Dynamic cluster division method and device for distribution network substations based on coupling coefficient index

Abstract

The present invention relates to the technical field of distribution network cluster division, and discloses a method and device for dynamic cluster division of distribution network substations based on coupling coefficient indicators. First, a cluster division indicator based on the sensitivity correlation coefficient is determined, and the reactive voltage sensitivity correlation coefficient, the cluster's autocoupling coefficient, and the mutual coupling coefficient between clusters are calculated; then, a cluster division comprehensive indicator is determined based on the autocoupling coefficient and the mutual coupling coefficient, and the maximization of the cluster division comprehensive indicator is used as the objective function to determine the constraints of the number of clusters and the cluster's reactive compensation capacity, thereby constructing a distribution network cluster division model that considers the equal distribution of regional photovoltaic reactive compensation capacity, and using a genetic algorithm to solve it; finally, based on the dynamic changes in the operating conditions of the distribution network, a method of aggregating different cluster division results under multiple time sections is used to perform dynamic cluster division. The present invention realizes cluster division applicable to any operating conditions of the distribution network substation, and can reduce the amount of partition calculation without losing accuracy.

Description

Technical Field

[0001] The present invention relates to the technical field of distribution network cluster division, and in particular to a method and device for dynamically dividing distribution network substations into clusters based on a coupling coefficient index. Background Art

[0002] Using PV inverters for reactive power compensation and adjusting the voltage levels at the PV grid connection point and nearby nodes is a flexible and feasible voltage regulation method. Dividing the distribution network into multiple clusters and clearly defining the relationship between PV and clusters is a prerequisite for zoning PV control.

[0003] However, using PV inverters for reactive power compensation faces challenges in voltage regulation efficiency and the computational complexity of PV output decision-making. First, a single PV inverter's voltage regulation effect varies for different nodes. Based on the characteristics of the voltage sensitivity matrix, the voltage regulation effect of PV is related to the electrical distance between nodes. PV provides better voltage regulation for nodes that are electrically closer to it, while for nodes farther away, more reactive power compensation is required to achieve the same voltage amplitude increase. Second, in distributed voltage control, cluster controllers receive control commands from a cloud-based controller center to regulate voltage. They also receive and analyze PV and load data uploaded by cluster devices, issuing commands to control the PV output and voltage within the cluster, achieving autonomous operation in each cluster region. However, if the affiliation between devices and cluster controllers changes dynamically, the controller's decision-making computational complexity will increase significantly. In reality, the operating characteristics of the distribution network may be similar at different time intervals. Scenario aggregation can be used to unify PV-cluster affiliations across different time intervals, reducing computational complexity.

[0004] Current research on distribution network clustering primarily encompasses clustering metrics and clustering methods. Most current clustering metrics are coupled indices based on sensitivity matrices, which fully reflect the impact of changes in node active and reactive power on node voltage. These metrics have been widely used as indicators for exploring sensitive node clusters. Clustering methods primarily encompass clustering algorithms, optimization algorithms, and complex network community detection algorithms. K-means is one of the most common methods, characterized by simple logic and ease of implementation. However, it is sensitive to noise and outliers, resulting in potentially inaccurate results. Optimization algorithms can globally search for optimal solutions and are highly adaptable, making them well-suited for complex optimization problems. However, these algorithms can be computationally intensive and time-consuming. Complex network community detection algorithms are suitable for processing large-scale network data, but some community detection algorithms have resolution limitations and may not detect smaller communities.

[0005] In summary, the shortcomings of the prior art are as follows:

[0006] 1) Most existing cluster division indicators rely on network topology and line parameters. Inaccurate parameters will affect the calculation results.

[0007] 2) The existing cluster division method does not take into account the photovoltaic reactive power compensation capability within each cluster.

[0008] 3) Existing technologies rarely consider the impact of changes in distribution substation operating conditions on the clustering results of the distribution network. In particular, when using relatively high-accuracy optimization algorithms for real-time dynamic clustering of distribution networks, the amount of calculation will increase dramatically and the time consumption will be long.

