Whole county photovoltaic cluster voltage control method, system and device based on improved differential evolution algorithm and medium

Abstract

The invention relates to the technical field of power distribution network control, and discloses a whole county photovoltaic cluster voltage control method, system and device based on an improved differential evolution algorithm, and a medium, and the method comprises the steps: calculating electrical data between nodes in a power distribution network, and carrying out the preprocessing of the electrical data; performing cluster division through modularity optimization, and setting a target function of cluster division; and constructing a cluster voltage control model by improving a differential evolution algorithm in each cluster and taking photovoltaic inverter residual capacity reactive compensation and distributed photovoltaic active power reduction as control variables, and performing voltage optimization control of the power distribution network. According to the method, the influence of active power and reactive power is comprehensively considered, adaptive clustering of the power distribution network is realized, and the calculation speed is higher; the differential evolution algorithm is improved, so that better optimization precision and calculation stability are achieved; on the basis of cluster voltage control of the virtual balance nodes, the problem that calculation errors are too large due to the fact that global coordination is ignored by independent control is avoided to a certain extent.

Description

Technical Field

[0001] This invention relates to the field of power distribution network control technology, and in particular to a method, system, equipment and medium for voltage control of a county-wide photovoltaic cluster based on an improved differential evolution algorithm. Background Technology

[0002] Photovoltaic power generation boasts advantages such as cleanliness and low carbon emissions, making it a significant development potential in low-carbon power grid construction. Distributed photovoltaic (PV) power generation follows the principle of "local generation, local grid connection, and local utilization," not only meeting the growing electricity demand of residents but also generating additional revenue for installed users through surplus grid-connected electricity. However, large-scale distributed PV grid connection presents new challenges to the distribution network. From a technical perspective, firstly, distributed PV is characterized by numerous nodes, wide coverage, and localized high-density grid connection, leading to high management costs and difficulties. Secondly, traditional power grids lack the effective absorption capacity to support distributed energy, and large-scale distributed PV grid connection can cause bidirectional power flow in the distribution network, potentially triggering risks such as voltage exceeding limits, power fluctuations, and power quality degradation. Furthermore, the volatility of solar energy is also a key factor restricting the promotion of distributed PV.

[0003] Currently, to address the voltage exceedance problem caused by distributed photovoltaic grid connection, much research focuses on reactive power optimization (voltage control) of distribution networks, mainly divided into centralized control methods and cluster control methods. Centralized control methods optimize the reactive power of the entire distribution network through a central controller, achieving globally optimal control through unified resource allocation. Cluster control methods divide the entire distribution network into multiple clusters, each controlled independently without interference. Cluster control methods typically neglect the overall coordination of the distribution network and generally cannot obtain a globally optimal solution, but can achieve locally optimal solutions within each cluster area, and their control efficiency is far higher than that of centralized control methods.

[0004] Therefore, the main technical problems are: under the background of "county-wide promotion", the high penetration rate, multiple nodes and wide coverage of distributed photovoltaics are becoming increasingly prominent, and the problems of low efficiency and high dependence on communication system of traditional centralized control methods are further amplified; existing cluster control methods rarely consider the impact of active power-voltage sensitivity and usually ignore the overall coordination of the distribution network, which generally cannot obtain the global optimal solution, resulting in excessive calculation error. Summary of the Invention

[0005] In view of the aforementioned existing problems, this invention is proposed. Therefore, this invention provides a county-wide photovoltaic cluster voltage control method based on an improved differential evolution algorithm to address the problems of voltage exceeding limits and power quality degradation caused by the high penetration rate of distributed photovoltaics in the distribution network, which seriously affect the safety and reliability of the power system. Traditional centralized control methods for distribution networks suffer from low efficiency and insufficient practicality in engineering applications.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0007] In a first aspect, the present invention provides a county-wide photovoltaic cluster voltage control method based on an improved differential evolution algorithm, comprising: calculating electrical data between nodes in a distribution network and preprocessing the electrical data;

[0008] Based on the preprocessed electrical data between nodes, cluster partitioning is performed through modularity optimization. The objective function for cluster partitioning is set to obtain the power grid cluster partitioning results.

[0009] Based on the grid cluster division results, an improved differential evolution algorithm is used within each cluster to construct a cluster voltage control model with reactive power compensation of remaining capacity of photovoltaic inverters and active power reduction of distributed photovoltaics as control variables, so as to optimize the voltage control of the distribution network.

[0010] As a preferred embodiment of the county-wide photovoltaic cluster voltage control method based on the improved differential evolution algorithm described in this invention, the calculation of electrical data between nodes in the distribution network and the preprocessing of the electrical data include: obtaining the electrical distance between nodes through the sensitivity relationship between voltage amplitude and power, normalizing the electrical distance, and establishing a side weight matrix.

