can cover the characteristics of the entire network, it suffers from drawbacks such as complex model construction, extremely low solution efficiency, and insufficient accuracy in representing feasible domains when facing the problem of a large number of base stations and an explosion of adjustment variable dimensions in ultra-dense heterogeneous networks. In contrast, feasible domain aggregation methods based on coupling degree clustering divide clusters based on the coupling correlation between base stations. This not only groups strongly coupled base stations into the same cluster to preserve synergistic adjustment characteristics, but also simplifies the solution process through cluster dimensionality reduction, while taking into account the complementary characteristics of macro and micro base stations. This provides a new technical path for efficient and accurate aggregation of feasible domains in 5G base station virtual power plants. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing feasible region aggregation methods for 5G base station virtual power plants by proposing a method based on coupling degree clustering. This invention uses the coupling correlation characteristics between base stations as the core criterion for cluster partitioning. It employs a greedy clustering algorithm combined with k-means++ to optimize initial values ​​for base station clustering, and then uses multi-objective optimization and weighted Minkowski summation to aggregate the feasible regions of each cluster, ultimately obtaining the overall feasible region of the 5G base station virtual power plant. This method can efficiently and accurately aggregate the adjustable power feasible region of 5G base stations in ultra-dense heterogeneous networks, which is of great significance for improving power system flexibility, promoting the coordinated optimization of communication infrastructure and power grid, and supporting its participation in energy market transactions.

[0005] This invention proposes a feasible domain aggregation method for 5G base station virtual power plants based on coupling degree clustering. The method includes the following steps: 1) For a 5G ultra-dense heterogeneous network containing M macro base stations and N micro base stations, construct a network model and extract base station locations, user distribution, channel parameters, and quality of service constraints to obtain the core parameter set of the base station cluster. ; 2) The concept of base station coupling degree is proposed, and the strength of the coupling relationship is measured by the amount of communication traffic of users jointly covered by two base stations. A coupling degree calculation model is constructed and the base station coupling degree matrix is ​​obtained. The specific steps are as follows: 2-1) First, represent the base stations as nodes in the graph, and the set is... Where M represents the number of macro base stations and N represents the number of micro base stations. This represents the i-th base station node. Macro base stations and micro base stations are uniformly included in the node set, and their types are not distinguished; they only participate in coupling degree calculation and cluster analysis based on their node identity.

[0006] 2-2) Base station coupling Defined as the sum of communication traffic for users jointly covered by two base stations, the coupling matrix medium elements The expression is: (1) In the formula, For base stations and The set of user terminals that are covered by the same network and Base stations and The degree of coupling is the communication traffic provided to user k in common coverage. The larger the value, the higher the user overlap between base stations, the closer the communication traffic correlation, and the stronger the coupling relationship; conversely, the smaller the value, the weaker the coupling relationship. 3) Using coupling degree as a similarity metric, a greedy clustering method combined with k-means++ to optimize initial values ​​is employed to cluster base stations, and dynamic grouping is performed based on traffic prediction to obtain the optimal base station cluster partitioning; the specific steps are as follows: 3-1) The k-means++ algorithm is used to extract data from the base station set. Select K initial cluster centers, denoted as K. The specific selection rules are as follows: a) From the base station set The first initial cluster center is randomly selected. ; b) Calculate the remaining base stations The minimum coupling degree with the selected cluster centers; the smaller the coupling degree, the lower the similarity between the base station and the selected cluster centers. c) Based on probability Randomly select the next initial cluster center, until K initial cluster centers are selected; 3-2) Traverse the remaining unclustered base stations, calculate the average coupling degree between each base station and each existing cluster, and assign it to the cluster with the highest average coupling degree, until all base stations have been clustered, thus obtaining the initial base station cluster partition. ; 3-3) Obtain changes in base station communication traffic over future periods. When the rate of change in intra-cluster coupling exceeds the threshold ε, re-execute the clustering process to update and obtain the optimal base station cluster partitioning. ; 4) For each sub-cluster in the optimal cluster division group, and in combination with the base station's ability to reduce communication energy consumption and energy storage resources, an optimization model for base station cluster communication energy consumption and energy storage regulation is constructed with the goal of maximizing the total cluster-level regulation power.