[0009] 4) The current real-time cluster division method not only increases the amount of calculation, but also causes the mapping relationship between photovoltaics and clusters to change frequently, increasing the difficulty of photovoltaic control. Summary of the Invention

[0010] To address the above issues, the present invention aims to provide a method and device for dynamically clustering distribution network substations based on the coupling coefficient indicator. This method aggregates clustering results with similar structures in different time periods, thereby achieving clustering suitable for distribution network substations under any operating conditions, and reducing the amount of partitioning calculations without sacrificing accuracy. The technical solution is as follows:

[0011] A method for dynamic clustering of distribution network substations based on coupling coefficient indicators includes:

[0012] Determine the cluster division index based on the sensitivity correlation coefficient, calculate the reactive power and voltage sensitivity correlation coefficient, the cluster autocoupling coefficient, and the mutual coupling coefficient between clusters;

[0013] Determining a comprehensive index for cluster division based on the autocoupling coefficient and the mutual coupling coefficient, taking maximization of the comprehensive index as the objective function, determining constraints on the number of clusters and cluster reactive compensation capacity, constructing a distribution network cluster division model that considers equal distribution of regional photovoltaic reactive compensation capacity, and solving it using a genetic algorithm;

[0014] Based on the dynamic changes in the operating conditions of the distribution network, the distribution network cluster partitioning model is used to perform dynamic cluster partitioning by adopting an aggregation method of different cluster partitioning results under multiple time sections.

[0015] Furthermore, the determining of the cluster division index based on the sensitivity correlation coefficient, and the calculation of the reactive voltage sensitivity correlation coefficient, the cluster autocoupling coefficient, and the mutual coupling coefficient between clusters, include:

[0016] Define the reactive voltage sensitivity correlation coefficient r between nodes i and j in the distribution network at time t: VQ.ij (t) is expressed as follows:

[0017]

[0018] Where n is the total number of distribution network nodes, and They represent the reactive voltage sensitivity vectors of all nodes in the distribution network to node i and node j at time t respectively; and They represent the reactive voltage sensitivity of node k to node i and node j at time t, k = 1, 2, …, n; and Reactive power and voltage sensitivity vectors are and The standard deviation of the sample in and Reactive power and voltage sensitivity vectors are and The mean of the sample in ;

[0019] Reactive voltage sensitivity vector and The expression is:

[0020]

[0021] in, and are the reactive voltage sensitivities of node k to node i and node j respectively;

[0022] Define R VQ.X (t) is the self-coupling coefficient of cluster X at time t, R VQ.XY (t) is the mutual coupling coefficient between cluster X and cluster Y at time t;

[0023] Autocoupling coefficient R VQ.X (t) and mutual coupling coefficient R VQ.XY The expression of (t) is:

[0024]

[0025]

[0026]

[0027] Among them, n x and n y are the number of nodes in cluster X and cluster Y respectively; N X and N Y are the sets of nodes in cluster X and cluster Y respectively; is an empty set;

[0028] Let the number of clusters be n c , then the self-coupling coefficient within the cluster and the mutual coupling coefficient between clusters are averaged:

[0029]

[0030]

[0031] Among them, R X,avg (t) is the average value of the self-coupling coefficient within the cluster; R XY,avg (t) are the average mutual coupling coefficients between clusters.

[0032] Furthermore, a distribution network cluster partitioning model considering equal distribution of regional photovoltaic reactive compensation capacity is constructed and solved as follows:

[0033] Determine the objective function:

[0034] Based on the indicators of self-coupling coefficient and mutual coupling coefficient, a comprehensive index f for cluster division is proposed R (t) as the fitness function, and f R (t) Maximum as the optimization goal:

[0035] max[f R (t)]=max[(R X，avg (t)-R XY，avg (t))] (8)

[0036] Set up constraints:

[0037] a) Constraints on the number of clusters:

[0038] Under each number of clusters, the cluster partitioning model is solved, and the number of clusters is limited as follows:

[0039] n c，min ≤n c ≤n c，max (9)

[0040] Among them, n c,min is the minimum number of clusters, n c,max is the maximum number of clusters;

[0041] b) Cluster reactive power compensation capacity constraints:

[0042] To ensure that each cluster has a certain photovoltaic reactive power compensation capacity and that the overall reactive power compensation is reasonably distributed among the clusters, the following constraints are set:

[0043]

[0044] Among them, Q load (t) and Q PV (t) are the maximum reactive power of the load at time t and the maximum capacity of the distribution network photovoltaic that can participate in reactive power compensation at time t, QX,load (t) is the maximum reactive power of cluster X at time t, Q X,PV (t) is the maximum capacity of photovoltaic power plant in cluster X that can participate in reactive power compensation at time t; Q X,PV The expression of (t) is:

[0045]

[0046] in, is the capacity of the h-th PV inverter in cluster X, N X.PV is the set of photovoltaic nodes in cluster X, is the maximum tracking power of the h-th PV in cluster X at time t;

[0047] Solve:

[0048] The node number is used as the decision variable and the comprehensive index of cluster division is used as the fitness function. The genetic algorithm is used to solve the problem: first, the population is initialized, and the individual with the largest fitness function in the population is selected as the optimal individual in the first iteration; then the initial population is subjected to selection, crossover, mutation and inversion operations to update the population information, and the optimal individual in the second generation is calculated, and so on, until the termination condition is met.

