Classified aggregation equivalent modeling method for photovoltaic station

Abstract

The invention belongs to the technical field of new energy power generation grid connection, and particularly discloses a photovoltaic station classification aggregation equivalent modeling method, which comprises the following steps: acquiring control parameters and sensitivity of a photovoltaic power generation unit, and calculating a clustering index according to the control parameters and the sensitivity; dividing the clustering indexes into a plurality of clusters, inputting the plurality of clusters into a pre-trained improved FCM clustering algorithm model, and obtaining corresponding cluster centers under different cluster numbers by adopting a spore propagation search mode; obtaining a plurality of division schemes for clustering indexes; and calculating division coefficients and classification entropies under corresponding cluster numbers in different division schemes, determining a classification aggregation result of the photovoltaic power generation units, and performing equivalent modeling on the photovoltaic power generation units according to the classification aggregation result. According to the method, the global search capability and convergence efficiency of a traditional clustering algorithm are improved, and accurate clustering of the dynamic characteristics of the photovoltaic power generation units is realized.

Description

Technical Field

[0001] The present invention belongs to the technical field of new energy power generation grid connection, and specifically relates to a photovoltaic station classification aggregation equivalent modeling method. Background Art

[0002] With the rapid development of new energy power generation technology, photovoltaic power generation has become an important driving force for the transformation of energy structure. However, the randomness and intermittent nature of photovoltaic power generation makes its output power have significant fluctuations, which poses a serious challenge to the security and stability of the power system. Especially in the context of the gradual expansion of photovoltaic station scale, the interaction between its dynamic response characteristics and the power grid is becoming more and more serious. In order to more efficiently characterize the grid-related dynamic characteristics of photovoltaic stations, modeling and classification aggregation technology have become key research directions.

[0003] Traditional photovoltaic (PV) plant modeling methods often rely on simplified assumptions, making it difficult to fully reflect dynamic operational characteristics. While clustering algorithms such as the fuzzy C-means (FCM) have been widely used for classification and aggregation, they are sensitive to initial cluster centers and tend to converge to local optima. When processing high-dimensional data, they often suffer from slow convergence and insufficient global search capabilities, leading to low modeling efficiency and accuracy. Furthermore, in the current scenario of high renewable energy grid integration, the adaptability of PV plant modeling to system-level simulation analysis is poor. Therefore, seeking an optimized classification and aggregation modeling method is essential for analyzing PV plant dynamic characteristic data. Summary of the Invention

[0004] The purpose of the present invention is to provide a photovoltaic station classification aggregation equivalent modeling method, improve the global search capability and convergence efficiency of traditional clustering algorithms, and improve the accuracy and efficiency of classification and aggregation of dynamic characteristic data of photovoltaic stations.

[0005] To achieve the above objectives, the present invention adopts the following technical solutions: According to one aspect of the present invention, a photovoltaic station classification aggregation equivalent modeling method is provided, comprising the following steps: Obtaining control parameters and sensitivities of photovoltaic power generation units, and calculating clustering indicators based on the control parameters and sensitivities; The clustering indicators are divided into a number of clusters, and the clustering indicators corresponding to the clusters are input into a pre-trained improved fuzzy C-means (FCM) clustering algorithm model to obtain a division scheme of the clustering indicators under different numbers of clusters; In the improved FCM clustering algorithm model, the spore propagation search method in the moss growth optimization algorithm is used to obtain the corresponding cluster centers under different cluster numbers; Calculate the division coefficient and classification entropy under the corresponding cluster number in different division schemes; determine the optimal cluster number according to the division coefficient and classification entropy, and determine the classification aggregation result of the photovoltaic power generation unit according to the division scheme of the clustering index corresponding to the optimal cluster number.

[0006] According to one embodiment of the present invention, the control parameters of the photovoltaic power generation unit include an active power setting value, a reactive power setting value, an active power regulation rate, and a reactive power regulation rate; The clustering indicators include the product of the active power setting value, the active power regulation rate and the corresponding sensitivity, and the product of the reactive power setting value, the reactive power regulation rate and the corresponding sensitivity.