[0011] As a preferred embodiment of the county-wide photovoltaic cluster voltage control method based on the improved differential evolution algorithm described in this invention, the cluster is partitioned through modularity optimization, and the objective function for cluster partitioning includes:

[0012] Based on the edge weight matrix, with the goal of minimizing the change in modularity, the objective function for cluster partitioning is set to partition the clusters, and the number and boundaries of the clusters are automatically determined.

[0013] Each node is sequentially assigned to all adjacent clusters, and the change in modularity is calculated. If the change in modularity is greater than zero, the current node is assigned to the cluster with the largest change in modularity; otherwise, it remains unchanged.

[0014] Determine whether the cluster dependencies and cluster structure are constant. If they are constant, complete the cluster partitioning. If they are not constant, repeat the above steps until they are constant.

[0015] The advantage of this preferred solution is that it eliminates the need to pre-specify the number of clusters and can adaptively complete cluster partitioning based on the distribution network attributes.

[0016] As a preferred embodiment of the county-wide photovoltaic cluster voltage control method based on the improved differential evolution algorithm described in this invention, the improved differential evolution algorithm includes:

[0017] The improved differential evolution algorithm optimizes the mutation process by introducing a mutation operation based on Cauchy distribution.

[0018] Select multiple optimal solution sets from the solution set, and then apply Cauchy perturbation to these solutions according to the Cauchy distribution function to generate the Cauchy perturbation set of the optimal solution set;

[0019] Based on the Cauchy perturbation set, combined with the worst solution set eliminated in the previous iteration, a Cauchy mutation set is generated for mutation operation.

[0020] The advantages of this preferred solution are that it can solve problems such as insufficient optimization ability, premature convergence, and poor stability.

[0021] As a preferred embodiment of the county-wide photovoltaic cluster voltage control method based on the improved differential evolution algorithm described in this invention, the construction of the cluster voltage control model includes:

[0022] The objective function of voltage optimization control is set with the weighted sum of node voltage deviation, network active power loss and photovoltaic active power reduction as the control objective. Newton-Raphson power flow calculation is performed to obtain the active power loss of the branch.

[0023] Constraints are set for the reduction of active power generation from photovoltaic power generation and the use of the remaining capacity of the inverter for reactive power compensation, and the photovoltaic reduction rate is used as an indicator of the distributed photovoltaic absorption capacity.

[0024] The advantage of this preferred solution is that voltage control is based on virtual balance nodes, which effectively avoids the problem of excessive calculation errors caused by neglecting global coordination due to independent control.

[0025] As a preferred embodiment of the county-wide photovoltaic cluster voltage control method based on the improved differential evolution algorithm described in this invention, the voltage optimization control of the distribution network includes:

[0026] Based on the objective function of the voltage optimization control, independent voltage control is performed on each independent cluster;

[0027] Power flow calculation is performed from the front-end cluster to the back-end cluster with branch connections. After the front-end cluster completes power flow calculation and voltage control, the node connected to the back-end cluster is found in the front-end cluster and used as the balancing node of the back-end cluster to obtain the reactive power compensation and active power reduction of each photovoltaic node. The effectiveness of the cluster voltage control strategy is verified by comparing the differences between cluster control and centralized control.

[0028] When performing voltage control, if the voltage exceeds the limit, the remaining reactive capacity of each photovoltaic inverter is first used for reactive power compensation. If the voltage still exceeds the limit, active power reduction is then performed according to sensitivity ranking until the voltage of each node is limited to the set range.

[0029] As a preferred embodiment of the county-wide photovoltaic cluster voltage control method based on the improved differential evolution algorithm described in this invention, the objective function of the voltage optimization control is expressed as:

[0030] min F=α1F dv +α2F Ploss +α3F Pcu +F pt

[0031]

[0032] Among them, F dv U represents the sum of squares of voltage deviations at each node of the distribution network. i F represents the voltage at each node; Ploss This represents the sum of active power losses in each branch of the distribution network. Indicates the active power loss of each branch; F Pcu This indicates the total reduction in active power from photovoltaic power generation. Indicates the active power reduction at each photovoltaic node; F pt U represents the voltage over-limit penalty. vu,i and U vl,j These represent the node voltage values ​​when the voltage exceeds the upper and lower limits, respectively; U up and U low α1 represents the set upper and lower voltage thresholds; α2 and α3 represent the weighting coefficients for node voltage deviation, active power loss, and photovoltaic active power reduction, respectively; α4 is the voltage over-limit penalty factor.