[0007] The objective function is: (2) In the formula, The communication power consumption that can be reduced for base station b. Adjust the charging and discharging power of the energy storage device at base station b; 5) The decision variables of the model include the signal transmission power between each resource block of the base station and each user. The connection relationship between each resource block of the base station and each user The connection relationship between each base station and each user and the sleep status of each base station. .Right now: (3) In the formula, and These represent the signal transmission power (in W) between the p-th RB of macro base station m or the q-th RB of micro base station n and user u. and These represent the connection relationship between the p-th RB of macro base station m or the q-th RB of micro base station n and user u, respectively. A value of 1 indicates a connection, and a value of 0 indicates no connection. and This represents the connection relationship between macro base station m or micro base station n and user u, where a value of 1 indicates a connection and a value of 0 indicates no connection. and These represent the sleep state of macro base station m or micro base station n, with a value of 1 indicating sleep and a value of 0 indicating activity.

[0008] 6) Construct model constraints and solve for the feasibility of adjusting the internal power allocation through multi-objective optimization. Obtain the cluster-level feasible region. The model constraints include: user service quality constraints, base station transmit power constraints, energy storage device state of charge constraints, and energy consumption reduction threshold constraints. 6-1) User Service Quality Constraints. User service quality constraints refer to ensuring that the communication quality (such as minimum transmission rate) meets preset requirements by dynamically adjusting the transmit power of each user or channel, while optimizing resource utilization and avoiding excessive interference to other users.

[0009] When user u connects to the p-th RB of macro base station m or the q-th RB of micro base station n, the signal-to-interference-plus-noise ratio (SIR) of the communication is: (4) (5) In the formula, Noise power (in W); and These represent the channel gain between the p-th RB of macro base station m or the q-th RB of micro base station n and user u, respectively, and are related to distance-based path impairment. This is relevant when the distance between the base station and the user is less than the reference distance. At this time, the channel gain is a fixed path loss value, where the fixed path losses of macro base stations and micro base stations are respectively... and If the distance between the base station and the user is greater than Channel gain is based on path loss exponent. ( Attenuation occurs with distance between the base station and the user: (6) (7) In the formula, and These represent the distances between the macro base station m or the micro base station n and the user u, respectively.

[0010] According to Shannon's theorem, the rate at which user u acquires data from the p-th RB of macro base station m can be expressed as: (8) (9) In the formula, and , respectively, represent the data transmission rate (in Mbps) between the p-th RB of the macro base station m or the q-th RB of the micro base station and user u; W is the bandwidth of each RB. Since the RB is the smallest radio resource allocation unit in the base station, the bandwidth of each RB is a fixed value W (in MHz).

[0011] To ensure the quality of communication services, the data rate obtained by users across all base stations must meet the user service quality constraints, namely: (10) In the formula This indicates the user's minimum data transfer rate requirement.

[0012] 6-2) User-RB allocation constraints. Connection variables between users and base stations (RBs). and This is a binary integer variable; a value of 1 indicates concatenation, and a value of 0 indicates no concatenation. (11) (12) Furthermore, an RB can only be assigned to one user at a time. (13) (14) 6-3) User-Base Station Allocation Constraints. Connection variables between users and base stations. and This is a binary integer variable; a value of 1 indicates concatenation, and a value of 0 indicates no concatenation. (15) (16) Each user terminal can only access one base station at a time. (17) A user terminal can only have a connection with an RB (Radio Station) of a base station if and only if the user terminal has accessed that base station: (18) (19) In the formula, P and Q are the number of RBs for macro base stations and micro base stations, respectively.

[0013] 6-4) Base station transmit power constraints. The total transmit power of each base station cannot exceed the maximum transmit power, where the maximum transmit power for macro base stations and micro base stations are respectively... and : (20) (twenty one) 6-5) RB power constraint. When or When =1, that is, when there is a connection between the base station RB and the user, Only then can it take a value greater than 0, that is: (twenty two) (twenty three) 6-6) Base Station Sleep Constraints. When no users are accessing any of the base station's RBs, the base station can choose to sleep: (twenty four) (25) 6-7) Connection distance constraint. If the distance between the user and the base station is greater than the connectable distance, a connection cannot be established. like (26) like (27) 6-8) Total Throughput Constraint. Due to base station service limitations, the total throughput of a single base station cannot exceed the maximum throughput limit. The maximum throughput for macro base stations and micro base stations are respectively... and (Unit: Mbps): (28) (29) 7) The NSGA-Ⅲ algorithm is used to solve the optimization model to obtain the feasible region of adjustable power for each base station cluster. ; 8) Using the weighted Minkowski summation method, with the size and adjustment potential of each cluster base station as weights, the feasible regions of all clusters are summed to obtain the overall feasible region of the 5G base station virtual power plant. The specific steps are as follows: 8-1) Construct a cluster weight calculation model, weight The expression is: (30) In the formula, For weighting coefficients, , These represent the number of base stations in the c-th cluster and the total number of base stations, respectively. , Let be the regulatory potential of the c-th cluster and the total regulatory potential, respectively. This ensures the physical meaning and numerical rationality of the feasible region after aggregation; 8-2) Based on the weighted Minkowski summation operator By combining weighting coefficients and performing aggregation operations on the feasible regions of each cluster, the overall feasible region of the 5G base station virtual power plant is obtained. The expression is (31) Specifically, the feasible region of each cluster is first determined. Multiply by weight Then, Minkowski summation is performed on all weighted feasible regions to obtain the overall regulated power feasible region of the virtual power plant, ensuring that the aggregation result satisfies the constraints of each cluster and is physically feasible.

