A distributed photovoltaic power generation cluster division method considering grid-connected management mode

Abstract

The present invention proposes a distributed photovoltaic power cluster division method considering the grid-connected management mode, collects information such as line parameters, topological structure, node type and distributed photovoltaic power grid-connected management mode of the distribution network; performs calculation and analysis based on the collected information, and organizes and forms a network structure diagram of the cluster distribution network to be divided; classifies the distributed photovoltaic power according to the grid-connected management mode; takes the maximum modularity of the distributed photovoltaic power cluster as the goal, establishes a distributed photovoltaic power cluster division model considering the grid-connected management mode, and uses a genetic algorithm to solve to obtain a cluster division plan. The present invention combines the impact of distributed photovoltaic power on operation and management due to different grid-connected management modes in actual situations, and performs cluster division on the basis of considering the grid-connected management mode, which is more in line with reality and is beneficial to the operation and management of power grid companies.

Description

Technical Field

[0001] The present invention belongs to the field of distributed power supply management in a power distribution network, and in particular relates to a distributed photovoltaic power supply cluster division method considering a grid-connected management mode. Background Art

[0002] In order to effectively respond to the energy crisis and control environmental pollution, the development of distributed power sources based on renewable energy generation has become a global consensus. Due to the different grid-connected management methods of distributed power sources, multi-point decentralized access, intermittent and random characteristics, the complexity and control difficulty of the power grid have been greatly increased. With the continuous growth of distributed power generation, it has brought huge challenges to the safe and stable operation of the new power system. Distributed photovoltaic power sources connected to the grid in cluster control mode can effectively alleviate the volatility and uncertainty of a single distributed photovoltaic power source, and is the development direction of orderly and efficient grid connection of renewable power sources in the future. The existing distributed photovoltaic power cluster division method is mainly used for dispatching control, focusing on the electrical coupling between nodes, lacking the grid-connected management method of distributed photovoltaic power sources, and cannot provide an effective solution for the efficient operation and management of large-scale distributed photovoltaic power sources. Therefore, a distributed photovoltaic power cluster division method that takes into account both distributed photovoltaic power grid-connected management and network coupling is needed. Summary of the invention

[0003] The present invention discloses a distributed photovoltaic cluster division method considering the grid-connected management mode, comprising:

[0004] Step A: Collect information on line parameters, topology, node types and distributed photovoltaic power grid connection management methods of the distribution network;

[0005] Step B. Calculate and analyze the information collected in step A to form a network structure diagram of the cluster distribution network to be divided;

[0006] Step C: Classify distributed photovoltaic power supply nodes according to different grid-connected management methods of distributed photovoltaic power supplies;

[0007] Step D. Taking the maximum modularity of the distributed photovoltaic power cluster as the goal, a distributed photovoltaic power cluster partition model considering the grid-connected management method is established, and a cluster partition scheme is obtained by using a genetic algorithm.

[0008] Furthermore, the line parameters of the distribution network in step A include the network single-line diagram, the model, installation method, length and number of loops of the power cable; the topological structure of the distribution network is the network connection relationship between the load nodes and distributed photovoltaic power supply nodes in the distribution network, and the node types are load nodes and distributed photovoltaic power supply nodes; the distributed photovoltaic power supply grid-connected management methods include self-generation and self-use surplus power access mode, full access mode and complete self-generation and self-use mode.

[0009] Furthermore, in step B, the impedance and admittance of the branches between each node in the distribution network are calculated according to the model, installation method, length and number of loops of the power cable; based on the topological structure of each network, the node type and the grid-connected management method of the distributed photovoltaic power source, a network structure diagram of the distribution network is formed, and the type of each node and the grid-connected management method of the distributed photovoltaic power source must be indicated in the diagram.

[0010] Furthermore, in step C, classifying distributed photovoltaic power nodes according to their different grid-connected management methods is a prerequisite for dividing distributed photovoltaic power clusters. The principle of classification is that distributed photovoltaic power nodes with the same grid-connected management method can be divided into the same cluster or into different clusters; distributed photovoltaic power nodes with different grid-connected management methods cannot be divided into the same cluster.

