Distributed photovoltaic cluster division method and device

Abstract

The invention provides a distributed photovoltaic cluster division method and device, and the method comprises the steps: obtaining the modularity, power balance degree and line impedance distance of a distributed photovoltaic system according to the information of each photovoltaic power generation node in the distributed photovoltaic system; and dividing the distributed photovoltaic system into a plurality of clusters by using an improved fuzzy clustering algorithm based on feature weighting according to the modularity, the power balance degree and the line impedance distance. According to the method, multiple indexes are used for cluster division, so that electrical connection between nodes in the clusters is close, connection between the clusters is loose, and operation management of the clusters is facilitated; the improved fuzzy clustering algorithm based on feature weighting is adopted to quantify each clustering index, the weight of each index is obtained, cluster division is achieved through iteration, and the accuracy of distributed photovoltaic cluster division is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of power systems, and in particular to a distributed photovoltaic cluster division method and device. Background Art

[0002] As the penetration rate of distributed photovoltaic power generation in the distribution network increases, the problems of power flow reverse transmission and overvoltage are particularly prominent, and the distributed photovoltaic clusters need to be regulated. Therefore, the division of distributed photovoltaic clusters is of great significance to improving the overall benefits of distributed photovoltaic systems and the regulation between clusters.

[0003] The research on distributed photovoltaic cluster partitioning methods mainly includes two aspects: cluster partitioning index and partitioning algorithm. Most existing research schemes are based on a single index for cluster partitioning, and the partitioning algorithm has randomness in the selection of cluster centers and the number of partitioned clusters, which makes it easy for isolated nodes to appear in the distributed photovoltaic cluster partitioning, and the partitioning accuracy is not high. Summary of the invention

[0004] The present invention provides a distributed photovoltaic cluster division method and device, which are used to solve the defects of the distributed photovoltaic cluster division in the prior art that isolated nodes are easy to appear and the division accuracy is not high, and realize the effective and accurate division of the distributed photovoltaic cluster.

[0005] The present invention provides a distributed photovoltaic cluster division method, comprising:

[0006] According to the information of each photovoltaic power generation node in the distributed photovoltaic system, the modularity, power balance and line impedance distance of the distributed photovoltaic system are obtained;

[0007] According to the modularity, power balance and line impedance distance, the distributed photovoltaic system is divided into a plurality of clusters using an improved fuzzy clustering algorithm based on feature weighting.

[0008] According to a distributed photovoltaic cluster division method provided by the present invention, the step of obtaining the modularity of the distributed photovoltaic system includes:

[0009] Determining the electrical distance between any two photovoltaic power generation nodes in the distributed photovoltaic system;

[0010] Determining the weight of the edge between the arbitrary two photovoltaic power generation nodes according to the electrical distance between the arbitrary two photovoltaic power generation nodes;

[0011] Determine the sum of the weights of the edges connecting each photovoltaic power generation node in the arbitrary two photovoltaic power generation nodes;

[0012] The modularity is determined according to the weight of the edge between the arbitrary two photovoltaic power generation nodes, the weight sum, and whether the arbitrary two photovoltaic power generation nodes belong to the same cluster.

[0013] According to a distributed photovoltaic cluster division method provided by the present invention, the step of determining the electrical distance between any two photovoltaic power generation nodes in the distributed photovoltaic system includes:

[0014] When determining the power change of each photovoltaic power generation node in the distributed photovoltaic system, the ratio between the voltage change value of each photovoltaic power generation node and the voltage change value of each photovoltaic power generation node in the arbitrary two photovoltaic power generation nodes;

[0015] Determine the difference between the ratios corresponding to the two photovoltaic power generation nodes to the photovoltaic power generation nodes;

[0016] The electrical distance between the arbitrary two photovoltaic power generation nodes is determined according to the difference.

[0017] According to a distributed photovoltaic cluster division method provided by the present invention, the step of obtaining the power balance degree of the distributed photovoltaic system includes:

[0018] Determining the net power of each cluster according to the net power of the photovoltaic power generation nodes in each cluster within a preset time period;

[0019] Determining the active power balance degree of the distributed photovoltaic system according to the net power of each cluster;

[0020] Determining the reactive power supply value of each cluster according to the reactive power output by each photovoltaic power generation node in each cluster;

[0021] Determining the reactive power demand value of each cluster according to the reactive power demand value of each photovoltaic power generation node in each cluster;

[0022] Determining the reactive power balance of each cluster according to the reactive power supply value and the reactive power demand value of each cluster;

[0023] Determining the reactive power balance degree of the distributed photovoltaic system according to the reactive power balance degree of each cluster;

[0024] The power balance degree of the distributed photovoltaic system is determined according to the active power balance degree and the reactive power balance degree of the distributed photovoltaic system.