[0049] Furthermore, the dynamic clustering method using the aggregation method of different clustering results under multiple time sections is specifically as follows:

[0050] Perform power flow calculation based on distribution network operating parameters to obtain the voltage sensitivity matrix within a set time period, and perform initial clustering of the distribution network using the distribution network clustering model to obtain the initial clustering results for the first time section.

[0051] According to the cluster division result of the first time section, the corresponding comprehensive index value is obtained, which is recorded as f R (t - ); Replace the sensitivity matrix of the previous time section with the voltage sensitivity matrix of the next time section, and recalculate the comprehensive index under the cluster division result of the previous time section based on the revised voltage sensitivity matrix, which is recorded as f R (t + );

[0052] Calculate f R (t + ) and f R (t - ), if the difference satisfies the following formula, the cluster division result at the later moment is the same as the cluster division result at the previous moment, otherwise the cluster division result at the current moment shall prevail;

[0053] |f R (t +)-f R (t - )|≤f T (12)

[0054] Among them, f T The threshold representing the difference between the objective function values ​​at the previous and next moments.

[0055] A device for dynamically dividing distribution network substations into clusters based on a coupling coefficient index, comprising:

[0056] Cluster division index determination module: determines the cluster division index based on the sensitivity correlation coefficient, calculates the reactive voltage sensitivity correlation coefficient, the cluster autocoupling coefficient and the mutual coupling coefficient between clusters;

[0057] Distribution network cluster partitioning model construction module: Determines the cluster partitioning comprehensive index based on the autocoupling coefficient and mutual coupling coefficient, takes maximizing the cluster partitioning comprehensive index as the objective function, determines the constraints on the number of clusters and cluster reactive compensation capacity, and thus constructs a distribution network cluster partitioning model that considers the equal distribution of regional photovoltaic reactive compensation capacity, and uses a genetic algorithm to solve it;

[0058] Mutual coupling coefficient between clusters: Based on the dynamic changes in the operating conditions of the distribution network, dynamic cluster division is performed by aggregating the results of different cluster divisions under multiple time sections.

[0059] Furthermore, in the cluster division index determination module, the reactive voltage sensitivity correlation coefficient r between node i and node j in the distribution network at time t is first defined. VQ.ij (t) is expressed as follows:

[0060]

[0061] Where n is the total number of distribution network nodes, and They represent the reactive voltage sensitivity vectors of all nodes in the distribution network to node i and node j at time t respectively; and They represent the reactive voltage sensitivity of node k to node i and node j at time t, k = 1, 2, …, n; and Reactive power and voltage sensitivity vectors are and The standard deviation of the sample in and Reactive power and voltage sensitivity vectors are and The mean of the sample in ;

[0062] Reactive voltage sensitivity vector and The expression is:

[0063]

[0064] in, and are the reactive voltage sensitivities of node k to node i and node j respectively;

[0065] Then define R VQ.X (t) is the self-coupling coefficient of cluster X at time t, R VQ.XY (t) is the mutual coupling coefficient between cluster X and cluster Y at time t;

[0066] Autocoupling coefficient R VQ.X (t) and mutual coupling coefficient R VQ.XY The expression of (t) is:

[0067]

[0068]

[0069]

[0070] Among them, n x and n y are the number of nodes in cluster X and cluster Y respectively; N X and N Y are the sets of nodes in cluster X and cluster Y respectively; is an empty set;

[0071] Let the number of clusters be n c , then the self-coupling coefficient within the cluster and the mutual coupling coefficient between clusters are averaged:

[0072]

[0073]

[0074] Among them, R X,avg (t) is the average value of the self-coupling coefficient within the cluster; R XY,avg (t) are the average mutual coupling coefficients between clusters.