[0014] Compared with the prior art, the beneficial effects of the present invention are as follows: The present invention makes full use of the dynamic clustering advantages and multi-resource collaborative optimization characteristics of coupling degree clustering, which significantly improves the accuracy and efficiency of feasible domain aggregation of 5G base station virtual power plant, and provides key technical support for the large-scale participation of 5G base stations in the power system ancillary services market. Attached Figure Description

[0015] Figure 1 This is a base station distribution diagram in an embodiment of the present invention; Figure 2 As described in the embodiments of the present invention Figure 1 Spatial distribution diagram of user points and base stations within a 1000m×1000m area in the lower left corner; Figure 3 This is a schematic diagram of the base station clustering results obtained in this embodiment of the invention, where a clustering size of 30 is used. Figure 4 This is a schematic diagram illustrating the aggregation speed across multiple scenarios in an embodiment of the present invention; Figure 5 This is a schematic diagram of the feasible region accuracy in an embodiment of the present invention. Detailed Implementation

[0016] The technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] A feasible domain aggregation method for 5G base station virtual power plants based on coupling degree clustering, the method includes the following steps: 1) For a 5G ultra-dense heterogeneous network containing M macro base stations and N micro base stations, construct a network model and extract base station locations, user distribution, channel parameters, and quality of service constraints to obtain the core parameter set of the base station cluster. ; 2) The concept of base station coupling degree is proposed, and the strength of the coupling relationship is measured by the amount of communication traffic of users jointly covered by two base stations. A coupling degree calculation model is constructed and the base station coupling degree matrix is ​​obtained. The specific steps are as follows: 2-1) First, represent the base stations as nodes in the graph, and the set is... Where M represents the number of macro base stations and N represents the number of micro base stations. This represents the i-th base station node. Macro base stations and micro base stations are uniformly included in the node set, and their types are not distinguished; they only participate in coupling degree calculation and cluster analysis based on their node identity.

[0018] 2-2) Base station coupling Defined as the sum of communication traffic for users jointly covered by two base stations, the coupling matrix medium elements The expression is: (1) In the formula, For base stations and The set of user terminals that are covered by the same network and Base stations and The degree of coupling is the communication traffic provided to user k in common coverage. The larger the value, the higher the user overlap between base stations, the closer the communication traffic correlation, and the stronger the coupling relationship; conversely, the smaller the value, the weaker the coupling relationship. 3) Using coupling degree as a similarity metric, a greedy clustering method combined with k-means++ to optimize initial values ​​is employed to cluster base stations, and dynamic grouping is performed based on traffic prediction to obtain the optimal base station cluster partitioning; the specific steps are as follows: 3-1) The k-means++ algorithm is used to extract data from the base station set. Select K initial cluster centers, denoted as K. The specific selection rules are as follows: a) From the base station set The first initial cluster center is randomly selected. ; b) Calculate the remaining base stations The minimum coupling degree with the selected cluster centers; the smaller the coupling degree, the lower the similarity between the base station and the selected cluster centers. c) Based on probability Randomly select the next initial cluster center, until K initial cluster centers are selected; 3-2) Traverse the remaining unclustered base stations, calculate the average coupling degree between each base station and each existing cluster, and assign it to the cluster with the highest average coupling degree, until all base stations have been clustered, thus obtaining the initial base station cluster partition. ; 3-3) Obtain changes in base station communication traffic over future periods. When the rate of change in intra-cluster coupling exceeds the threshold ε, re-execute the clustering process to update and obtain the optimal base station cluster partitioning. ; 4) For each sub-cluster in the optimal cluster division group, and in combination with the base station's ability to reduce communication energy consumption and energy storage resources, an optimization model for base station cluster communication energy consumption and energy storage regulation is constructed with the goal of maximizing the total cluster-level regulation power.