[0025] According to a distributed photovoltaic cluster division method provided by the present invention, the step of obtaining the line impedance distance of the distributed photovoltaic system includes:

[0026] Determine the ratio between the voltage and current of any two photovoltaic power generation nodes in the distributed photovoltaic system, and determine the line impedance distance of the distributed photovoltaic system according to the difference between the ratios corresponding to the any two photovoltaic power generation nodes; or,

[0027] The self-impedance of each of the arbitrary two photovoltaic power generation nodes and the mutual impedance between the arbitrary two photovoltaic power generation nodes are determined, and the line impedance distance of the distributed photovoltaic system is determined according to the self-impedance and the mutual impedance of the arbitrary two photovoltaic power generation nodes.

[0028] According to a distributed photovoltaic cluster division method provided by the present invention, the distributed photovoltaic system is divided into multiple clusters using an improved fuzzy clustering algorithm based on feature weighting according to the modularity, power balance and line impedance distance, including:

[0029] According to the membership of each photovoltaic power generation node in the distributed photovoltaic system to each cluster, a membership matrix is ​​constructed;

[0030] Constructing a cluster center matrix according to the cluster centers of each cluster;

[0031] Constructing a weight matrix according to the weight of the modularity, the weight of the power balance degree and the weight of the line impedance distance;

[0032] Constructing a priori weight matrix according to the a priori weight of the modularity, the a priori weight of the power balance degree and the a priori weight of the line impedance distance determined by the gradient descent method;

[0033] Determining the intra-cluster divergence of the distributed photovoltaic system according to the distance between each photovoltaic power generation node and the cluster center of each cluster, the membership degree of each cluster and the weight matrix;

[0034] Determine a difference between the weight matrix and the prior weight matrix, and determine an objective function according to the intra-cluster divergence and the difference;

[0035] According to the objective function, using the Lagrange multiplier method to obtain the iterative formula of the membership degree, the iterative formula of the cluster center and the iterative formula of each weight;

[0036] The membership matrix is ​​updated according to the iterative formula of the membership, the cluster center matrix is ​​updated according to the iterative formula of the cluster center, and the weight matrix is ​​updated according to the iterative formula of each weight until a preset condition is met;

[0037] The distributed photovoltaic system is divided into a plurality of clusters according to the updated membership matrix.

[0038] According to a distributed photovoltaic cluster division method provided by the present invention, after the distributed photovoltaic system is divided into a plurality of clusters using an improved fuzzy clustering algorithm based on feature weighting according to the modularity, power balance degree and line impedance distance, the method further includes:

[0039] One or more of a DBI index, a Xie-Beni index, a partition coefficient, and a classification entropy are determined for the plurality of clusters.

[0040] The present invention also provides a distributed photovoltaic cluster division device, comprising:

[0041] An acquisition module, used to acquire the modularity, power balance and line impedance distance of the distributed photovoltaic system according to the information of each photovoltaic power generation node in the distributed photovoltaic system;

[0042] A partitioning module is used to partition the distributed photovoltaic system into a plurality of clusters according to the modularity, power balance and line impedance distance using an improved fuzzy clustering algorithm based on feature weighting.

[0043] The present invention also provides an electronic 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 program, any of the above-mentioned distributed photovoltaic cluster division methods is implemented.

[0044] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the distributed photovoltaic cluster division method as described in any one of the above is implemented.

[0045] The present invention also provides a computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the distributed photovoltaic cluster division method as described above is implemented.

[0046] The distributed photovoltaic cluster division method and device provided by the present invention consider the structural strength of the cluster, use the modularity index, consider that the nodes inside the cluster can achieve active and reactive complementarity, use the power balance index, consider the line impedance, use the line impedance distance index between nodes, and use the three indicators to construct a comprehensive performance index system for cluster division, so that the electrical connection between the nodes inside the cluster is close, and the connection between clusters is loose, which is convenient for cluster operation management; the comprehensive performance index system is input into the improved fuzzy clustering algorithm based on feature weighting, each clustering index is quantified, and the weight of each index is obtained. Cluster division is implemented through iteration, thereby improving the accuracy of distributed photovoltaic cluster division. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0048] Figure 1 It is a schematic diagram of the process of the distributed photovoltaic cluster division method provided by the present invention;

[0049] Figure 2 It is a schematic diagram of constructing a comprehensive clustering index system in the distributed photovoltaic cluster division method provided by the present invention;

[0050] Figure 3 It is a schematic diagram of the calculation process of the electrical distance in the distributed photovoltaic cluster division method provided by the present invention;

[0051] Figure 4 It is a schematic diagram of the flow calculation process in the distributed photovoltaic cluster division method provided by the present invention;

[0052] Figure 5 It is a flow chart of the Newton-Raphson power flow calculation algorithm in the distributed photovoltaic cluster partitioning method provided by the present invention;

[0053] Figure 6 It is a schematic diagram of a distributed photovoltaic cluster division process based on feature learning in the distributed photovoltaic cluster division method provided by the present invention;

[0054] Figure 7 It is a structural schematic diagram of a distributed photovoltaic cluster division device provided by the present invention;

[0055] Figure 8 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0056] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0057] The growing demand for new energy and the public's concern about environmental issues have promoted the rapid development of photovoltaic power generation. The penetration rate of distributed photovoltaic power generation in distribution networks is getting higher and higher. However, the stable operation of distribution networks also faces many challenges, among which power flow backflow and overvoltage problems are particularly prominent.