[0075] Furthermore, the objective function determined in the distribution network cluster partitioning model construction module is:

[0076] Based on the indicators of self-coupling coefficient and mutual coupling coefficient, a comprehensive index f for cluster division is proposed R (t) as the fitness function, and f R (t) Maximum as the optimization goal:

[0077] max[f R(t)]=max[(R x,avg (t)-R XY,avg (t))] (20)

[0078] The constraints set include:

[0079] a) Cluster number constraints

[0080] Under each number of clusters, the cluster partitioning model is solved, and the number of clusters is limited as follows:

[0081] n c，min ≤n c ≤n c，max (twenty one)

[0082] Among them, n c,min is the minimum number of clusters, n c,max is the maximum number of clusters;

[0083] b) Cluster reactive power compensation capacity constraints

[0084] To ensure that each cluster has a certain photovoltaic reactive power compensation capacity and that the overall reactive power compensation is reasonably distributed among the clusters, the following constraints are set:

[0085]

[0086] Among them, Q load (t) and Q PV (t) are the maximum reactive power of the load at time t and the maximum capacity of the distribution network photovoltaic that can participate in reactive power compensation at time t, Q X,load (t) is the maximum reactive power of cluster X at time t, Q X,PV (t) is the maximum capacity of photovoltaic power plant in cluster X that can participate in reactive power compensation at time t; Q X,PV The expression of (t) is:

[0087]

[0088] in, is the capacity of the h-th PV inverter in cluster X, N X.PV is the set of photovoltaic nodes in cluster X, is the maximum tracking power of the h-th PV in cluster X at time t;

[0089] The solution process is as follows: using the node number as the decision variable and the comprehensive index of cluster division as the fitness function, the genetic algorithm is used for solution: first, the population is initialized, and the individual corresponding to the largest fitness function in the population is selected as the optimal individual in the first iteration; then the initial population is subjected to selection, crossover, mutation and inversion operations to update the population information, and the optimal individual in the second generation is calculated, and so on, until the termination condition is met.

[0090] Furthermore, the dynamic clustering module first performs power flow calculation based on the distribution network operating parameters to obtain the voltage sensitivity matrix within a set time period, and performs initial clustering on the distribution network using the distribution network clustering model to obtain the initial clustering result for the first time section;

[0091] Then, according to the cluster division result of the first time section, the corresponding comprehensive index value is obtained, which is recorded as f R (t - ); Replace the sensitivity matrix of the previous time section with the voltage sensitivity matrix of the next time section, and recalculate the comprehensive index under the cluster division result of the previous time section based on the revised voltage sensitivity matrix, which is recorded as f R (t + );

[0092] Finally calculate f R (t + ) and f R (t - ), if the difference satisfies the following formula, the cluster division result at the later moment is the same as the cluster division result at the previous moment, otherwise the cluster division result at the current moment shall prevail;

[0093] |f R (t + )-f R (t - )|≤f T (twenty four)

[0094] Among them, f T The threshold representing the difference between the objective function values ​​at the previous and next moments.

[0095] A storage medium stores a computer program, wherein the computer program is configured to execute the above-mentioned method for dynamically dividing distribution network substations into clusters when running.

[0096] An electronic device includes a processor, which is used to process the above-mentioned method for dynamically dividing distribution network areas into clusters.

[0097] The beneficial effects of the present invention are:

[0098] 1) To address the problem of inaccurate parameters in the existing technology, the indicator proposed in this invention is based on known sensitivity data. From the perspective of the reactive power-voltage correlation of the nodes, a coupling coefficient indicator is proposed for cluster division and used as the optimization target. It highly couples the reactive power-voltage relationship of the nodes within the cluster and accurately clusters the photovoltaic and its sensitive response nodes.

[0099] 2) The present invention evenly distributes the total photovoltaic reactive compensation capacity to each cluster, achieving balanced reactive compensation capabilities of each cluster.

[0100] 3) The multi-time section dynamic clustering result aggregation method proposed in the present invention realizes real-time dynamic clustering applicable to any operating conditions in the distribution station area, and can reduce the partitioning calculation amount without losing accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0101] Figure 1 This is a flow chart of the method for dynamically dividing distribution network substations into clusters based on the coupling coefficient index of the present invention. DETAILED DESCRIPTION

[0102] The present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0103] In order to ensure the accuracy of cluster division, the present invention adopts the genetic algorithm in the optimization algorithm for cluster division. In order to overcome the problem of large amount of calculation, the present invention proposes a cluster result aggregation method for different time periods, thereby reducing the amount of calculation for real-time partitioning. The dynamic cluster division of the distribution network of the present invention takes into account the changes in photovoltaic and load operating conditions. The process is as follows: Figure 1 The specific steps are as follows:

[0104] Step 1: Clustering index based on sensitivity correlation coefficient

[0105] The reactive voltage sensitivity correlation coefficient between node i and node j at time t is defined as follows:

[0106]

[0107] Where n is the total number of distribution network nodes, and They represent the reactive voltage sensitivity vectors of all nodes in the distribution network to node i and node j at time t respectively; and They represent the reactive voltage sensitivity of node k to node i and node j at time t, k = 1, 2, …, n; and Reactive power and voltage sensitivity vectors are and The standard deviation of the sample in and Reactive power and voltage sensitivity vectors are and The mean of the sample in .