[0019] The objective function is: (2) In the formula, The communication power consumption that can be reduced for base station b. Adjust the charging and discharging power of the energy storage device at base station b; 5) The decision variables of the model include: the signal transmission power between each resource block of the base station and each user. The connection relationship between each resource block of the base station and each user The connection relationship between each base station and each user and the sleep status of each base station. .Right now: (3) In the formula, and These represent the signal transmission power (in W) between the p-th RB of macro base station m or the q-th RB of micro base station n and user u. and These represent the connection relationship between the p-th RB of macro base station m or the q-th RB of micro base station n and user u, respectively. A value of 1 indicates a connection, and a value of 0 indicates no connection. and This represents the connection relationship between macro base station m or micro base station n and user u, where a value of 1 indicates a connection and a value of 0 indicates no connection. and These represent the sleep state of macro base station m or micro base station n, with a value of 1 indicating sleep and a value of 0 indicating activity.

[0020] 6) Construct model constraints and solve for the feasibility of adjusting the internal power allocation through multi-objective optimization. Obtain the cluster-level feasible region. The model constraints include: user service quality constraints, base station transmit power constraints, energy storage device state of charge constraints, and energy consumption reduction threshold constraints. 6-1) User Service Quality Constraints. User service quality constraints refer to ensuring that the communication quality (such as minimum transmission rate) meets preset requirements by dynamically adjusting the transmit power of each user or channel, while optimizing resource utilization and avoiding excessive interference to other users.

[0021] When user u connects to the p-th RB of macro base station m or the q-th RB of micro base station n, the signal-to-interference-plus-noise ratio (SIR) of the communication is: (4) (5) In the formula, Noise power (in W); and These represent the channel gain between the p-th RB of macro base station m or the q-th RB of micro base station n and user u, respectively, and are related to distance-based path impairment. This is relevant when the distance between the base station and the user is less than the reference distance. At this time, the channel gain is a fixed path loss value, where the fixed path losses of macro base stations and micro base stations are respectively... and If the distance between the base station and the user is greater than Channel gain is based on path loss exponent. ( Attenuation occurs with distance between the base station and the user: (6) (7) In the formula, and These represent the distances between the macro base station m or the micro base station n and the user u, respectively.

[0022] According to Shannon's theorem, the rate at which user u acquires data from the p-th RB of macro base station m can be expressed as: (8) (9) In the formula, and , respectively, represent the data transmission rate (in Mbps) between the p-th RB of the macro base station m or the q-th RB of the micro base station and user u; W is the bandwidth of each RB. Since the RB is the smallest radio resource allocation unit in the base station, the bandwidth of each RB is a fixed value W (in MHz).

[0023] To ensure the quality of communication services, the data rate obtained by users across all base stations must meet the user service quality constraints, namely: (10) In the formula This indicates the user's minimum data transfer rate requirement.

[0024] 6-2) User-RB allocation constraints. Connection variables between users and base stations (RBs). and This is a binary integer variable; a value of 1 indicates concatenation, and a value of 0 indicates no concatenation. (11) (12) Furthermore, an RB can only be assigned to one user at a time. (13) (14) 6-3) User-Base Station Allocation Constraints. Connection variables between users and base stations. and This is a binary integer variable; a value of 1 indicates concatenation, and a value of 0 indicates no concatenation. (15) (16) Each user terminal can only access one base station at a time. (17) A user terminal can only have a connection with an RB (Radio Station) of a base station if and only if the user terminal has accessed that base station: (18) (19) In the formula, P and Q are the number of RBs for macro base stations and micro base stations, respectively.