[0058] Flow backflow is when the power generation of the distributed energy system exceeds the demand on the load side, the excess power will flow to the power supply side, resulting in flow backflow. When flow backflow occurs, it means that the control strategy is abnormal and needs to be readjusted to ensure the stability and safety of the system.

[0059] Overvoltage refers to instantaneous or continuous voltage fluctuations that exceed the voltage in the power system or electrical equipment. Overvoltage not only directly affects the safe and stable operation of the line, but also affects the photovoltaic absorption capacity of the distribution network and the operating efficiency of the photovoltaic system itself.

[0060] It can be seen that the distributed photovoltaic cluster division scheme is of great significance for improving the overall benefits of distributed photovoltaic systems and the regulation between clusters, reducing the risk of system failures, and ensuring the safe and sustainable operation of the system.

[0061] The present invention mainly studies the distributed photovoltaic cluster partitioning method, including two aspects: cluster partitioning index and partitioning algorithm. The goal of cluster partitioning is to ensure that the internal connection of the cluster is close and the connection between clusters is sparse. In terms of cluster partitioning index, there are mainly electrical distance indexes, electrical modularity indexes, active balance indexes and reactive balance indexes based on address location or administrative area distance, Newton-Raphson power flow calculation of reactive power and active power to voltage. In terms of partitioning algorithm, the current cluster partitioning algorithms mainly include community discovery algorithm, intelligent optimization algorithm, k-means algorithm and fuzzy clustering algorithm.

[0062] Most existing research schemes are based on a single indicator, and the partitioning algorithm is random in the selection of cluster centers and the number of partitioned clusters, which leads to isolated nodes easily appearing after using the partitioning algorithm and the partitioning accuracy is not high.

[0063] There are many influencing factors in the distributed photovoltaic scenario, including temperature, light irradiation intensity, address location, cluster modularity and net load, etc. The present invention comprehensively considers the impact of various factors on cluster division, combines the improved fuzzy clustering algorithm based on feature weighting, and constructs a comprehensive division index system for each influencing factor through weights to make up for the defects in distributed photovoltaic cluster division.

[0064] Combine the following Figure 1 A distributed photovoltaic cluster division method of the present invention is described, comprising:

[0065] Step 101, obtaining the modularity, power balance and line impedance distance of the distributed photovoltaic system according to the information of each photovoltaic power generation node in the distributed photovoltaic system;

[0066] Load the data set of distributed photovoltaic system to obtain the numerical items such as active power and voltage of distributed photovoltaic system. Use Newton-Raphson power flow calculation method to calculate the electrical sensitivity matrix composed of the influence of active power and reactive power on voltage amplitude in distributed photovoltaic system, and then construct electrical distance matrix, and combine the modularity concept to obtain the modularity index of distributed photovoltaic system.

[0067] Load the data set of the distributed photovoltaic system, obtain the net power numerical item of the distributed photovoltaic system, and combine the active balance index formula to obtain the active balance of the distributed photovoltaic system. According to the connectivity between nodes and the reactive balance index formula, the reactive balance of the distributed photovoltaic system is obtained. The active balance and reactive balance can be averaged to obtain the power balance index of the distributed photovoltaic system;

[0068] According to the admittance matrix obtained in the process of calculating the modularity index, the impedance distance matrix of the distributed photovoltaic system is obtained.

[0069] The obtained modularity index, power balance index and line impedance distance index of the distributed photovoltaic system constitute a comprehensive performance index system.

[0070] Step 102: Divide the distributed photovoltaic system into a plurality of clusters using an improved fuzzy clustering algorithm based on feature weighting according to the modularity, power balance and line impedance distance.

[0071] The comprehensive performance index system is passed as a parameter into the improved fuzzy clustering algorithm based on feature weighting. The algorithm quantifies the weight corresponding to each index and obtains the clustering result after iteration. The fuzzy clustering algorithm based on feature learning is applied to the field of distributed photovoltaic cluster division for cluster control.

[0072] like Figure 2 As shown, this embodiment selects a distributed photovoltaic cluster division scenario, designs the active power balance within the cluster and the requirements for voltage control of the subsequent operation scheduling, takes full advantage of the cluster's autonomy as the goal, conducts in-depth research on cluster division criteria, indicators and algorithm implementation processes, and proposes a comprehensive performance indicator system and an improved fuzzy clustering method. This method comprehensively considers the complementarity and correlation between nodes within the cluster, and on the basis of ensuring a reasonable match between source and load within the cluster, reduces line losses after division, and ensures the coupling connection and voltage regulation capability between nodes.