[0108] Reactive voltage sensitivity vector and The expression is:

[0109]

[0110] in, and are the reactive voltage sensitivities of node k to node i and node j respectively.

[0111] Define R VQ.X (t) is the self-coupling coefficient of cluster X at time t, R VQ.XY (t) is the mutual coupling coefficient between cluster X and cluster Y at time t. VQ.X (t) and R VQ.XY The expression of (t) is:

[0112]

[0113]

[0114]

[0115] Among them, n x and n y are the number of nodes in cluster X and cluster Y respectively; N X and N Y are the sets of nodes in cluster X and cluster Y respectively; Is an empty set.

[0116] R VQ.X The larger (t) is, the higher the reactive power-voltage coupling of nodes in cluster X is, and the reactive power compensation characteristics of nodes in the cluster are strongly correlated. VQ.XY The smaller (t) is, the lower the reactive-voltage coupling between nodes is, and the reactive compensation characteristics between clusters are weakly correlated. X and N Y are the node sets in cluster X and cluster Y respectively, n x and n y are the number of nodes in cluster X and cluster Y respectively.

[0117] If the number of clusters is n c , then the self-coupling coefficient within the cluster and the mutual coupling coefficient between clusters are averaged:

[0118]

[0119]

[0120] Among them, R X,avg (t) is the average value of the self-coupling coefficient within the cluster; R XY,avg (t) are the average mutual coupling coefficients between clusters.

[0121] Step 2: Distribution network cluster division model considering equal distribution of regional photovoltaic reactive compensation capacity

[0122] The coupling coefficient reflects the degree of coupling between the reactive power and voltage relationships of each cluster node and is a key indicator for determining the effective voltage regulation area for photovoltaic power generation. A distribution network cluster partitioning model based on the coupling coefficient ensures the rational allocation of reactive power compensation capacity within each cluster and limits the number of clusters in the model solution, reducing the computational effort.

[0123] (1) Objective function

[0124] Based on the autocoupling coefficient index and the mutual coupling coefficient index, a cluster division comprehensive index f is proposed. R (t) as the fitness function, and f R (t) Maximum as the optimization goal:

[0125] max[f R (t)]=max[(R X，avg (t)-R XY，avg (t))] (8)

[0126] Comprehensive index f R A larger (t) indicates a stronger reactive-voltage coupling within the cluster and a lower voltage-reactive coupling between nodes in the cluster. PV reactive compensation has the most significant effect on improving node voltage within the PV cluster and has a smaller overall impact on node voltages in other clusters.

[0127] (2) Constraints

[0128] a) Cluster number constraints

[0129] The cluster partitioning model must be solved for each number of clusters. To reduce the computational complexity, the number of clusters needs to be limited. The number of clusters is set manually, and in this embodiment is set to [3, 8].

[0130] n c，min ≤n c ≤n c，max (9)

[0131] Among them, n c,min is the minimum number of clusters, n c,max is the maximum number of clusters.

[0132] b) Cluster reactive power compensation capacity constraints

[0133] To ensure that each cluster has a certain photovoltaic reactive power compensation capacity and that the overall reactive power compensation is reasonably distributed among the clusters, the following constraints are set:

[0134]

[0135] Among them, Q load (t) and Q PV (t) are the maximum reactive power of the load at time t and the maximum capacity of the distribution network photovoltaic that can participate in reactive power compensation at time t, Q X,load (t) is the maximum reactive power of cluster X at time t, Q X,PV (t) is the maximum capacity of photovoltaic power plant in cluster X that can participate in reactive power compensation at time t. X,PV The expression of (t) is:

[0136]

[0137] in, is the capacity of the h-th PV inverter in cluster X, N X.PV is the set of photovoltaic nodes in cluster X, is the maximum tracking power of the h-th PV cell in cluster X at time t.