[0011] Furthermore, the distributed photovoltaic power cluster partitioning model considering the grid-connected management mode established in step D is as follows:

[0012]

[0013] in,

[0014]

[0015]

[0016] e ij =1-d ij / max(d)

[0017] Z ij =Z ii +Z jj -2Z ij

[0018]

[0019] Where ρ is the modularity value of the network after the distributed photovoltaic power generation is divided into clusters; m is the sum of the edge weights of the network; k i represents the sum of all edge weights connected to node i; e ij is the weight of the edge connecting node i and node j, referred to as edge weight, d ij Represents the electrical distance between node i and node j; when node i and node j are directly connected, e ij =1, when not connected e ij =0; δ is the reachability matrix, when node i and node j are in the same cluster, δ(i,j)=1, otherwise δ(i,j)=0.

[0020] Furthermore, the electrical distance is given by the input impedance Z of the two-port network.ij Indicates that

[0021] d ij =Z ij

[0022] Among them, Z ii is the self-impedance of node i, Z ij is the mutual impedance between node i and node j, and the electrical distance is inversely proportional to the edge weight.

[0023] Furthermore, after the distributed photovoltaic power sources are divided into clusters, the modularity value ρ of the network is greater than 0.7.

[0024] Furthermore, an improved genetic algorithm encoding method is used in the model to encode chromosomes based on the adjacency matrix of the network. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 is a schematic diagram of a network structure of a specific example of the method of the present invention;

[0026] Figure 2 is a flow chart of the method of the present invention;

[0027] Figure 3 is the electrical distance from node 1 to the remaining nodes in the network in a specific example of the method of the present invention;

[0028] Figure 4 It is the genetic algorithm encoding method adopted by the method of the present invention;

[0029] Figure 5 It is a flow chart of the classification of distributed photovoltaic power sources considering the grid-connected management mode of the method of the present invention;

[0030] Figure 6 It is a genetic algorithm solution flow chart adopted by the method of the present invention;

[0031] Figure 7 is a diagram showing changes in modularity values ​​before and after the grid connection management method is considered in a specific example of the method of the present invention;

[0032] Figure 8 The specific example of the method of the present invention does not consider the cluster division result of the grid connection management mode;

[0033] Fig. 9 It is the cluster division result of the specific example of the method of the present invention considering the grid connection management mode; DETAILED DESCRIPTION

[0034] The present invention will be described in detail below in conjunction with the accompanying drawings.

[0035] Figure 1The network structure diagram of a specific example of the method of the present invention is shown. The network is an IEEE 33-node feeder system, with a total of 32 branches, node 0 being the root node of the system, and a total of 15 distributed photovoltaic power nodes connected in the system. The figure shows the location distribution of photovoltaic nodes in three grid-connected management modes, with the triangle representing the self-generation and self-use surplus power access mode, the diamond representing the self-generation and self-use mode, and the five-pointed star representing the full amount access mode.

[0036] Figure 2 The figure shows a flow chart of distributed photovoltaic power cluster division considering the grid-connected management mode, which includes the following steps:

[0037] Step A: Collect information such as line parameters, topology, node types and distributed photovoltaic power grid-connected management methods of the distribution network.

[0038] In the actual distribution network, cluster division is carried out, and the data collection needs to be connected with the power supply company where the distribution network is located, which can be collected on the automation and intelligent platform of the corresponding department. The line parameters of the distribution network include the network single line diagram, the model, installation method, length and number of loops of the power cable. The topological structure of the distribution network is the network connection relationship between each load node and distributed photovoltaic power supply node in the distribution network. The node type is load node and distributed photovoltaic power supply node. The distributed photovoltaic power supply grid-connected management mode includes self-generation and self-use surplus power access mode, full access mode and complete self-generation and self-use mode.

[0039] Step B: Perform calculation and analysis based on the information collected in step A, and organize and form a network structure diagram of the cluster distribution network to be divided.

[0040] Step B1: Calculate the impedance and admittance of the branches between nodes in the distribution network according to the model, installation method, length and number of loops of the power cable. The impedance of the cable is calculated by the following formula:

[0041]

[0042]

[0043] z=(r+x)l

[0044] In the formula, r is the unit resistance of the cable, ρ is the conductor resistivity, S is the nominal cross-sectional area of ​​the conductor, the resistivity of aluminum is 31.5Ω·mm2 / km, and the resistivity of copper is 18.8Ω·mm2 / km. x is the unit reactance of the cable, D m is the geometric mean distance, d is the wire radius, z is the total impedance of the branch, and l is the length of the branch.