[0033] Secondly, this invention provides a county-wide photovoltaic cluster voltage control system based on an improved differential evolution algorithm, comprising:

[0034] The calculation module is used to calculate the electrical data between nodes in the power distribution network and to preprocess the electrical data.

[0035] The partitioning module is used to partition the cluster based on the preprocessed electrical data between nodes through modularity optimization. The objective function for cluster partitioning is set to obtain the power grid cluster partitioning result.

[0036] The optimization control module is used to construct a cluster voltage control model within each cluster based on the grid cluster division results, using an improved differential evolution algorithm and taking the reactive power compensation of the remaining capacity of photovoltaic inverters and the active power reduction of distributed photovoltaics as control variables, to perform voltage optimization control of the distribution network.

[0037] Thirdly, the present invention provides a computer device, comprising:

[0038] Memory and processor;

[0039] The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the computer-executable instructions are executed by the processor, they implement the steps of the county-wide photovoltaic cluster voltage control method based on the improved differential evolution algorithm.

[0040] Fourthly, the present invention provides a computer-readable storage medium storing computer-executable instructions that, when executed by a processor, implement the steps of the county-wide photovoltaic cluster voltage control method based on the improved differential evolution algorithm.

[0041] Compared with existing technologies, the beneficial effects of this invention are as follows: Addressing the limitation of existing cluster partitioning methods that rarely consider the influence of active power-voltage sensitivity, the cluster partitioning index proposed in this invention comprehensively considers the influence of both active and reactive power, and establishes a normalized distribution network edge weight matrix based on this. This not only enables adaptive clustering of the distribution network but also has a faster computation speed. The differential evolution algorithm is improved and applied to distribution network cluster control, resulting in better optimization accuracy and computational stability, effectively solving the problem of premature convergence in traditional optimization algorithms. Based on virtual balance nodes, cluster voltage control allows each cluster to exchange voltage information through terminal nodes, which can, to some extent, avoid the problem of excessive computational errors caused by independent control neglecting global coordination. Attached Figure Description

[0042] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0043] Figure 1 This is a schematic diagram of the overall process of a county-wide photovoltaic cluster voltage control method based on an improved differential evolution algorithm according to an embodiment of the present invention;

[0044] Figure 2 This is a schematic diagram of the specific process of the county-wide photovoltaic cluster voltage control method based on the improved differential evolution algorithm according to an embodiment of the present invention;

[0045] Figure 3 This is a schematic diagram of the overall process of a county-wide photovoltaic cluster voltage control method based on an improved differential evolution algorithm according to an embodiment of the present invention;

[0046] Figure 4 This is a schematic diagram of the overall process of a county-wide photovoltaic cluster voltage control method based on an improved differential evolution algorithm according to an embodiment of the present invention;

[0047] Figure 5This is a schematic diagram of the group partitioning result of an IEEE 33 bus system for a county-wide photovoltaic cluster voltage control method based on an improved differential evolution algorithm, as described in an embodiment of the present invention.

[0048] Figure 6 This is a schematic diagram of the convergence path of the objective function value of the county-wide photovoltaic cluster voltage control method based on the improved differential evolution algorithm according to an embodiment of the present invention;

[0049] Figure 7 The results of different voltage control methods in the IEEE 33 bus system of the county-wide photovoltaic cluster voltage control method based on the improved differential evolution algorithm described in one embodiment of the present invention under the condition of low penetration rate of DPV power grid;

[0050] Figure 8 This is a schematic diagram of the solution parameters for two control methods of the IEEE 33 bus system under high penetration conditions of DPV power grid, based on the improved differential evolution algorithm for county-wide photovoltaic cluster voltage control, as described in an embodiment of the present invention.

[0051] Figure 9 This is a schematic diagram of the node voltage distribution of an IEEE33 bus system with a3 values ​​of 0, 0.01, 0.1 and 1, based on an improved differential evolution algorithm for controlling the voltage of a county-wide photovoltaic cluster, as described in an embodiment of the present invention. Detailed Implementation

[0052] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0053] Example 1

[0054] Reference Figures 1-4 As an embodiment of the present invention, a county-wide photovoltaic cluster voltage control method based on an improved differential evolution algorithm is provided, such as... Figure 1 , Figure 2 As shown, it includes:

[0055] S101, calculate the electrical distance between nodes in the distribution network and preprocess the electrical distance data;

[0056] S102, Based on the preprocessed electrical data between nodes, cluster partitioning is performed through modularity optimization. The objective function for cluster partitioning is set to obtain the power grid cluster partitioning result.