[0025] 6-4) Base station transmit power constraints. The total transmit power of each base station cannot exceed the maximum transmit power, where the maximum transmit power for macro base stations and micro base stations are respectively... and : (20) (twenty one) 6-5) RB power constraint. When or When =1, that is, when there is a connection between the base station RB and the user, Only then can it take a value greater than 0, that is: (twenty two) (twenty three) 6-6) Base Station Sleep Constraints. When no users are accessing any of the base station's RBs, the base station can choose to sleep: (twenty four) (25) 6-7) Connection distance constraint. If the distance between the user and the base station is greater than the connectable distance, a connection cannot be established. like (26) like (27) 6-8) Total Throughput Constraint. Due to base station service limitations, the total throughput of a single base station cannot exceed the maximum throughput limit. The maximum throughput for macro base stations and micro base stations are respectively... and (Unit: Mbps): (28) (29) 7) The NSGA-Ⅲ algorithm is used to solve the optimization model to obtain the feasible region of adjustable power for each base station cluster. ; 8) Using the weighted Minkowski summation method, with the size and adjustment potential of each cluster base station as weights, the feasible regions of all clusters are summed to obtain the overall feasible region of the 5G base station virtual power plant. The specific steps are as follows: 8-1) Construct a cluster weight calculation model, weight The expression is: (30) In the formula, For weighting coefficients, , These represent the number of base stations in the c-th cluster and the total number of base stations, respectively. , Let be the regulatory potential of the c-th cluster and the total regulatory potential, respectively. This ensures the physical meaning and numerical rationality of the feasible region after aggregation; 8-2) Based on the weighted Minkowski summation operator By combining weighting coefficients and performing aggregation operations on the feasible regions of each cluster, the overall feasible region of the 5G base station virtual power plant is obtained. The expression is (31) Specifically, the feasible region of each cluster is first determined. Multiply by weight Then, Minkowski summation is performed on all weighted feasible regions to obtain the overall regulated power feasible region of the virtual power plant, ensuring that the aggregation result satisfies the constraints of each cluster and is physically feasible.

[0026] As a preferred embodiment of the present invention, this section will evaluate the adjustable DR potential of a 5G base station network in a certain city in China. The case study analysis was performed on a PC with an Intel(R) Core(TM) i5-13400 CPU with a main frequency of 2.50GHz. The program was developed based on Python 3.9.5, and the optimized solver was Gurobi 9.5.2.

[0027] Taking a 5G base station in a certain area of ​​China as an example, the base station distribution is as follows: Figure 1 As shown in the figure, the remaining parameters are shown in Table 1.

[0028] In statistics, because the spatial distribution of users is random, we can integrate all users within a small area into a single user aggregation point and connect it to an equivalent aggregation RB. In this case, a total of 3000 user points are generated within the area, each representing a user cluster. The user points are generated randomly. Figure 2 Detailed demonstration Figure 1 Spatial distribution of user points and base stations within the 1000m×1000m area in the lower left corner.

[0029] Scene settings To analyze and compare the accuracy and computation speed of the feasible region of 5G base stations under different clustering conditions, this paper sets up the following three scenarios: Scenario 1 (Original Feasible Domain): No clustering is performed on all base stations. The overall feasible domain of all base stations in the network is solved directly as a benchmark reference for the aggregation effect.

[0030] Scenario 2 (k-means clustering): The traditional k-means clustering algorithm is used to divide the clusters according to the spatial location characteristics of the base stations. Then, the feasible regions at the cluster level are solved separately and the overall aggregation is completed. This is used to compare the advantages and disadvantages of clustering algorithms.

[0031] Scenario 3 (Coupling Degree Clustering): The method proposed in this invention, which uses coupling degree as a similarity measure and combines greedy clustering with k-means++ to optimize the initial value, is used to divide the base station clusters. Subsequently, the cluster-level and overall feasible regions are solved to verify the effectiveness of the method.

[0032] The clustering effect analysis is as follows: To reduce computational complexity and accelerate parallel computing, this paper employs a coupling-degree-based clustering method for base station networks. This study uses a cluster size of 30 and obtains the base station clustering results, such as... Figure 3 As shown.

[0033] The dashed lines in the figure fit the geographical distribution range of each base station cluster through a convex hull. Research shows that the coupling-based clustering method can group base stations with high signal interference into one class. Since the signal interference intensity of macro base stations is significantly higher than that of micro base stations, the convex hull contour formed by this method is mainly composed of macro base stations. Furthermore, the base station size in each cluster remains relatively balanced, which provides a good foundation for sub-problem decomposition and parallel computing acceleration.

[0034] The polymerization rate analysis is as follows: like Figure 4 As shown, Scenario 1 is the original feasible region solution, which does not require clustering modeling, so the modeling time is the shortest. However, it directly solves the overall feasible region of all 1029 base stations in the network (196 macro base stations + 833 micro base stations), which faces the problems of variable dimension explosion and extremely high model complexity, resulting in a solution time that is much higher than the two scenarios after clustering.

[0035] Scenario 2 uses the k-means clustering method, which reduces the dimensionality of the base station cluster and significantly reduces the solution dimensionality. The solution time is reduced by an order of magnitude compared to Scenario 1. The modeling time is consumed by the clustering operation. However, it only uses distance as the clustering basis and does not consider the coupling and correlation characteristics between base stations based on communication traffic, thus severing the power regulation coupling and the complementary coupling relationship of the feasible region of the base station.