[0045] Table 1 Calculation table of impedance values ​​of each branch in the network

[0046]

[0047]

[0048] Step B2: Based on the topological structure of each network, the node type and the grid-connected management method of the distributed photovoltaic power source, a network structure diagram of the distribution network is formed. The type of each node and the grid-connected management method of the distributed photovoltaic power source must be indicated in the diagram, such as Figure 1 shown.

[0049] Step C: Classify distributed photovoltaic power sources according to different grid-connected management methods.

[0050] Distributed photovoltaic power nodes are classified according to their different grid-connected management methods. This classification serves as a prerequisite for the division of distributed photovoltaic power clusters. The classification principle is that distributed photovoltaic power nodes with the same grid-connected management method can be divided into the same cluster or into different clusters, but distributed photovoltaic power nodes with different grid-connected management methods cannot be divided into the same cluster.

[0051] Step D. Taking the maximum modularity of the distributed photovoltaic power cluster as the goal, a distributed photovoltaic power cluster partition model considering the grid-connected management mode is established, and a cluster partition scheme is obtained by using a genetic algorithm.

[0052] Step D1. Based on the network node admittance matrix, calculate the electrical distance between each node.

[0053] The input impedance Z of the two-port network is used to represent the electrical distance d between nodes. ij (i≠j) can be expressed as the difference between the voltage and current ratios between the two nodes when a certain current is input to nodes i and j. The node impedance matrix Z is calculated from the node admittance matrix Y, as shown in the following formula:

[0054] Z=Y -1

[0055] Z ij =Z ii +Z jj -2Z ij

[0056] In the formula, Z ii is the self-impedance of node i, Z ij is the mutual impedance between node i and node j. Figure 3 In a specific example network of the method of the present invention, the electrical distances from node 1 to the remaining nodes are shown, and the electrical distance from node 1 to node 17 is the farthest, while the electrical distance to node 18 is the shortest.

[0057] Step D2: Perform individual coding based on the network adjacency matrix.

[0058] The present invention adopts genetic algorithm to solve the problem, and the key is chromosome coding. Because the problem model of clustering the network is different from the conventional problem model, the topological structure of the network also needs to be reflected in the coding, so an improved genetic algorithm coding method is adopted to encode the chromosome based on the adjacency matrix of the network. The network adjacency matrix A represents the connection relationship between the nodes in the network. When A ij = 0, indicating that there is no direct connection between node i and node j. ij =1, it means that node i and node j are directly connected. Considering the actual topological relationship of the network, when encoding and constructing the initial population, the genetic algorithm only needs to search for 1 in the adjacency matrix and randomly modify it to 0 or keep it unchanged, which means disconnection or connection respectively. Figure 4 After the search is completed, a new matrix is ​​formed. This matrix is ​​an encoded individual and also represents a cluster division scheme.

[0059] During the coding process, photovoltaic classifications with different grid-connected management methods are considered simultaneously. Figure 5 The flow chart of the classification of distributed photovoltaic power sources considering the grid-connected management mode of the method of the present invention is shown. Distributed photovoltaic nodes with the same grid-connected management mode can be in the same cluster or in different clusters, but distributed photovoltaic nodes with different grid-connected management modes cannot be classified in the same cluster. For example: the grid-connected management modes of photovoltaic nodes i and photovoltaic nodes j are different, then in the adjacency matrix A(i,j)=0, A(j,i)=0, and if node k is in the same cluster as node i, then A(i,k)=1, A(k,i)=1, and A(k,j)=0, A(j,k)=0, to ensure that distributed photovoltaics with different grid-connected management modes will not be classified into the same cluster.

[0060] Step D3. Establish a distributed photovoltaic power cluster partition model that takes into account the grid-connected management method, and use a genetic algorithm to solve and obtain a cluster partition solution.

[0061] The distributed photovoltaic power cluster division model considering the grid-connected management method is as follows:

[0062]

[0063] in

[0064]

[0065]

[0066] e ij =1-d ij / max(d)

[0067] d ij =Z ij

[0068] Z ij =Z ii +Z jj -2Z ij

[0069]

[0070] In the formula, ρ is the modularity value of the network after the distributed photovoltaic power generation is divided into clusters; m is the sum of the edge weights of the network; k i represents the sum of all edge weights connected to node i; e ij is the weight of the edge connecting node i and node j, referred to as edge weight, d ij Represents the electrical distance between node i and node j. The edge weight based on electrical distance is selected to calculate the modularity index. On the one hand, it measures the structural performance of the divided clusters, and on the other hand, it also ensures the electrical connection strength of the divided clusters. The larger the modularity value, the stronger the structural strength of the divided clusters. Generally, it is better to be greater than 0.7. The electrical distance is determined by the input impedance Z of the two-port network. ij Indicates that Z ii is the self-impedance of node i, Z ij is the mutual impedance between node i and node j. The electrical distance is inversely proportional to the edge weight. When node i and node j are directly connected, e ij =1, when not connected e ij =0; δ is the reachability matrix, when node i and node j are in the same cluster, δ(i,j)=1, otherwise δ(i,j)=0.