[0057] S103, based on the grid cluster division results, uses an improved differential evolution algorithm within each cluster to construct a cluster voltage control model with reactive power compensation of remaining capacity of photovoltaic inverters and active power reduction of distributed photovoltaics as control variables, and performs voltage optimization control of the distribution network.

[0058] In a preferred embodiment, calculating electrical data between nodes in the distribution network and preprocessing the electrical data includes: obtaining the electrical distance between nodes through the sensitivity relationship between voltage amplitude and power, normalizing the electrical distance, and establishing a side weight matrix.

[0059] Specifically, the electrical distance between nodes in a distribution network reflects the degree of electrical coupling, which is usually obtained through the relationship between voltage amplitude and power sensitivity. Taking reactive power-voltage sensitivity as an example, its expression is:

[0060]

[0061] in, Represents the reactive power-voltage sensitivity matrix, ΔV i ΔQ represents the voltage magnitude change at node i. j The formula represents the change in unit reactive power at node j. This formula indicates the impact of injecting unit reactive power at node j on the voltage amplitude change at node i, reflecting the degree of coupling between the two nodes to some extent. Similarly, the active power-voltage sensitivity can be defined. Power-voltage sensitivity matrix Determined by both the reactive power-voltage sensitivity matrix and the active power-voltage sensitivity matrix, it can be expressed as:

[0062]

[0063] Based on the definition of power-voltage sensitivity, electrical distance is further defined as follows:

[0064]

[0065] In the formula, d ij The common logarithm of d represents the ratio of the voltage change at node j to the voltage change at node i after a change in power at node j. In other words, d ij The larger d is, the smaller the influence of node j on node i, and the greater the electrical distance between the two nodes. Therefore, d ij It can be used as a basis for the division of the distribution network; ΔV j This represents the voltage amplitude change at node j.

[0066] Furthermore, based on this, the electrical distance is normalized, and a boundary weight matrix is ​​established, represented as:

[0067]

[0068] Where, d max and d min These are the maximum and minimum elements in the electrical distance matrix, respectively, which can restrict the elements in the edge weight matrix to the range [0,1].

[0069] In one alternative implementation, electrical data in the distribution network can be power coupling strength, i.e., the degree of mutual influence of power flow between nodes, such as the active / reactive power flow coupling coefficient, and the power interaction relationship between nodes can be analyzed through power flow calculation; mutual impedance matrix, the impedance values ​​between nodes reflect the electrical coupling strength, wherein the equivalent impedance between nodes is calculated based on the impedance matrix of the distribution network, and the power flow distribution data between nodes is obtained through actual measurement or simulation.

[0070] In another alternative implementation, the electrical data in the distribution network can be voltage correlation coefficients, i.e., the statistical correlation of node voltage fluctuations. The correlation coefficients between nodes can be calculated using historical voltage data, or voltage scenarios can be generated based on stochastic power flow or Monte Carlo simulations, followed by statistical analysis.

[0071] In a preferred embodiment, such as Figure 3 As shown, cluster partitioning is performed through modularity optimization, and the objective function for cluster partitioning includes:

[0072] Based on the edge weight matrix, with the goal of minimizing the change in modularity, the objective function for cluster partitioning is set to partition the clusters, and the number and boundaries of the clusters are automatically determined.

[0073] Each node is assigned to all adjacent clusters in turn. The change in modularity is calculated. If the change in modularity is greater than zero, the current node is assigned to the cluster with the largest change in modularity; otherwise, it remains unchanged.

[0074] Determine whether the cluster dependencies and cluster structure are constant. If they are constant, complete the cluster partitioning. If they are not constant, repeat the above steps until they are constant.

[0075] It should be noted that, based on the edge weight index of electrical distance quantification, a cluster partitioning method based on modularity optimization is proposed. This method does not require pre-specifying the number of clusters and can adaptively complete the cluster partitioning according to the distribution network attributes, thus having stronger scalability.

[0076] Specifically, the objective function for cluster partitioning is expressed as:

[0077]

[0078] Where ρ is the modularity; a ij This represents the elements of the edge weight matrix, i.e., the weights of the edges connecting nodes i and j; This represents the sum of all elements in the edge weight matrix; k is the sum of the weights of all edges connected to node i; j The modularity is the sum of the weights of all edges connected to node j. ψ(i,j) = 1 when nodes i and j belong to the same cluster, and ψ(i,j) = 0 otherwise. The modularity value typically ranges from 0 to 1; a larger value indicates better clustering performance and higher partitioning accuracy.

[0079] In one alternative implementation, to achieve coordination and complementarity between clusters, the power balance principle of the distribution network and the electrical connection requirements between nodes must be met. In this case, the electrical connection between nodes within the same cluster is tight, while the connection between clusters is loose, which is conducive to the effective implementation of cluster voltage control. Most related studies use k-means and its derivative algorithms to pre-specify the number of clusters and complete the cluster division.