[0138] (3) Solution method

[0139] The cluster partitioning mathematical model proposed in the present invention is essentially a nonlinear optimization problem, which can be solved by a nonlinear optimization algorithm. Genetic algorithm is a heuristic optimization algorithm that searches the solution space of the problem by simulating natural selection and genetic mechanisms to find the optimal solution or a solution close to the optimal solution. Therefore, the present invention uses the node number as the decision variable and the comprehensive index of cluster partitioning as the fitness function, and adopts a genetic algorithm to solve it. First, the population is initialized, and the individual with the largest fitness function in the population is selected as the optimal individual in the first iteration; then the initial population is subjected to selection, crossover, mutation and inversion operations to update the population information, and the optimal individual in the second generation is calculated, and so on, until the termination condition is met. Step 3: Aggregation of cluster partitioning results of multi-time section distribution networks

[0140] Performing dynamic clustering on a minute or hourly time scale not only increases the computational complexity of clustering, but also frequently changes the affiliation between photovoltaic inverters and clusters, increasing the difficulty of control. In fact, the clustering results within a certain time scale may be similar. Therefore, the present invention proposes an aggregation method for different clustering results under multiple time sections, which not only simplifies the clustering results under multiple time sections, but also ensures the rationality of the clustering results facing the dynamic changes in the operating conditions of the distribution network. The aggregation process is as follows:

[0141] Step 3.1: First, perform power flow calculation based on the distribution network operating parameters to obtain the 24-hour voltage sensitivity matrix. Second, perform initial clustering on the distribution network to obtain the initial clustering results for the first time section.

[0142] Step 3.2: Obtain the corresponding comprehensive index value based on the cluster division result of the first time section, denoted as f R (t - ). The voltage sensitivity matrix of the next time section is replaced by the sensitivity matrix of the previous time section, and the comprehensive index under the cluster division result of the previous time section is recalculated based on the revised voltage sensitivity matrix, which is recorded as f R (t + ).

[0143] Step 3.3: Calculate f R (t + ) and f R (t - ), if the difference satisfies formula (12), the cluster division result at the later moment is the same as the cluster division result at the previous moment, otherwise the cluster division result at the current moment shall prevail.

[0144] |f R (t + )-f R (t - )|≤0.1 (12)

[0145] Among them, 0.1 represents the threshold value of the difference between the objective function values ​​before and after, and the value is set manually according to the actual situation.

[0146] In summary, the present invention proposes a method for clustering distribution network substations based on the coupling coefficient indicator, and aggregates the clustering results with similar cluster structures in different time periods, thereby achieving clustering applicable to any operating conditions of the distribution network substation, and can reduce the amount of partitioning calculations without losing accuracy. In addition, through capacity constraints, the total photovoltaic compensation capacity is evenly distributed to each cluster, ensuring that each cluster has sufficient reactive compensation capacity reserves. The present invention can be used in the software development of smart terminals for distribution network substations, and the distribution network substation can be partitioned in real time through smart terminal devices.

[0147] It will be understood by those skilled in the art that the embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of the present invention may be implemented in various computer languages, for example, the object-oriented programming language Java and the interpreted scripting language JavaScript.

[0148] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0149] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0150] These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are executed on the computer or other programmable device to produce a computer.

Claims

1. A method for dynamic cluster division of distribution network substations based on coupling coefficient index, characterized in that: include: Determine the cluster division index based on the sensitivity correlation coefficient, calculate the reactive power and voltage sensitivity correlation coefficient, the cluster autocoupling coefficient, and the mutual coupling coefficient between clusters; Determining a comprehensive cluster partitioning index based on the autocoupling coefficient and the mutual coupling coefficient, and taking maximization of the comprehensive cluster partitioning index as the objective function, determining constraints on the number of clusters and cluster reactive compensation capacity, thereby constructing a distribution network cluster partitioning model that considers equal distribution of regional photovoltaic reactive compensation capacity, and solving it using a genetic algorithm; Based on the dynamic changes of the distribution network operating conditions, dynamic clustering is performed by using the distribution network clustering model and adopting an aggregation method of different clustering results under multiple time sections; The determining of the cluster division index based on the sensitivity correlation coefficient, and the calculating of the reactive voltage sensitivity correlation coefficient, the cluster autocoupling coefficient, and the mutual coupling coefficient between clusters, include: The reactive voltage sensitivity correlation coefficient () between nodes i and j in the distribution network at time t is defined as follows: Where n is the total number of distribution network nodes, and They represent the reactive voltage sensitivity vectors of all nodes in the distribution network to node i and node j at time t respectively; and They represent the reactive voltage sensitivity of node k to node i and node j at time t, k = 1, 2, …, n; and Reactive power and voltage sensitivity vectors are and The standard deviation of the sample in and Reactive power and voltage sensitivity vectors are and The mean of the sample in ; Reactive voltage sensitivity vector and The expression is: in, and are the reactive voltage sensitivities of node k to node i and node j respectively; Define R VQ.X (t) is the self-coupling coefficient of cluster X at time t, R VQ.XY (t) is the mutual coupling coefficient between cluster X and cluster Y at time t; Autocoupling coefficient R VQ.X (t) and mutual coupling coefficient R VQ.XY The expression of (t) is: Among them, n x and n y are the number of nodes in cluster X and cluster Y respectively; N X and N Y are the sets of nodes in cluster X and cluster Y respectively; is an empty set; Let the number of clusters be n c , then the self-coupling coefficient within the cluster and the mutual coupling coefficient between clusters are averaged: Among them, R X,avg (t) is the average value of the self-coupling coefficient within the cluster; R XY,avg (t) are the average mutual coupling coefficients between clusters.