[0071] Figure 6 The flowchart of genetic algorithm solution adopted by the method of the present invention is shown. In a network with complex structure, the changes of the number of clusters and the combination of nodes in the cluster have an impact on the cluster performance, and it is difficult for the conventional partitioning algorithm to obtain the global optimal solution. Compared with the conventional algorithm, the genetic algorithm can give full play to the global search ability by reasonably setting parameters, and will not fall into the local optimal solution, ensuring that it gradually approaches the global optimal solution as the number of iterations increases. In addition, when the number of iterations reaches a certain value, the change range of fitness in the genetic algorithm will gradually slow down. In the case of not strictly requiring the global optimal solution, the slow value can be selected as the approximate optimal solution, which will greatly reduce the calculation time.

[0072] In order to improve the global search capability and convergence speed of the genetic algorithm, the idea of ​​adaptive genetic algorithm is adopted to make the crossover rate and mutation rate of the algorithm adaptively adjusted. Adjustment principle: If the individual fitness is less than the average fitness of the population, set a larger crossover rate and a smaller mutation rate; if the individual fitness is greater than or equal to the average fitness, the crossover rate and mutation rate can be set according to the current number of iterations. The crossover rate decreases with the increase of the number of iterations, and the mutation rate increases with the increase of the number of iterations. In the early stage of iteration, the overall fitness of the population is relatively low. Setting a larger crossover rate and a smaller mutation rate for individuals with fitness greater than or equal to the average fitness is conducive to retaining excellent individuals, thereby ensuring that the population inherits in an increasingly optimal direction. In the later stage of iteration, the population gradually approaches the optimal, and individuals gradually become similar and tend to be consistent. At this time, a larger crossover rate becomes meaningless, and a larger mutation rate is conducive to enabling the algorithm to jump out of the local optimal solution and obtain the global optimal solution. Therefore, a smaller crossover rate and a larger mutation rate should be set in the later stage of iteration. Adaptive crossover probability p c and mutation probability p m It is calculated by the following formula:

[0073]

[0074]

[0075] Where: p c , p m is the crossover rate and mutation rate; p cmax , p cmin is the maximum and minimum value of the crossover rate; p mmax , p mmin is the maximum and minimum value of the mutation rate; I max , I is the maximum number of iterations and the current number of iterations; f c is the larger fitness value of the two individuals in the crossover; f m is the fitness value of the individual undergoing mutation; f avg is the average fitness of the population.

[0076] Figure 7 The comparison of modularity values ​​before and after considering the grid-connected management method is shown. Figure 8 This is the cluster division result without considering the grid connection management method. Fig. 9 To consider the cluster division results of the grid-connected management method, Figure 8 and Fig. 9 The length of the branches between the nodes represents the electrical distance.

[0077] from Figure 7It can be seen that the maximum modularity value has been reached when the number of iterations reaches 150. The maximum modularity when the grid-connected management method is considered is slightly smaller than the maximum modularity when the grid-connected management method is not considered. This is because when the cluster division is performed without considering the grid-connected management method, the division result is shown in Figure 8 , there is a cluster containing two photovoltaic nodes with different grid-connected management modes, such as node 1 and node 3 are in the self-generation and self-use surplus power grid-connected mode, and node 23 is in the full grid-connected mode, but nodes 1, 3 and 23 are divided into the same cluster; when considering the grid-connected management mode, the division results are shown in Fig. 9 , distributed photovoltaic nodes with different grid-connected management methods will not be divided into the same cluster, which means that node 23, node 1 and node 3 cannot be divided into the same cluster, and the overall cluster is increased by 1, resulting in a decrease in the maximum modularity.