[0007] According to one embodiment of the present invention, in the step of obtaining cluster centers corresponding to different numbers of clusters using a spore propagation search, the spore propagation search optimizes the cluster center to minimize the Euclidean distance between the majority of clustering indicators in the cluster. In a specific implementation, the optimization direction is generally to minimize the Euclidean distance between the cluster center and more than 90% of the clustering indicators in the cluster.

[0008] According to one embodiment of the present invention, the step of inputting the clustering indices corresponding to the plurality of clusters into a pre-trained improved FCM clustering algorithm model to obtain a partitioning scheme for clustering indices under different numbers of clusters includes: Determine the maximum number of iterations and the maximum number of clusters allowed K max ; Let the number of iterations be 1 and the initial number of clusters be K , K is a positive integer, and 1≤ K < K max ; Search and update using spore propagation K The cluster center of each cluster; If the cluster center changes, spore propagation is used again to search and update the cluster center; If the cluster center does not change, the output cluster number is K The membership matrix of the cluster center and each photovoltaic power generation unit to each cluster center at the time; and let K = K +1, restart spore propagation search and update cluster center until K = K max ; According to the cluster centers corresponding to different cluster numbers and the membership matrix of each photovoltaic power generation unit to each cluster center, a division scheme of clustering indicators under different cluster numbers is obtained.

[0009] Generally, the initial number of clusters K It is the minimum initial value determined artificially, usually the initial number of clusters K =1.

[0010] The maximum allowed number of clusters is the total number of photovoltaic power generation units.

[0011] According to one embodiment of the present invention, when the initial number of clusters is K ≠1, the clustering index is divided into K When clusters are formed, the clustering indicators are aggregated in a weighted manner, and objects with similar values are classified into the same cluster.

[0012] The spore propagation search and update K In the cluster center step of each cluster, the spore propagation search includes the following steps: Spore individuals are randomly generated according to the number of clustering indicators to form a spore population. The adaptability of the spores is calculated according to the fitness value of the spore individuals. The spores with the best adaptability are selected from the population, and their corresponding cluster centers are obtained. The obtained cluster centers are combined with the results obtained by spore propagation search. It is judged whether the cluster center has changed. If the cluster center has changed, the spore propagation search is re-used and the cluster center is updated until the cluster center no longer changes.

[0013] According to an embodiment of the present invention, the step of calculating the partition coefficient and classification entropy corresponding to the number of clusters in different partitioning schemes includes: Obtain the membership matrix of each photovoltaic power generation unit to each cluster center in the partitioning scheme, and calculate the partition coefficient according to the membership matrix D pc and classification entropy D ce , the calculation formula is as follows: ; in, δ ij For the j The photovoltaic power generation unit belongs to i The membership value of the cluster corresponding to the cluster center point, K is the number of clusters, N is the number of photovoltaic power generation units; δ ij Calculated according to the following formula: ; Where, K is the number of clusters, W is the clustering index, C j For the j The cluster center of the cluster to which the photovoltaic power generation units belong; Ck For the k The cluster centers of the clusters.

[0014] According to an embodiment of the present invention, the step of determining the optimal number of clusters based on the partition coefficient and classification entropy includes: the number of clusters corresponding to the maximum partition coefficient or the minimum classification entropy is the optimal number of clusters.

[0015] According to one embodiment of the present invention, the step of performing equivalent modeling on photovoltaic power generation units based on the classification aggregation results includes: establishing a multi-machine equivalent model, the multi-machine equivalent model includes a number of equivalent power generation units, the number of equivalent power generation units is the optimal cluster number, and the equivalent power generation units correspond one-to-one to the cluster centers under the optimal cluster number.