2. The method for dynamic cluster division of distribution network substations based on coupling coefficient index according to claim 1, characterized in that: The construction and solution of the distribution network cluster partition model considering the equal distribution of regional photovoltaic reactive compensation capacity are as follows: Determine the objective function: Based on the indicators of self-coupling coefficient and mutual coupling coefficient, a comprehensive index f for cluster division is proposed R (t) as the fitness function, and f R (t) Maximum as the optimization goal: max[f R (t)]=max[(R X,avg (t)-R XY,avg (t))] (8) Set up constraints: a) Constraints on the number of clusters: Under each number of clusters, the cluster partitioning model is solved, and the number of clusters is limited as follows: n c,min ≤n c ≤n c,max (9) Among them, n c,min is the minimum number of clusters, n c,max is the maximum number of clusters; b) Cluster reactive power compensation capacity constraints: To ensure that each cluster has a certain photovoltaic reactive power compensation capacity and that the overall reactive power compensation is reasonably distributed among the clusters, the following constraints are set: Among them, Q load (t) and Q PV (t) are the maximum reactive power of the load at time t and the maximum capacity of the distribution network photovoltaic that can participate in reactive power compensation at time t, Q X,load (t) is the maximum reactive power of cluster X at time t, Q X,PV (t) is the maximum capacity of photovoltaic power plant in cluster X that can participate in reactive power compensation at time t; Q X,PV The expression of (t) is: in, is the capacity of the h-th PV inverter in cluster X, N X.PV is the set of photovoltaic nodes in cluster X, is the maximum tracking power of the h-th PV in cluster X at time t; Solve: The node number is used as the decision variable and the comprehensive index of cluster division is used as the fitness function. The genetic algorithm is used to solve the problem: first, the population is initialized, and the individual with the largest fitness function in the population is selected as the optimal individual in the first iteration; then the initial population is subjected to selection, crossover, mutation and inversion operations to update the population information, and the optimal individual in the second generation is calculated, and so on, until the termination condition is met.

3. The method for dynamic cluster division of distribution network substations based on coupling coefficient index according to claim 2, characterized in that: The dynamic clustering method using the aggregation method of different clustering results under multiple time sections is specifically as follows: Based on the distribution network operating parameters, the power flow calculation is performed to obtain the voltage sensitivity matrix within the set time period, and the distribution network is initially clustered using the distribution network clustering model to obtain the initial clustering result of the first time section; the corresponding comprehensive index value is obtained based on the clustering result of the first time section, which is recorded as f R (t - ); Replace the sensitivity matrix of the previous time section with the voltage sensitivity matrix of the next time section, and recalculate the comprehensive index under the cluster division result of the previous time section based on the revised voltage sensitivity matrix, which is recorded as f R (t + ); Calculate f R (t + ) and f R (t - ), if the difference satisfies the following formula, the cluster division result at the later moment is the same as the cluster division result at the previous moment, otherwise the cluster division result at the current moment shall prevail; |f R (t + )-f R (t - )|≤f T (12) Among them, f T The threshold representing the difference between the objective function values ​​at the previous and next moments.