[0073] Specifically, the proposed method meets the following principles: ① Logical principle, that is, there are no isolated nodes within the cluster, there are no overlapping nodes between clusters, and connectivity must be guaranteed between nodes; ② Structural principle, that is, in terms of geographical space or electrical coupling, the internal connections of the cluster are close, and the connections between clusters are sparse; ③ Functional principle, that is, the characteristics of the cluster are comprehensively expressed by the characteristics of each node in the cluster, and the nodes in the cluster have good collaboration capabilities; ④ Minimum loss principle, that is, to minimize the loss within the cluster.

[0074] In common hard clustering and soft clustering algorithms, each feature of the data is usually considered equally important. The algorithm assigns the same feature weight to different features, without considering that different features should have different importance. In the improved fuzzy clustering algorithm based on feature weighting, feature learning technology is used to improve their clustering performance. Feature learning technology is usually divided into feature selection and feature weighting. Feature selection technology assumes that each selected feature has the same importance. As an extension of feature selection, feature weighting allows feature weights to take values ​​in the interval [0,1], and the more important the feature, the larger the weight value.

[0075] Feature learning allows the algorithm to automatically learn and obtain different feature weights when building a clustering algorithm or training a classifier. In cluster analysis, feature learning methods are used to expand k-means and FCM (Fuzzy C-means clustering algorithm).

[0076] This embodiment considers the structural strength of the cluster, uses the modularity index, considers that the nodes within the cluster can achieve active and reactive complementarity, uses the power balance index, considers the line impedance, uses the line impedance distance index between nodes, and uses the comprehensive performance index system constructed by the three indicators to perform cluster division, so that the electrical connection between the nodes within the cluster is close, and the connection between clusters is loose, which is convenient for cluster operation management; the comprehensive performance index system is input into the improved fuzzy clustering algorithm based on feature weighting, each clustering index is quantified, and the weight of each index is obtained. Cluster division is achieved through iteration, thereby improving the accuracy of distributed photovoltaic cluster division.

[0077] Based on the above embodiment, the steps of obtaining the modularity of the distributed photovoltaic system in this embodiment include:

[0078] Determining the electrical distance between any two photovoltaic power generation nodes in the distributed photovoltaic system;

[0079] Determining the weight of the edge between the arbitrary two photovoltaic power generation nodes according to the electrical distance between the arbitrary two photovoltaic power generation nodes;

[0080] Determine the sum of the weights of the edges connecting each photovoltaic power generation node in the arbitrary two photovoltaic power generation nodes;

[0081] The modularity is determined according to the weight of the edge between the arbitrary two photovoltaic power generation nodes, the weight sum, and whether the arbitrary two photovoltaic power generation nodes belong to the same cluster.

[0082] The concept of modularity ρ is used and extended to weighted network partitioning to measure the strength of the community structure of complex networks. The larger the modularity index, the higher the degree of coupling within the cluster; the smaller the modularity index, the lower the degree of coupling between clusters. The definition of modularity is as follows:

[0083]

[0084] Among them, e ij is the weight of the edge between nodes i and j, by changing e ij The modularity function will have a different meaning if k i is the sum of the weights of all edges connecting node i; m is the sum of the weights of the edges connecting all nodes in the network; when nodes i and j are in the same distributed generation cluster (DGC), δ(i, j) = 1, otherwise δ(i, j) = 0.

[0085] Measuring the structural performance of the cluster only based on the topological relationship of the distribution network lacks consideration of the electrical connection between nodes. The electrical distance can be used to measure the tightness of the electrical coupling between nodes. The weight of the edge between photovoltaic power generation nodes is determined according to the electrical distance between the two nodes. The longer the electrical distance, the smaller the weight of the edge; the closer the electrical distance, the greater the weight of the edge.

[0086] Based on the above embodiment, the method of determining the electrical distance between any two photovoltaic power generation nodes in the distributed photovoltaic system in this embodiment includes:

[0087] When determining the power change of each photovoltaic power generation node in the distributed photovoltaic system, the ratio between the voltage change value of each photovoltaic power generation node and the voltage change value of each photovoltaic power generation node in the arbitrary two photovoltaic power generation nodes;

[0088] Determine the difference between the ratios corresponding to the two photovoltaic power generation nodes to the photovoltaic power generation nodes;

[0089] The electrical distance between the arbitrary two photovoltaic power generation nodes is determined according to the difference.

[0090] The specific expression for using electrical distance to measure the tightness of electrical coupling between nodes is:

[0091] ΔV=S QV ΔQ;

[0092]

[0093] Among them, S QV is the sensitivity matrix, S QV，ij is the element in the i-th row and j-th column of the matrix, indicating the voltage change value of node i caused by the change of node j power; ΔQ is the change of reactive power, ΔV is the change of voltage; d ij is the ratio of the voltage change at node j to the voltage change at node i when the reactive power at node j changes.