[0036] Scenario 3 uses coupling degree clustering, which takes slightly longer to model than Scenario 2. This is because the method requires first calculating the base station coupling degree matrix, then combining k-means++ to optimize the initial values ​​with greedy clustering to complete dynamic grouping. The clustering process is more in line with the actual coupling characteristics of base station operation. However, by grouping strongly coupled base stations into the same cluster, it preserves the cooperative adjustment characteristics of base stations within the cluster and the complementary characteristics of macro and micro base stations. The polymerization accuracy analysis is as follows: The experimental results from the three scenarios clearly demonstrate the differentiated impact of different clustering methods on the aggregation accuracy of the feasible domain of the 5G base station virtual power plant. Figure 5 As shown.

[0037] From the perspective of feasible domain aggregation accuracy, the three scenarios show clear gradient differences: Scenario 1, as the original feasible domain solution benchmark, has not undergone clustering dimensionality reduction processing and fully retains all the features of the feasible domain of the 5G base station virtual power plant, so the aggregation accuracy reaches 100%.

[0038] Scenario 2 uses the k-means clustering method, which only uses spatial distance as the basis for division and ignores the power regulation coupling between base stations and the complementary characteristics of feasible regions. This results in feature loss in the aggregation of feasible regions after clustering, and the accuracy drops to 95.4%.

[0039] Scenario 3: Based on coupling degree clustering, the clustering process fully considers the coupling relationship of communication traffic between base stations and the complementary characteristics of macro and micro base stations. Only a small amount of accuracy loss occurs in the clustering calculation stage. The final aggregation accuracy reaches 97.8%, which is 2.4 percentage points higher than k-means clustering. It not only achieves effective dimensionality reduction of the solution dimension, but also retains the true characteristics of the feasible region to the greatest extent, achieving a better balance between accuracy loss and solution efficiency.

[0040] As can be seen from the above embodiments: 1) The present invention provides a feasible domain aggregation method for 5G base station virtual power plants based on coupling degree clustering. It achieves accurate clustering of ultra-dense heterogeneous 5G base stations by relying on the coupling degree of base stations, and completes feasible domain aggregation by combining multi-objective optimization and weighted Minkowski summation. It can achieve efficient and accurate characterization of the feasible domain of 5G base station virtual power plants while preserving the collaborative adjustment characteristics of base station clusters. The present invention is compatible with the macro-micro base station hybrid deployment characteristics of 5G ultra-dense heterogeneous networks, meets the technical requirements of the power system ancillary service market for flexible resource aggregation, and facilitates base station operators to build 5G base station virtual power plants on a large scale and participate in power market transactions.

[0041] 2) The 5G base station virtual power plant feasible domain aggregation method based on coupling degree clustering of the present invention, compared with traditional feasible domain aggregation methods such as single-unit aggregation and whole-network aggregation, achieves dimensionality reduction solution of base station clusters for base station operators, greatly reduces the model construction and solution cost, and significantly improves the efficiency and accuracy of 5G base station virtual power plant feasible domain aggregation, which is of positive significance for solving the technical bottleneck problem of large-scale participation of 5G base stations in the power market; For power grid companies, the feasible domain obtained by this method can truly reflect the actual adjustment potential of 5G base station clusters, providing accurate flexible resource data support for power grid dispatch and effectively improving the power grid's utilization efficiency of distributed flexible resources.

[0042] In an embodiment of the feasible domain aggregation method for 5G base station virtual power plants based on coupling degree clustering of the present invention, the feasible domain aggregation accuracy based on coupling degree clustering reaches 97.8%, which is far superior to the traditional k-means clustering method. In the long run, this method can realize the dynamic updating and accurate aggregation of the feasible domain of 5G base station virtual power plants, which helps to tap the power regulation potential of communication infrastructure and promote the coordinated optimization and development of communication networks and power grids.

[0043] 3) When implementing this method, only the core operational parameters of 5G base stations, such as location and traffic, need to be extracted. Base station clustering can be completed through coupling degree matrix construction and clustering algorithms. Then, the overall feasible region can be obtained by combining cluster-level multi-objective optimization and weighted aggregation. Each base station cluster retains its own coupling and cooperative characteristics in the aggregation process, and the feasible region aggregation result conforms to the physical operation constraints of the power system. These characteristics make this method adaptable to 5G ultra-dense heterogeneous networks of different scales, enabling normalized and engineering-based implementation.