[0016] According to one aspect of the present invention, a photovoltaic station classification aggregation equivalent modeling device is provided, comprising: A data acquisition module, configured to obtain control parameters and sensitivities of photovoltaic power generation units, and calculate clustering indices based on the control parameters and sensitivities; A clustering partitioning module is used to divide the clustering index into a plurality of clusters, and input the clustering indexes corresponding to the plurality of clusters into a pre-trained improved FCM clustering algorithm model to obtain partitioning schemes of the clustering index under different numbers of clusters, wherein the plurality of partitioning schemes have different numbers of clusters; in the improved FCM clustering algorithm model, a spore propagation search method is used to obtain the cluster centers corresponding to different numbers of clusters; The analysis output module is used to calculate the division coefficient and classification entropy under the corresponding number of clusters in different division schemes; determine the optimal number of clusters based on the division coefficient and classification entropy, determine the classification aggregation results of photovoltaic power generation units based on the division scheme of clustering indicators corresponding to the optimal number of clusters, and perform equivalent modeling on the photovoltaic power generation units based on the classification aggregation results to obtain a multi-machine equivalent model.

[0017] According to another aspect of the present invention, a computer device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, any one of the above-mentioned photovoltaic station classification aggregation equivalent modeling methods is implemented.

[0018] According to another aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, any one of the above photovoltaic station classification aggregation equivalent modeling methods is implemented.

[0019] Compared with the prior art, the present invention has at least the following beneficial effects: 1. The present invention improves the global search capability and convergence efficiency of traditional clustering algorithms by introducing a moss growth optimization algorithm, thereby achieving accurate classification and aggregation of dynamic characteristic data of photovoltaic stations. The present invention utilizes the environmental adaptive mechanism of the moss growth optimization algorithm to be closer to the characteristics of dynamic changes in data distribution in clustering problems, applies spore propagation search to the FCM clustering algorithm model, and utilizes its global search capability to optimize the cluster center. The spore propagation search is used to simulate the process of moss diffusion and resource competition, thereby improving the global search capability and convergence efficiency of traditional clustering algorithms. This can not only meet the requirement for stable membership during the convergence process of the FCM clustering algorithm, but also reduce the sensitivity of the FCM clustering algorithm to the initial cluster center, overcome the local convergence defects of traditional methods, avoid the occurrence of local optimal solutions, significantly improve the accuracy of photovoltaic unit dynamic characteristic clustering, and achieve efficient classification and aggregation of dynamic characteristic data of photovoltaic stations.

[0020] 2. This paper proposes a hierarchical aggregation equivalent modeling system that integrates multidimensional feature clustering with an improved FCM clustering algorithm. This method innovatively combines a moss growth optimization algorithm with global optimization capabilities with traditional fuzzy clustering theory to achieve intelligent identification and precise clustering of the dynamic characteristics of photovoltaic power generation units. Furthermore, an equivalent model reduction strategy based on weighted aggregation of dynamic parameters is employed to construct a multi-machine equivalent model for a regional centralized photovoltaic power generation system that balances simulation accuracy and computational efficiency.

[0021] 3. The implementation of the present invention can significantly improve the accuracy of transient stability analysis of power systems containing photovoltaic power sources, reduce the model order while ensuring simulation accuracy, and effectively solve the problem of limited scale of electromagnetic transient simulation caused by large-scale new energy grid connection. It is of great significance to promote the large-scale access and application of clean energy. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] The accompanying drawings, which constitute part of the present invention, are intended to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are intended to explain the present invention and do not constitute an undue limitation of the present invention. In the accompanying drawings: Figure 1 This is a flow chart of the FCM clustering algorithm based on the moss growth optimization algorithm according to Example 1 of the present invention; Figure 2 This is a schematic structural diagram of a regional photovoltaic power generation grid-connected system according to Example 1 of the present invention; Figure 3 This is a schematic structural diagram of a multi-machine equivalent model of a regional photovoltaic power generation grid-connected system according to Example 1 of the present invention; Figure 4 for Figure 2 Detailed model of the photovoltaic power generation grid-connected system in the shown area and Figure 3 Comparison of active power output of equivalent models of photovoltaic power generation grid-connected systems in the shown areas; Figure 5 for Figure 2 Detailed model of the photovoltaic power generation grid-connected system in the shown area and Figure 3 Comparison of reactive power output of equivalent models of photovoltaic power generation grid-connected systems in the shown area. DETAILED DESCRIPTION