[0094] Considering that the electrical distance between two nodes is not only related to themselves, but also to other nodes in the network. Assuming there are n nodes in the distributed photovoltaic system, the electrical distance between node i and node i is defined as:

[0095]

[0096] Among them, L ij is the electrical distance between node i and node j. The larger the value, the farther the electrical distance between nodes is, and the smaller the edge weight in the corresponding modularity function is, and vice versa. The calculation process of electrical distance is as follows: Figure 3 As shown in the figure, the power flow calculation process in the electrical distance calculation process is as follows Figure 4 As shown in the figure, the schematic diagram of the Newton-Raphson power flow calculation algorithm in the power flow calculation process is as follows Figure 5 shown.

[0097] The relationship between the edge weight and the electrical distance between nodes is expressed as:

[0098] e ij =1-L ij / max(L);

[0099] Where L is the electrical distance matrix.

[0100] Based on the above embodiment, the steps of obtaining the power balance degree of the distributed photovoltaic system in this embodiment include:

[0101] Determining the net power of each cluster according to the net power of the photovoltaic power generation nodes in each cluster within a preset time period;

[0102] Determining the active power balance degree of the distributed photovoltaic system according to the net power of each cluster;

[0103] Determining the reactive power supply value of each cluster according to the reactive power output by each photovoltaic power generation node in each cluster;

[0104] Determining the reactive power demand value of each cluster according to the reactive power demand value of each photovoltaic power generation node in each cluster;

[0105] Determining the reactive power balance of each cluster according to the reactive power supply value and the reactive power demand value of each cluster;

[0106] Determining the reactive power balance degree of the distributed photovoltaic system according to the reactive power balance degree of each cluster;

[0107] The power balance degree of the distributed photovoltaic system is determined according to the active power balance degree and the reactive power balance degree of the distributed photovoltaic system.

[0108] The access of a large number of distributed generation (DG) may cause problems such as power reverse transmission and voltage over-limit in the distribution network. Power reverse transmission will not only sharply increase line losses, but also pose the risk of equipment overload and burning.

[0109] When dividing clusters, the power balance problem within the cluster should be fully considered, the active power transmission of the cluster should be reduced, and a certain voltage regulation capability should be provided when the voltage exceeds the limit. When the voltage exceeds the limit, the voltage level should be restored by adjusting the reactive power first; when the reactive power regulation margin is exceeded, the active power should be adjusted to maintain the stable operation of the distribution network.

[0110] Therefore, the power balance index needs to consider the balance degree of active and reactive power within the cluster. The larger the active and reactive power balance index is, the higher the power balance degree within the cluster is. The formula for the power balance index is:

[0111] β=(β P +β Q ) / 2;

[0112] Among them, β is the power balance index; β P is the total active power balance index of the distributed photovoltaic system; β Q It is the total reactive power balance index of the distributed photovoltaic system.

[0113] The active power balance index is used to describe the balance relationship between active power demand and supply within the cluster. The calculation formula is:

[0114]

[0115] Among them, P clu (t) i is the net power characteristic of cluster i, T is the duration of the typical time-varying scenario, and c represents the number of clusters.

[0116] The reactive power balance index is used to describe the balance between reactive power demand and supply within the cluster:

[0117]

[0118]

[0119] Among them, Q sup，i and Q need，i are the maximum reactive power output and reactive power demand of the ith cluster respectively; Q i is the reactive balance degree of cluster i.

[0120] Based on the above embodiment, the step of obtaining the line impedance distance of the distributed photovoltaic system in this embodiment includes:

[0121] Determine the ratio between the voltage and current of any two photovoltaic power generation nodes in the distributed photovoltaic system, and determine the line impedance distance of the distributed photovoltaic system according to the difference between the ratios corresponding to the any two photovoltaic power generation nodes; or,

[0122] The self-impedance of each of the arbitrary two photovoltaic power generation nodes and the mutual impedance between the arbitrary two photovoltaic power generation nodes are determined, and the line impedance distance of the distributed photovoltaic system is determined according to the self-impedance and the mutual impedance of the arbitrary two photovoltaic power generation nodes.

[0123] For large-scale distributed photovoltaic power stations, line loss is a factor that cannot be ignored. Considering that the topological structure of distributed photovoltaic power stations is often complex, and the location of the access point where the photovoltaic power generation unit is located will affect the dynamic characteristics of the photovoltaic power generation unit. This embodiment considers the equivalent line impedance of the common connection point of each photovoltaic power generation unit, so that photovoltaic power generation units with similar distances to the common connection point are classified into one category.

[0124] The input impedance Z′ of the two-port network can be used ij (i≠j) is used to represent the electrical distance between non-grounded nodes.

[0125] Z′ ij The calculation of the value can be expressed as the difference between the voltage and current ratios between the two nodes when a certain current is injected into nodes i and j.

[0126] Z′ ij You can also use the node self-impedance Z ii and mutual impedance Z ij To express, that is, Z′ ij =Z ii +Z jj -2Z ij The electrical distance represented by this method effectively solves the drawbacks of using only the node impedance matrix to represent the electrical distance, and the result is more representative.