[0078] When the modularity index based on electrical distance is used for cluster division without considering the grid-connected management mode, the electrical connection between nodes in each cluster is close, while the electrical connection between clusters is loose, and the structural strength of the cluster is the strongest. At this time, the maximum modularity value is ρ max =0.77511, but in the division results, there are PV nodes with two different grid-connected management methods in the same cluster, such as PV nodes 1, 3 and PV node 23 in cluster 1, which have different grid-connected management methods, which is not conducive to the operation and management of the power grid company. When the grid-connected management method is considered and the modularity index based on electrical distance is used for cluster division, although the number of clusters obtained in the optimal division result increases by 1 compared with the case without considering the grid-connected management method, the structural strength of the cluster decreases slightly, and the maximum modularity value is still greater than 0.7, ρ max =0.76465, the modularity value decreased by 1.35%, but in the division results, all photovoltaic nodes in the clusters have the same grid-connected management method, and there is no situation where photovoltaic nodes with different grid-connected management methods are divided into the same cluster, which is very beneficial for the power grid company to operate and manage the distribution network.

[0079] The present invention is general in modeling, does not have any special application conditions, has no requirements on system parameters, does not require any inspection before application, has a wide range of applications, and can be easily extended to the field of distributed photovoltaic power cluster division.

[0080] The above embodiments are only preferred specific implementations of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed by the present invention should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A distributed photovoltaic power cluster division method considering the grid-connected management mode, comprising: Step A: Collect information on line parameters, topology, node types and distributed photovoltaic power grid connection management methods of the distribution network; Step B. Perform calculation and analysis based on the information collected in step A, and organize and form a network structure diagram of the cluster distribution network to be divided; Step C. Classifying distributed photovoltaic power supply nodes according to different grid-connected management modes of distributed photovoltaic power supplies; the principle of classification is that distributed photovoltaic power supply nodes with the same grid-connected management mode can be divided into the same cluster or different clusters; distributed photovoltaic power supply nodes with different grid-connected management modes cannot be divided into the same cluster; Step D. With the goal of maximizing the modularity of the distributed photovoltaic power cluster, a distributed photovoltaic power cluster partition model considering the grid-connected management mode is established, and a cluster partition scheme is obtained by using a genetic algorithm; the partition model is as follows: in, e ij =1-d ij / max(d) WITH ij =Z ii +Z jj -2Z ij Where ρ is the modularity value of the network after the distributed photovoltaic power generation is divided into clusters, and it is greater than 0.7; m is the sum of the edge weights of the network; k i represents the sum of all edge weights connected to node i; e ij is the weight of the edge connecting node i and node j, referred to as edge weight, d ij Represents the electrical distance between node i and node j; when node i and node j are directly connected, e ij =1, when not connected e ij =0; δ is the reachability matrix. When node i and node j are in the same cluster, δ(i, j) = 1, otherwise δ(i, j) = 0.

2. A distributed photovoltaic power cluster division method considering the grid-connected management mode according to claim 1, characterized in that: The line parameters of the distribution network in step A include the network single-line diagram, the model, installation method, length and number of loops of the power cable; the topological structure of the distribution network is the network connection relationship between the load nodes and distributed photovoltaic power supply nodes in the distribution network, and the node types are load nodes and distributed photovoltaic power supply nodes; the distributed photovoltaic power supply grid-connected management methods include self-generation and self-use surplus power access mode, full access mode and complete self-generation and self-use mode.

3. The method for dividing distributed photovoltaic power clusters considering the grid-connected management mode according to claim 2 is characterized in that: In step B, the impedance and admittance of the branches between each node in the distribution network are calculated according to the model, installation method, length and number of loops of the power cable; based on the topological structure of each network, the node type and the grid-connected management method of the distributed photovoltaic power source, a network structure diagram of the distribution network is formed, and the type of each node and the grid-connected management method of the distributed photovoltaic power source must be indicated in the diagram.

4. The method for dividing distributed photovoltaic power clusters considering the grid-connected management mode according to claim 1 is characterized in that: The electrical distance is determined by the input impedance Z of the two-port network ij Indicates that d ij =Z ij Among them, Z ii is the self-impedance of node i, Z ij is the mutual impedance between node i and node j, and the electrical distance is inversely proportional to the edge weight.

5. The method for dividing distributed photovoltaic power clusters considering the grid-connected management mode according to claim 1 is characterized in that: The model adopts an improved genetic algorithm encoding method and encodes chromosomes based on the adjacency matrix of the network.

Images