[0023] The present invention will be described in detail below with reference to the accompanying drawings and in combination with embodiments. It should be noted that, in the absence of conflict, the embodiments and features of the embodiments of the present invention can be combined with each other.

[0024] The following detailed description is an exemplary description and is intended to provide further detailed description of the present invention. Unless otherwise indicated, all technical terms used in the present invention have the same meaning as those generally understood by those skilled in the art to which the present invention belongs. The terms used in the present invention are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present invention.

[0025] Example 1 A photovoltaic station classification aggregation equivalent modeling method includes the following steps: Obtaining control parameters and sensitivities of photovoltaic power generation units, and calculating clustering indices based on the control parameters and sensitivities; The clustering indicators are divided into several clusters, and the clustering indicators corresponding to the several clusters are input into the pre-trained improved fuzzy C-means (FCM) clustering algorithm model to obtain the division scheme of clustering indicators under different cluster numbers; In the improved FCM clustering algorithm model, the spore propagation search method in the moss growth optimization algorithm is used to obtain the corresponding cluster centers under different cluster numbers; The partition coefficient and classification entropy under the corresponding number of clusters in different partitioning schemes are calculated; the optimal number of clusters is determined based on the partition coefficient and classification entropy, and the classification aggregation results of photovoltaic power generation units are determined based on the partitioning scheme of the clustering index corresponding to the optimal number of clusters.

[0026] By adopting the above technical solution, a moss generation optimization algorithm is introduced on the basis of the traditional FCM clustering algorithm. The global search capability and convergence efficiency of spore propagation search are used to improve the classification aggregation effect, reduce the sensitivity of the FCM clustering algorithm to the initial cluster center, overcome the local convergence defects of the traditional method, avoid the occurrence of local optimal solutions, and significantly improve the dynamic characteristics clustering accuracy of photovoltaic units.

[0027] Specifically, the classification and aggregation equivalent modeling method of photovoltaic stations is as follows: Figure 1 The detailed steps are as follows: S10. Obtaining control parameters and sensitivities of photovoltaic power generation units, and calculating clustering indicators based on the control parameters and sensitivities.

[0028] The control parameters of the photovoltaic power generation unit include active power setting value, reactive power setting value, active power regulation rate and reactive power regulation rate, etc. Read the control parameters of the photovoltaic power generation unit. In order to avoid the phenomenon of "large number eats small number" between different control parameters, the control parameter vector of the photovoltaic power generation unit is converted into H Normalization is performed to eliminate dimensional differences.

[0029] Calculate the product of power output, power regulation rate and sensitivity to obtain the clustering index of photovoltaic power generation unit W ; That is, the clustering index W It includes the product of the active power setting value, the active power regulation rate and the corresponding sensitivity, and the product of the reactive power setting value, the reactive power regulation rate and the corresponding sensitivity.

[0030] S20. Clustering index W Divided into K clusters, K is a positive integer, and 1≤ K < K max ;in, K max is the maximum number of clusters allowed, generally, K max Equal to the total number of photovoltaic power generation units; K The value of is the minimum initial value of the FCM clustering algorithm model K The division standard is the clustering index W Perform weighted aggregation and group clustering indicators with similar values into the same cluster.

[0031] S30. K The clustering indicators of the clusters are input into the pre-trained improved FCM clustering algorithm model to obtain multiple partitioning schemes for the clustering indicators, and the number of clusters in the multiple partitioning schemes is different.