[0044] However, those skilled in the art should recognize that the above embodiments are only used to illustrate this application and are not intended to limit this application. Any changes or modifications to the above embodiments within the spirit and essence of this application will fall within the scope of the claims of this application.

Claims

1. A feasible domain aggregation method for 5G base station virtual power plants based on coupling degree clustering, characterized in that, Includes the following steps: 1) For a 5G ultra-dense heterogeneous network containing M macro base stations and N micro base stations, construct a network model and extract base station locations, user distribution, channel parameters, and quality of service constraints to obtain the core parameter set of the base station cluster. ; 2) The concept of base station coupling degree is proposed, and the strength of the coupling relationship is measured by the amount of communication traffic of users jointly covered by two base stations. A coupling degree calculation model is constructed and the base station coupling degree matrix is ​​obtained. ; 3) Using coupling degree as a similarity measure, greedy clustering combined with k-means++ to optimize initial values ​​is used to cluster base stations, and dynamic grouping is performed based on traffic prediction to obtain the optimal base station cluster division; 4) For each sub-cluster in the optimal cluster division group, and combining the base station's ability to reduce communication energy consumption and energy storage resources, with the goal of maximizing the total cluster-level regulation power, construct an optimization model for base station cluster communication energy consumption and energy storage regulation. 5) The decision variables of the model include the signal transmission power between each resource block of the base station and each user. The connection relationship between each resource block of the base station and each user The connection relationship between each base station and each user and the sleep status of each base station. ,Right now: In the formula, and These represent the signal transmission power between the p-th RB of macro base station m or the q-th RB of micro base station n and user u, respectively. and These represent the connection relationship between the p-th RB of macro base station m or the q-th RB of micro base station n and user u, respectively. A value of 1 indicates a connection, and a value of 0 indicates no connection. and This represents the connection relationship between macro base station m or micro base station n and user u, where a value of 1 indicates a connection and a value of 0 indicates no connection. and These represent the sleep state of macro base station m or micro base station n, with a value of 1 indicating sleep and a value of 0 indicating activity. 6) Construct model constraints and solve the feasibility of adjusting the internal power allocation through multi-objective optimization to obtain the cluster-level feasible region. The model constraints include: user service quality constraints, base station transmit power constraints, energy storage device state of charge constraints, and energy consumption reduction threshold constraints. 7) The NSGA-Ⅲ algorithm is used to solve the optimization model to obtain the feasible region of adjustable power for each base station cluster. ; 8) Using the weighted Minkowski summation method, with the size and adjustment potential of each cluster base station as weights, the feasible regions of all clusters are summed to obtain the overall feasible region of the 5G base station virtual power plant. .

2. The feasible domain aggregation method for 5G base station virtual power plant based on coupling degree clustering according to claim 1, characterized in that, Step 2) is as follows: 2-1) First, represent the base stations as nodes in the graph, and the set is... Where M represents the number of macro base stations and N represents the number of micro base stations. The i-th base station node is represented by the node set. Macro base stations and micro base stations are included in the node set and participate in coupling degree calculation and cluster analysis only as nodes, without distinguishing their types. 2-2) Base station coupling Defined as the sum of communication traffic for users jointly covered by two base stations, the coupling matrix medium elements The expression is: (1) In the formula, For base stations and The set of user terminals that are covered by the same network and Base stations and The degree of coupling is the communication traffic provided to user k in common coverage. The larger the value, the higher the user overlap between base stations, the closer the communication traffic correlation, and the stronger the coupling relationship; conversely, the smaller the value, the weaker the coupling relationship.

3. The feasible domain aggregation method for 5G base station virtual power plants based on coupling degree clustering according to claim 2, characterized in that, Step 3) is as follows: 3-1) The k-means++ algorithm is used to extract data from the base station set. Select K initial cluster centers, denoted as K. The specific selection rules are as follows: a) From the base station set The first initial cluster center is randomly selected. ; b) Calculate the remaining base stations The minimum coupling degree with the selected cluster centers; the smaller the coupling degree, the lower the similarity between the base station and the selected cluster centers. c) Based on probability Randomly select the next initial cluster center, until K initial cluster centers are selected; 3-2) Traverse the remaining unclustered base stations, calculate the average coupling degree between each base station and each existing cluster, and assign it to the cluster with the highest average coupling degree, until all base stations have been clustered, thus obtaining the initial base station cluster partition. ; 3-3) Obtain changes in base station communication traffic over future periods. When the rate of change in intra-cluster coupling exceeds the threshold ε, re-execute the clustering process to update and obtain the optimal base station cluster partitioning. .