[0127] According to the modularity index, power balance index and impedance distance, a weight-based objective function system is constructed. In order to ensure the structural and functional characteristics of the cluster while ensuring the balance of cluster size, a comprehensive cluster partitioning index is proposed by combining electrical modularity, power balance and impedance distance between nodes as the objective function of cluster partitioning. The expression of the comprehensive performance index τ is shown in the formula:

[0128] τ=w1ρ+w2β+w3Z;

[0129] Among them, w1, w2 and w3 are weight coefficients of each indicator, and w1+w2+w3=1. According to different clustering objectives, different weight coefficients are selected to obtain corresponding clustering results.

[0130] On the basis of the above embodiments, in this embodiment, the distributed photovoltaic system is divided into multiple clusters using an improved fuzzy clustering algorithm based on feature weighting according to the modularity, power balance and line impedance distance, including:

[0131] According to the membership of each photovoltaic power generation node in the distributed photovoltaic system to each cluster, a membership matrix is ​​constructed;

[0132] Constructing a cluster center matrix according to the cluster centers of each cluster;

[0133] Constructing a weight matrix according to the weight of the modularity, the weight of the power balance degree and the weight of the line impedance distance;

[0134] Constructing a priori weight matrix according to the a priori weight of the modularity, the a priori weight of the power balance degree and the a priori weight of the line impedance distance determined by the gradient descent method;

[0135] Determining the intra-cluster divergence of the distributed photovoltaic system according to the distance between each photovoltaic power generation node and the cluster center of each cluster, the membership degree of each cluster and the weight matrix;

[0136] Determine a difference between the weight matrix and the prior weight matrix, and determine an objective function according to the intra-cluster divergence and the difference;

[0137] According to the objective function, using the Lagrange multiplier method to obtain the iterative formula of the membership degree, the iterative formula of the cluster center and the iterative formula of each weight;

[0138] The membership matrix is ​​updated according to the iterative formula of the membership, the cluster center matrix is ​​updated according to the iterative formula of the cluster center, and the weight matrix is ​​updated according to the iterative formula of each weight until a preset condition is met;

[0139] The distributed photovoltaic system is divided into a plurality of clusters according to the updated membership matrix.

[0140] The gradient descent method is used to find the optimal solution of the feature evaluation function, and the solution is used as the prior weight of the feature. In the objective function of this embodiment, the intra-cluster divergence and the difference between the cluster weight and the feature prior weight are integrated, and then the local optimal solution of the objective function is found by cross iteration.

[0141] set up is the sample point x i and x j and define as follows:

[0142]

[0143] in

[0144] α is a positive parameter and can be calculated by the following formula:

[0145]

[0146] Optimal It is obtained by minimizing the following objective function:

[0147]

[0148] Where n is the number of samples in the data set, represents the feature vector, Represents the sample point x i and x j The similarity between Indicates when hour The value of .

[0149] When all feature weights are 1, the similarity between two sample points is called the old similarity. Now the weight value of each dimension of the data is regarded as a parameter, and their value range is [0, +∞). The similarity between the two sample points is called the new similarity. The similarity between two sample points varies depending on the weight parameter.

[0150] The process of using the gradient descent method to obtain the optimal solution of the above objective function is described as follows:

[0151]

[0152]

[0153]

[0154] set up for The increase of It can be calculated by the following formula:

[0155]

[0156] Where η is the learning rate and is determined by

[0157]

[0158] Let X = {x i |i=1,2,...,n) is R d The data set in space, the feature prior weight vector is recorded as The formula of the objective function is:

[0159]

[0160] Among them, U=[u ik ] c×n is the membership matrix, element u ik Represents the sample point x i The membership degree of the k-th cluster. V = [v1, v2, ..., v c ] T =[v kj ] c×d is the cluster center matrix, representing c cluster centers, and m is the fuzzy index (where m>1). ik Satisfy constraints as well as The first term in the above formula is the intra-cluster divergence, the second term is the difference between the structural characteristics of the feature prior weight and the cluster weight, and the γ parameter is used to balance the first and second terms.

[0161] The optimal partitioning and clustering problem of sample points can be described as a constrained optimization problem, that is, finding the optimal U, V, and W to meet the above constraints. Since U, V, and W are continuous variables, the Lagrange multiplier method is used to find the first-order partial derivative of the Lagrange function with respect to U, V, and W, and set it equal to 0, and then u is obtained. ik The iteration formula is:

[0162]

[0163] Get the cluster center v kj The iteration formula is:

[0164]

[0165] The iterative formula for feature weights is:

[0166]

[0167] like Figure 6 As shown in the figure, the distributed photovoltaic cluster division based on feature learning includes the following steps:

[0168] Initialization: Fixed ε>0, given the number of clusters c, randomly initialize V(0) and feature weights w in the clustering (0) , set t=0.

[0169] Calculate the membership matrix U (t) .

[0170] Update cluster center V (t) .

[0171] Update weight w (t) .

[0172] If ||U (t) -U (t-1) ||<ε, then exit; otherwise, set t=t+1 and return to the step of calculating the membership matrix.