[0032] In this embodiment, the optimization method of the FCM clustering algorithm model is optimized based on the moss growth algorithm, and the spore propagation search method is used for optimization. The optimization direction is K Most clustering metrics in a cluster have the shortest Euclidean distance to the cluster center. Traditional FCM is prone to falling into local optima and is sensitive to the initial cluster center. The moss growth optimization algorithm simulates the propagation and resource competition mechanisms of moss spores to enhance global search capabilities and avoid local optima. The FCM clustering algorithm model is optimized during the initialization strategy and dynamic adjustment of cluster centers.

[0033] The process of obtaining a partitioning scheme using a pre-trained improved FCM clustering algorithm model includes the following steps: Determine the maximum number of iterations m 0 and the maximum number of clusters allowed K max , K max is the total number of photovoltaic power generation units; Let the number of iterations m=1, the initial number of clusters K =1; Select the cluster center of the cluster and determine the optimization direction of the initial cluster. In this embodiment, the optimization direction is K More than 90% of the clustering indices in a cluster have the shortest Euclidean distance to the cluster center; The clustering index is input into the pre-trained improved FCM clustering algorithm model, and spore propagation is used to search and update the cluster center; If the cluster center changes, spore propagation is used again to search and update the cluster center; If the cluster center does not change or no longer changes, calculate the number of clusters K =1 under the division coefficient D pc and classification entropy D ce .and, K = K +1, re-use spore propagation to search and update the cluster center of each cluster until K = K max .

[0034] In this embodiment, the quality of photovoltaic power generation unit clustering is used to determine the optimal number of clusters. The partition coefficient is obtained based on the membership matrix of each photovoltaic power generation unit to each cluster center in the partition scheme. D pc and classification entropy D ce , partition coefficient D pc and classification entropy D ce Calculated according to the following formula: ; in, δ ij For the j The photovoltaic power generation unit belongs to i The membership value of the cluster corresponding to the cluster center point, K is the number of clusters, N is the number of photovoltaic power generation units; δ ij Calculated according to the following formula: ; Where,K is the number of clusters, W is the clustering index, C j For the j The cluster center of the cluster to which the photovoltaic power generation units belong; C k For the k The cluster centers of the clusters.

[0035] In general, the division coefficient D pc The larger the value of , the closer the membership is to 1 or 0, that is, the clearer the clustering is, the higher the membership of each sample point to a cluster, and the lower the degree of mixing. D pc The larger the value, the better the clustering effect. D ce is the uncertainty of the membership, D ce The smaller the value, the clearer the membership and the better the clustering effect.

[0036] exist K Under clusters, the FCM clustering algorithm model improved by the moss growth optimization algorithm is run on each cluster to obtain the membership matrix δ ij , calculate its D pc and D ce The value of D pc and D ce Perform statistics, draw curves, and observe D pc and D ce Follow K Trends in change, looking for D pc The maximum value and D ce The minimum value of, or the corresponding inflection point on the curve, which corresponds to K This is the optimal number of clusters.

[0037] Number of computing clusters K The partition coefficient under D pc and classification entropy D ce Then, determine the number of clusters K Whether the condition is met, the judgment condition is the number of clusters K < K max Is it true? If so, let the number of iterations be m =1, number of clustersK = K +1 and reselect K The cluster centers of the clusters are determined K The optimization direction of the initial cluster; spore propagation is used to search and update the cluster center. K < K max If not, then compare different clusters K Next D pc 、 D ce value, determine the optimal number of clusters, determine the classification and aggregation results of photovoltaic power generation units according to the division scheme of clustering indicators corresponding to the optimal number of clusters, and Figure 2 The photovoltaic power generation units in the regional photovoltaic power generation grid-connected system shown in the figure are modeled as equal value, and the results are as follows: Figure 3 The multi-machine equivalent model shown.

[0038] The steps for spore propagation search are as follows: S(1). Spore population initialization According to the input n Clustering indicators are randomly generated n spore individuals, each spore represents a set of initial cluster centers: ; Where, Shidi i spores, i =1,2,..., n ; Indicates the i The spores corresponding to K Cluster centers.