4. The feasible domain aggregation method for 5G base station virtual power plant based on coupling degree clustering according to claim 3, characterized in that, The objective function of the base station cluster communication energy consumption and energy storage regulation optimization model in step 4) is: (2) In the formula, The communication power consumption that can be reduced for base station b. Adjust the charging and discharging power of the energy storage device at base station b.

5. The feasible domain aggregation method for 5G base station virtual power plants based on coupling degree clustering according to claim 4, characterized in that, Step 6) is as follows: 6-1) User service quality constraints refer to ensuring that the communication quality of each user or channel meets the preset requirements by dynamically adjusting the transmission power of each user or channel, while optimizing resource utilization and avoiding excessive interference to other users; When user u connects to the p-th RB of macro base station m or the q-th RB of micro base station n, the signal-to-interference-plus-noise ratio (SIR) of the communication is: (3) (4) In the formula, Noise power; and These represent the channel gain between the p-th RB of macro base station m or the q-th RB of micro base station n and user u, respectively, and are related to distance-based path impairment. This is relevant when the distance between the base station and the user is less than the reference distance. At this time, the channel gain is a fixed path loss value, where the fixed path losses of macro base stations and micro base stations are respectively... and If the distance between the base station and the user is greater than Channel gain is based on path loss exponent. Attenuation occurs with distance between the base station and the user: (5) (6) In the formula, and These represent the distances between macro base station m or micro base station n and user u, respectively. According to Shannon's theorem, the rate at which user u acquires data from the p-th RB of macro base station m is expressed as: (7) (8) In the formula, and These represent the data transmission rates between the p-th RB of the macro base station m or the q-th RB of the micro base station and user u, respectively; W is the bandwidth of each RB. Since the RB is the smallest radio resource allocation unit in the base station, the bandwidth of each RB is a fixed value W. The data rate obtained by the user across all base stations meets the user's quality of service constraints, that is: (9) In the formula This indicates the user's minimum data transfer rate requirement. 6-2) User-RB allocation constraints, connection variables between users and base stations (RBs) and This is a binary integer variable; a value of 1 indicates concatenation, and a value of 0 indicates no concatenation. (10) (11) Furthermore, an RB can only be assigned to one user at a time. (12) (13) 6-3) User-Base Station Allocation Constraints, Connection Variables between Users and Base Stations and This is a binary integer variable; a value of 1 indicates concatenation, and a value of 0 indicates no concatenation. (14) (15) Each user terminal can only access one base station at a time. (16) A user terminal can only have a connection with an RB (Radio Station) of a base station if and only if the user terminal has accessed that base station: (17) (18) In the formula, P and Q are the number of RBs for macro base stations and micro base stations, respectively; 6-4) Base station transmit power constraint: The total transmit power of each base station cannot exceed the maximum transmit power, where the maximum transmit power of macro base stations and micro base stations are respectively... and : (19) (20) 6-5) RB power constraint, when or When =1, that is, when there is a connection between the base station RB and the user, Only then can it take a value greater than 0, that is: (21) (22) 6-6) Base station sleep constraint: When no users access any of the base station's RBs, the base station chooses to sleep. (23) (24) 6-7) Connection distance constraint: If the distance between the user and the base station is greater than the connectable distance, a connection cannot be established. like (25) like (26) 6-8) Total throughput constraint: Due to base station service limitations, the total throughput of a single base station cannot exceed the maximum throughput limit. The maximum throughput for macro base stations and micro base stations are respectively... and : (27) (28)。 6. The feasible domain aggregation method for 5G base station virtual power plants based on coupling degree clustering according to claim 4, characterized in that, Step 8) is as follows: 8-1) Construct a cluster weight calculation model, weight The expression is: (29) In the formula, For weighting coefficients, , These represent the number of base stations in the c-th cluster and the total number of base stations, respectively. , Let be the regulatory potential of the c-th cluster and the total regulatory potential, respectively. This ensures the physical meaning and numerical rationality of the feasible region after aggregation; 8-2) Based on the weighted Minkowski summation operator By combining weighting coefficients and performing aggregation operations on the feasible regions of each cluster, the overall feasible region of the 5G base station virtual power plant is obtained. The expression is: (30) Specifically, the feasible region of each cluster is first determined. Multiply by weight Then, Minkowski summation is performed on all weighted feasible regions to obtain the overall regulated power feasible region of the virtual power plant, ensuring that the aggregation result satisfies the constraints of each cluster and is physically feasible.

Images