[0173] On the basis of the above embodiments, after the distributed photovoltaic system is divided into a plurality of clusters according to the modularity, power balance and line impedance distance using an improved fuzzy clustering algorithm based on feature weighting, the present embodiment further includes:

[0174] Determine the DBI (Davies-Bouldin Index, Davidson Bouldin Index) index, Xie-Beni index, and partition coefficient K of the multiple clusters PC and classification entropy K CE One or more of .

[0175] The calculation process of the DBI indicator involves the distance between cluster centers and the distance between nodes within a cluster. It regards each cluster as a spherical area and measures the separation of clusters by calculating the distance between all cluster balls. At the same time, it also considers the distance between the nodes within each cluster and the cluster center to measure the compactness of the cluster. The calculation formula of the DBI indicator is as follows:

[0176]

[0177] Among them, D i It can represent the distance between the i-th clustering ball and other clustering balls, as well as the sum of the distances between the nodes inside the i-th clustering ball and the cluster center.

[0178] The Xie-Beni index is the most representative evaluation index in the fuzzy clustering effectiveness evaluation function of the data set structure. The calculation formula is as follows:

[0179]

[0180] Introducing the partition coefficient K PC and classification entropy K CE The quality of clustering and the optimal number of clusters are evaluated and defined as follows:

[0181]

[0182]

[0183] The distributed photovoltaic cluster division device provided by the present invention is described below. The distributed photovoltaic cluster division device described below and the distributed photovoltaic cluster division method described above can be referred to each other.

[0184] like Figure 7 As shown, the device includes an acquisition module 701 and a division module 702, wherein:

[0185] The acquisition module 701 is used to acquire the modularity, power balance and line impedance distance of the distributed photovoltaic system according to the information of each photovoltaic power generation node in the distributed photovoltaic system;

[0186] The division module 702 is used to divide the distributed photovoltaic system into multiple clusters according to the modularity, power balance and line impedance distance by using an improved fuzzy clustering algorithm based on feature weighting.

[0187] The present invention considers the structural strength of the cluster, uses the modularity index, considers that active and reactive complementarity can be achieved between nodes within the cluster, uses the power balance index, considers the line impedance, uses the line impedance distance index between nodes, and uses a comprehensive performance index system constructed by the three indicators to perform cluster division, so that the electrical connection between nodes within the cluster is close, and the connection between clusters is loose, which is convenient for cluster operation management; the comprehensive performance index system is input into an improved fuzzy clustering algorithm based on feature weighting, each clustering index is quantified, and the weight of each index is obtained. Cluster division is achieved through iteration, thereby improving the accuracy of distributed photovoltaic cluster division.

[0188] Figure 8 An example of a physical structure diagram of an electronic device is shown in FIG. Figure 8As shown, the electronic device may include: a processor 810, a communication interface 820, a memory 830 and a communication bus 840, wherein the processor 810, the communication interface 820 and the memory 830 communicate with each other through the communication bus 840. The processor 810 may call the logic instructions in the memory 830 to execute the distributed photovoltaic cluster division method, which includes: obtaining the modularity, power balance and line impedance distance of the distributed photovoltaic system according to the information of each photovoltaic power generation node in the distributed photovoltaic system; and dividing the distributed photovoltaic system into multiple clusters using an improved fuzzy clustering algorithm based on feature weighting according to the modularity, power balance and line impedance distance.

[0189] In addition, the logic instructions in the above-mentioned memory 830 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0190] On the other hand, the present invention also provides a computer program product, which includes a computer program, which can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the distributed photovoltaic cluster division method provided by the above methods, and the method includes: according to the information of each photovoltaic power generation node in the distributed photovoltaic system, obtaining the modularity, power balance and line impedance distance of the distributed photovoltaic system; according to the modularity, power balance and line impedance distance, using an improved fuzzy clustering algorithm based on feature weighting to divide the distributed photovoltaic system into multiple clusters.

[0191] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to execute the distributed photovoltaic cluster division method provided by the above-mentioned methods, the method comprising: obtaining the modularity, power balance and line impedance distance of the distributed photovoltaic system according to the information of each photovoltaic power generation node in the distributed photovoltaic system; and dividing the distributed photovoltaic system into multiple clusters using an improved fuzzy clustering algorithm based on feature weighting according to the modularity, power balance and line impedance distance.

[0192] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0193] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0194] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A distributed photovoltaic cluster division method, characterized in that: include: According to the information of each photovoltaic power generation node in the distributed photovoltaic system, the modularity, power balance and line impedance distance of the distributed photovoltaic system are obtained; According to the modularity, power balance and line impedance distance, the distributed photovoltaic system is divided into a plurality of clusters using an improved fuzzy clustering algorithm based on feature weighting.

2. The distributed photovoltaic cluster division method according to claim 1, characterized in that: The steps of obtaining the modularity of the distributed photovoltaic system include: Determining the electrical distance between any two photovoltaic power generation nodes in the distributed photovoltaic system; Determining the weight of the edge between the arbitrary two photovoltaic power generation nodes according to the electrical distance between the arbitrary two photovoltaic power generation nodes; Determine the sum of the weights of the edges connecting each photovoltaic power generation node in the arbitrary two photovoltaic power generation nodes; The modularity is determined according to the weight of the edge between the arbitrary two photovoltaic power generation nodes, the weight sum, and whether the arbitrary two photovoltaic power generation nodes belong to the same cluster.