4. A device for dynamic cluster division of distribution network substations based on coupling coefficient index, characterized in that: include: Cluster division index determination module: determines the cluster division index based on the sensitivity correlation coefficient, calculates the reactive voltage sensitivity correlation coefficient, the cluster autocoupling coefficient and the mutual coupling coefficient between clusters; Distribution network cluster partitioning model construction module: Determines the cluster partitioning comprehensive index based on the autocoupling coefficient and mutual coupling coefficient, takes maximizing the cluster partitioning comprehensive index as the objective function, determines the constraints on the number of clusters and cluster reactive compensation capacity, and thus constructs a distribution network cluster partitioning model that considers the equal distribution of regional photovoltaic reactive compensation capacity, and uses a genetic algorithm to solve it; Mutual coupling coefficient between clusters: Based on the dynamic changes in the operating conditions of the distribution network, dynamic cluster division is performed through the distribution network cluster division model by aggregating different cluster division results under multiple time sections; In the cluster division index determination module, the reactive voltage sensitivity correlation coefficient () between nodes i and j in the distribution network at time t is first defined as follows: Where n is the total number of distribution network nodes, and They represent the reactive voltage sensitivity vectors of all nodes in the distribution network to node i and node j at time t respectively; and They represent the reactive voltage sensitivity of node k to node i and node j at time t, k = 1, 2, …, n; and Reactive power and voltage sensitivity vectors are and The standard deviation of the sample in and Reactive power and voltage sensitivity vectors are and The mean of the sample in ; Reactive voltage sensitivity vector and The expression is: in, and are the reactive voltage sensitivities of node k to node i and node j respectively; Then define R VQ.X (t) is the self-coupling coefficient of cluster X at time t, R VQ.XY (t) is the mutual coupling coefficient between cluster X and cluster Y at time t; Autocoupling coefficient R VQ.X (t) and mutual coupling coefficient R VQ.XY The expression of (t) is: Among them, n x and n y are the number of nodes in cluster X and cluster Y respectively; N X and N Y are the sets of nodes in cluster X and cluster Y respectively; is an empty set; Let the number of clusters be n c , then the self-coupling coefficient within the cluster and the mutual coupling coefficient between clusters are averaged: Among them, R X,avg (t) is the average value of the self-coupling coefficient within the cluster; R XY,avg (t) are the average mutual coupling coefficients between clusters.

5. The device for dynamic cluster division of distribution network substations based on coupling coefficient index according to claim 4, characterized in that: The objective function determined in the distribution network cluster partitioning model construction module is: Based on the indicators of self-coupling coefficient and mutual coupling coefficient, a comprehensive index f for cluster division is proposed R (t) as the fitness function, and f R (t) Maximum as the optimization goal: max[f R (t)]=max[(R X,avg (t)-R XY,avg (t))] (20) The constraints set include: a) Cluster number constraints Under each number of clusters, the cluster partitioning model is solved, and the number of clusters is limited as follows: n c,min ≤n c ≤n c,max (21) Among them, n c,min is the minimum number of clusters, n c,max is the maximum number of clusters; b) Cluster reactive power compensation capacity constraints To ensure that each cluster has a certain photovoltaic reactive power compensation capacity and that the overall reactive power compensation is reasonably distributed among the clusters, the following constraints are set: Among them, Q load (t) and Q PV (t) are the maximum reactive power of the load at time t and the maximum capacity of the distribution network photovoltaic that can participate in reactive power compensation at time t, Q X,load (t) is the maximum reactive power of cluster X at time t, Q X,PV (t) is the maximum capacity of photovoltaic power plant in cluster X that can participate in reactive power compensation at time t; Q X,PV The expression of (t) is: in, is the capacity of the h-th PV inverter in cluster X, N X.PV is the set of photovoltaic nodes in cluster X, is the maximum tracking power of the h-th PV in cluster X at time t; The solution process is as follows: using the node number as the decision variable and the comprehensive index of cluster division as the fitness function, the genetic algorithm is used for solution: first, the population is initialized, and the individual corresponding to the largest fitness function in the population is selected as the optimal individual in the first iteration; then the initial population is subjected to selection, crossover, mutation and inversion operations to update the population information, and the optimal individual in the second generation is calculated, and so on, until the termination condition is met.

6. The device for dynamic cluster division of distribution network substations based on coupling coefficient index according to claim 5, characterized in that: The dynamic clustering module first calculates the power flow based on the distribution network operating parameters, obtains the voltage sensitivity matrix within the set time period, and performs initial clustering of the distribution network using the distribution network clustering model to obtain the initial clustering results for the first time section. Then, according to the cluster division result of the first time section, the corresponding comprehensive index value is obtained, which is recorded as f R (t - ); Replace the sensitivity matrix of the previous time section with the voltage sensitivity matrix of the next time section, and recalculate the comprehensive index under the cluster division result of the previous time section based on the revised voltage sensitivity matrix, which is recorded as f R (t + ); Finally calculate f R (t + ) and f R (t - ), if the difference satisfies the following formula, the cluster division result at the later moment is the same as the cluster division result at the previous moment, otherwise the cluster division result at the current moment shall prevail; |f R (t + )-f R (t - )|≤f T (24) Among them, f T The threshold representing the difference between the objective function values ​​at the previous and next moments.

7. A storage medium, characterized in that: The storage medium stores a computer program, wherein the computer program is configured to execute the method according to any one of claims 1 to 3 when executed.

8. An electronic device, characterized in that: The method comprises a processor for processing the method according to any one of claims 1 to 3.

Images