[0039] S(2). Fitness evaluation Calculate the fitness value of each spore to measure its quality as the initial cluster center. The fitness value of the spore is calculated according to the following formula: ; Where, N is the number of photovoltaic power generation units; K is the number of clusters, W j For the j Clustering index of photovoltaic power generation units.

[0040] The smaller the fitness value, the better the cluster center division corresponding to the spore, that is, the shorter the distance between the corresponding data point and the center.

[0041] S(3). Spore propagation optimization S(31). Adaptability calculation. Spores with lower adaptability have more opportunities to spread. The adaptability of spores is calculated according to the following formula: ; Where, For the adaptability of spores, n is the number of spore individuals, For the i The fitness value of each spore; is the calculated optimal spore fitness value.

[0042] S(4).Optimal initial center selection Select the spores with the best fitness from the spore population x best , and its corresponding cluster center is: ; Compare the cluster centers obtained above with the cluster centers of the clusters selected before optimization. If the cluster centers do not change, proceed to the subsequent steps of the FCM clustering algorithm: calculate the number of clusters K The partition coefficient under D pc and classification entropy D ce ; If the cluster center changes, spore propagation is used again to search and update the cluster center.

[0043] In the process of re-using spore propagation to search and update the cluster center, based on the fitness calculation in the previous optimization, the top 50% of spores in fitness are retained and the spores with the highest fitness are eliminated. n / 2 remaining individuals, then randomly generated n / 2 sets of candidate cluster centers are completed to regain n The spore individuals form a new spore population, initialize the population, re-evaluate the fitness and recalculate the fitness, and select the spore with the best fitness from the updated spore population. x best and its corresponding cluster center.

[0044] The above method is used to Figure 2 The photovoltaic power generation grid-connected systems shown in the figure are classified and aggregated, and the following Figure 3 The multi-machine equivalent model shown in Figure 2 In the figure, the photovoltaic grid-connected system includes n photovoltaic power generation units, as well as the corresponding converters, step-up transformers, loads and power grids. n represents the converter corresponding to the nth photovoltaic power generation unit, T n Indicates the step-up transformer corresponding to the nth photovoltaic power generation unit, L nrepresents the line internal resistance corresponding to the nth photovoltaic power generation unit. The photovoltaic power generation unit is used to generate power. The converter uses dual-loop control to control the active power of the photovoltaic power generation unit, which is finally connected to the grid through the step-up transformer and the AC bus.

[0045] The optimal number of clusters obtained by adopting the photovoltaic station classification aggregation equal value construction in this embodiment is K; Figure 3 In the PV power generation equivalent model, there are K equivalent power generation units, as well as the corresponding step-up transformers, loads and power grids. The K equivalent power generation units correspond one-to-one to the photovoltaic power generation units corresponding to the cluster centers of the K clusters. In the figure, VSC K represents the converter corresponding to the K-th photovoltaic power generation unit, T K represents the step-up transformer corresponding to the K-th photovoltaic power generation unit, L K represents the line internal resistance corresponding to the Kth photovoltaic power generation unit. The photovoltaic power generation unit is used to generate power. The converter uses dual-loop control to control the active power of the photovoltaic power generation unit, which is finally connected to the grid through the step-up transformer and the AC bus.

[0046] In order to verify the photovoltaic station classification and aggregation equivalent modeling method implemented in this invention, the Simulink simulation platform was used to respectively Figure 2 Detailed model of the photovoltaic power generation grid-connected system in the area shown, Figure 3 The multi-machine equivalent model of the photovoltaic power generation grid-connected system in the area shown is simulated and tested. At the 0th second of normal operation, a 0.1MW load is added. The simulation results of active and reactive power are as follows: Figure 4 and Figure 5 shown.