3. The distributed photovoltaic cluster division method according to claim 2, characterized in that: The determining the electrical distance between any two photovoltaic power generation nodes in the distributed photovoltaic system comprises: When determining the power change of each photovoltaic power generation node in the distributed photovoltaic system, the ratio between the voltage change value of each photovoltaic power generation node and the voltage change value of each photovoltaic power generation node in the arbitrary two photovoltaic power generation nodes; Determine the difference between the ratios corresponding to the two photovoltaic power generation nodes to the photovoltaic power generation nodes; The electrical distance between the arbitrary two photovoltaic power generation nodes is determined according to the difference.

4. The distributed photovoltaic cluster division method according to claim 1, characterized in that: The step of obtaining the power balance degree of the distributed photovoltaic system includes: Determining the net power of each cluster according to the net power of the photovoltaic power generation nodes in each cluster within a preset time period; Determining the active power balance degree of the distributed photovoltaic system according to the net power of each cluster; Determining the reactive power supply value of each cluster according to the reactive power output by each photovoltaic power generation node in each cluster; Determining the reactive power demand value of each cluster according to the reactive power demand value of each photovoltaic power generation node in each cluster; Determining the reactive power balance of each cluster according to the reactive power supply value and the reactive power demand value of each cluster; Determining the reactive power balance degree of the distributed photovoltaic system according to the reactive power balance degree of each cluster; The power balance degree of the distributed photovoltaic system is determined according to the active power balance degree and the reactive power balance degree of the distributed photovoltaic system.

5. The distributed photovoltaic cluster division method according to claim 1, characterized in that: The step of obtaining the line impedance distance of the distributed photovoltaic system includes: Determine the ratio between the voltage and current of any two photovoltaic power generation nodes in the distributed photovoltaic system, and determine the line impedance distance of the distributed photovoltaic system according to the difference between the ratios corresponding to the any two photovoltaic power generation nodes; or, The self-impedance of each of the arbitrary two photovoltaic power generation nodes and the mutual impedance between the arbitrary two photovoltaic power generation nodes are determined, and the line impedance distance of the distributed photovoltaic system is determined according to the self-impedance and the mutual impedance of the arbitrary two photovoltaic power generation nodes.

6. The distributed photovoltaic cluster division method according to any one of claims 1 to 5, characterized in that: The distributed photovoltaic system is divided into a plurality of clusters using an improved fuzzy clustering algorithm based on feature weighting according to the modularity, power balance and line impedance distance, including: According to the membership of each photovoltaic power generation node in the distributed photovoltaic system to each cluster, a membership matrix is ​​constructed; Constructing a cluster center matrix according to the cluster centers of each cluster; Constructing a weight matrix according to the weight of the modularity, the weight of the power balance degree and the weight of the line impedance distance; Constructing a priori weight matrix according to the a priori weight of the modularity, the a priori weight of the power balance degree and the a priori weight of the line impedance distance determined by the gradient descent method; Determining the intra-cluster divergence of the distributed photovoltaic system according to the distance between each photovoltaic power generation node and the cluster center of each cluster, the membership degree of each cluster and the weight matrix; Determine a difference between the weight matrix and the prior weight matrix, and determine an objective function according to the intra-cluster divergence and the difference; According to the objective function, the iterative formula of the membership degree, the iterative formula of the cluster center and the iterative formula of each weight are obtained using the Lagrange multiplier method; The membership matrix is ​​updated according to the iterative formula of the membership, the cluster center matrix is ​​updated according to the iterative formula of the cluster center, and the weight matrix is ​​updated according to the iterative formula of each weight until a preset condition is met; The distributed photovoltaic system is divided into a plurality of clusters according to the updated membership matrix.

7. The distributed photovoltaic cluster division method according to any one of claims 1 to 5, characterized in that: After dividing the distributed photovoltaic system into a plurality of clusters according to the modularity, power balance and line impedance distance using an improved fuzzy clustering algorithm based on feature weighting, the method further includes: One or more of a DBI index, a Xie-Beni index, a partition coefficient, and a classification entropy are determined for the plurality of clusters.

8. A distributed photovoltaic cluster division device, characterized in that: include: An acquisition module, used to acquire the modularity, power balance and line impedance distance of the distributed photovoltaic system according to the information of each photovoltaic power generation node in the distributed photovoltaic system; A partitioning module is used to partition the distributed photovoltaic system into a plurality of clusters according to the modularity, power balance and line impedance distance by using an improved fuzzy clustering algorithm based on feature weighting.

9. An electronic 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 program, the distributed photovoltaic cluster division method according to any one of claims 1 to 7 is implemented.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the distributed photovoltaic cluster division method according to any one of claims 1 to 7 is implemented.

Images