[0047] observe Figure 4 and Figure 5 It can be seen that when the system load suddenly increases, the active and reactive power trends of the detailed model and the equivalent model of the photovoltaic grid-connected system are consistent, with small errors in the specific data. This shows that the FCM clustering algorithm based on moss growth optimization implemented in this invention can achieve high performance testing accuracy and system stability for photovoltaic stations in complex power grid environments. The classification and aggregation of dynamic characteristic data of photovoltaic stations are highly accurate, promoting the safety and reliability of large-scale photovoltaic power generation connected to the grid.

[0048] The photovoltaic station classification and aggregation equivalent modeling method provided in this embodiment is designed to accurately classify and aggregate the dynamic characteristic data of photovoltaic stations, thereby improving the performance test accuracy and system stability of photovoltaic stations in complex power grid environments, and promoting the safety and reliability of large-scale photovoltaic power generation connected to the power grid. This embodiment organically combines the moss growth optimization algorithm with global optimization capabilities with traditional fuzzy clustering theory to achieve intelligent identification and precise clustering of the dynamic characteristics of photovoltaic power generation units; on this basis, an equivalent model reduction strategy based on weighted aggregation of dynamic parameters is adopted to construct a multi-machine equivalent model of a regional centralized photovoltaic power generation system that takes into account both simulation accuracy and computational efficiency, significantly improving the accuracy of photovoltaic unit dynamic characteristic clustering and achieving efficient classification and aggregation of photovoltaic station dynamic characteristic data.

[0049] Example 2 A photovoltaic station classification aggregation equivalent modeling device is provided, comprising: A data acquisition module, configured to obtain control parameters and sensitivities of photovoltaic power generation units, and calculate clustering indices based on the control parameters and sensitivities; A clustering partitioning module is used to divide the clustering index into a plurality of clusters, and input the clustering indexes corresponding to the plurality of clusters into a pre-trained improved FCM clustering algorithm model to obtain partitioning schemes of the clustering index under different numbers of clusters, wherein the plurality of partitioning schemes have different numbers of clusters; in the improved FCM clustering algorithm model, a spore propagation search method is used to obtain the cluster centers corresponding to different numbers of clusters; The analysis output module is used to calculate the division coefficient and classification entropy under the corresponding number of clusters in different division schemes; determine the optimal number of clusters based on the division coefficient and classification entropy, determine the classification aggregation results of photovoltaic power generation units based on the division scheme of clustering indicators corresponding to the optimal number of clusters, and perform equivalent modeling on the photovoltaic power generation units based on the classification aggregation results to obtain a multi-machine equivalent model.

[0050] Example 3 A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, a photovoltaic station classification and aggregation equivalent modeling method is implemented.

[0051] Example 4 A computer-readable storage medium stores a computer program, which, when executed by a processor, implements a photovoltaic station classification aggregation equivalent modeling method.

[0052] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, ordinary technicians in the field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents. Any modification or equivalent replacement that does not depart from the spirit and scope of the present invention should be covered by the scope of protection of the claims of the present invention.

Claims

1. A photovoltaic station classification aggregation equivalent modeling method, characterized by: The following steps are involved: Obtaining control parameters and sensitivities of photovoltaic power generation units, and calculating clustering indicators based on the control parameters and sensitivities; The clustering indicators are divided into several clusters, and the clustering indicators corresponding to the several clusters are input into the pre-trained improved FCM clustering algorithm model to obtain the division scheme of the clustering indicators under different cluster numbers; In the improved FCM clustering algorithm model, a spore propagation search method is used to obtain the corresponding cluster centers under different cluster numbers; The partition coefficients and classification entropies under the corresponding numbers of clusters in different partitioning schemes are calculated; the optimal number of clusters is determined according to the partition coefficients and classification entropy, the classification aggregation results of the photovoltaic power generation units are determined according to the partitioning scheme of the clustering index corresponding to the optimal number of clusters, and the photovoltaic power generation units are equivalently modeled according to the classification aggregation results to obtain a multi-machine equivalent model.

2. The photovoltaic station classification aggregation equivalent modeling method according to claim 1 is characterized in that: The control parameters of the photovoltaic power generation unit include active power setting value, reactive power setting value, active power regulation rate and reactive power regulation rate; The clustering indexes include the product of the active power setting value, the active power regulation rate and the corresponding sensitivity, and the product of the reactive power setting value, the reactive power regulation rate and the corresponding sensitivity.

3. The photovoltaic station classification aggregation equivalent modeling method according to claim 1 is characterized in that: In the step of obtaining the corresponding cluster centers under different cluster numbers by using the spore propagation search method, the optimization direction of the spore propagation search is that the Euclidean distance between more than 90% of the clustering indicators in the cluster and the cluster center is the shortest.

4. The photovoltaic station classification aggregation equivalent modeling method according to claim 1 is characterized in that: The step of inputting the clustering indices corresponding to the plurality of clusters into a pre-trained improved FCM clustering algorithm model to obtain a partitioning scheme for clustering indices under different numbers of clusters includes: Determine the maximum number of iterations and the maximum number of clusters allowed K max ; Let the number of iterations be 1 and the initial number of clusters be K , K is a positive integer, and 1≤ K < K max ; The K The clustering indicators corresponding to the clusters are input into the pre-trained improved FCM clustering algorithm model, and spore propagation is used to search and update K The cluster center of each cluster; If the cluster center changes, spore propagation is used again to search and update the cluster center; If the cluster center does not change, the output cluster number is K The membership matrix of the cluster center and each photovoltaic power generation unit to each cluster center at the time; and let K = K +1, re-use spore propagation to search and update the cluster center of each cluster until K = K max ; According to the cluster centers corresponding to different cluster numbers and the membership matrix of each photovoltaic power generation unit to each cluster center, a division scheme of clustering indicators under different cluster numbers is obtained.

5. The photovoltaic station classification aggregation equivalent modeling method according to claim 4 is characterized in that: The maximum number of clusters allowed K max is the total number of photovoltaic power generation units; Initial number of clusters K =1.

6. The photovoltaic station classification aggregation equivalent modeling method according to claim 1 is characterized in that: The step of determining the optimal number of clusters according to the partition coefficient and the classification entropy includes: The number of clusters corresponding to the maximum partition coefficient or the minimum classification entropy is the optimal number of clusters.

7. The photovoltaic station classification aggregation equivalent modeling method according to claim 1 is characterized in that: The multi-machine equivalent model includes a plurality of equivalent power generation units, the number of the equivalent power generation units is the optimal cluster number, and the equivalent power generation units correspond one-to-one to cluster centers under the optimal cluster number.

8. A photovoltaic station classification aggregation equivalent modeling device, characterized by: include: A data acquisition module, configured to obtain control parameters and sensitivities of photovoltaic power generation units, and calculate clustering indices based on the control parameters and sensitivities; A clustering partitioning module is used to divide the clustering index into a plurality of clusters, and input the clustering indexes corresponding to the plurality of clusters into a pre-trained improved FCM clustering algorithm model to obtain partitioning schemes of the clustering index under different numbers of clusters, wherein the plurality of partitioning schemes have different numbers of clusters; in the improved FCM clustering algorithm model, a spore propagation search method is used to obtain the cluster centers corresponding to different numbers of clusters; The analysis output module is used to calculate the division coefficient and classification entropy under the corresponding number of clusters in different division schemes; determine the optimal number of clusters based on the division coefficient and classification entropy, determine the classification aggregation results of photovoltaic power generation units based on the division scheme of clustering indicators corresponding to the optimal number of clusters, and perform equivalent modeling on the photovoltaic power generation units based on the classification aggregation results to obtain a multi-machine equivalent model.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the photovoltaic station classification and aggregation equivalent modeling method according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium storing a computer program, wherein when the computer program is executed by a processor, the photovoltaic station classification aggregation equivalent modeling method according to any one of claims 1 to 7 is implemented.

Images