[0080] In another alternative implementation, cluster partitioning can also be achieved through hierarchical clustering. A tree diagram can be constructed from bottom to top (aggregation) or from top to bottom (split), and clusters can be formed by cutting according to a threshold. This can generate a multi-scale cluster structure, which is suitable for hierarchical coordination of voltage control.

[0081] In a preferred embodiment, the improved differential evolution algorithm includes:

[0082] The improved differential evolution algorithm optimizes the mutation process by introducing a mutation operation based on Cauchy distribution.

[0083] Select multiple optimal solution sets from the solution set, and then apply Cauchy perturbation to these solutions according to the Cauchy distribution function to generate the Cauchy perturbation set of the optimal solution set;

[0084] Based on the Cauchy perturbation set, combined with the worst solution set eliminated in the previous iteration, a Cauchy mutation set is generated for mutation operation.

[0085] In one alternative implementation, the Differential Evolution (DE) algorithm is used to control the voltage of the distribution network. Its process includes initialization, mutation, crossover, and selection. This algorithm requires very few control variables and has advantages such as fast convergence, good robustness, and strong optimization capabilities. However, when dealing with multi-parameter collaborative optimization problems in distribution networks, the DE algorithm may still face problems such as insufficient optimization capability, premature convergence, and poor stability.

[0086] This embodiment improves the Cauchy perturbation algorithm and optimizes the mutation process of the basic DE algorithm. The difference between the Improved Differential Evolution (IDE) algorithm and the basic DE algorithm lies in the introduction of a mutation operation based on the Cauchy distribution and a crossover operation based on the central solution. The Cauchy perturbation is based on the Cauchy distribution function, expressed as:

[0087]

[0088] Where x is a random variable representing the position in the solution space; x0 is the position parameter, representing the center position of the Cauchy distribution, usually taken as the current optimal solution; γ is the scale parameter, which controls the perturbation amplitude, and the larger the value, the wider the perturbation range.

[0089] When performing Cauchy perturbation, we first need to select bNP optimal solution sets as bX from the solution set X, and then perturb these solutions according to the above formula, expressed as:

[0090] dX=bX+etatan(π×(rand(bNP,n)-0.5))

[0091] Where η is the Cauchy perturbation coefficient; n is the dimension of the solution set (which can be understood as the number of nodes in the distribution network); rand(bNP,n) is a random number matrix of rows and columns bNP, and bNP is the size of the elite solution set, representing the number of optimal solutions selected from the population. Considering that centralized control involves more nodes while cluster control has fewer nodes per cluster, the expression for bNP is defined as follows:

[0092]

[0093] The Cauchy perturbation coefficient can be expressed as:

[0094]

[0095] Where gen is the current iteration number; G is the maximum iteration number; and δ is the adjustment parameter, expressed as:

[0096]

[0097] Among them, F worst F represents the worst value of the objective function. best F represents the optimal value of the objective function. mean ζ represents the average value of the objective function; ζ represents the parameter of the small perturbation.

[0098] The Cauchy perturbation set dX generated by the above operations can be understood as several solutions near the optimal solution in the solution space. Its purpose is to search for better solutions within the neighborhood of the optimal solution in the solution space, and further refine this search based on bX, dX, and the worst solution set wX eliminated in the previous iteration. (-1) Generate the Cauchy mutation set mX, represented as:

[0099] mX=(X-bX)∪dX∪wX (-1)

[0100] Basic DE typically employs the following mutation strategy m i :

[0101] m i =Xr1 +M(X r2 -X r3 )

[0102] Where M is the coefficient of variation, X r1 X r2 X r3 Let X represent the mutation parameters. In this strategy, all three mutation parameters originate from the original solution set X. The mutation process is non-directional and highly random, resulting in slow convergence and a tendency to get trapped in local optima. Therefore, after generating the Cauchy perturbation set dX and the Cauchy mutation set mX, the following strategy is used for the mutation operation, expressed as:

[0103] m i =X i +M(X best -X i +X r1 -mX r2 )

[0104] Among them, X best The optimal solution in X or (X-bX)∪dX; X i Let X represent the i-th individual (solution vector) in the population during the current iteration. r2 X represents the variation parameter. r1 and mX r2 They come from solution sets X and mX, respectively. best The mutation operation is directional, guiding the population towards the current optimal solution set, thus improving convergence speed and algorithm accuracy. Furthermore, `mX` mitigates the potential loss of solution information caused by greedy selection, recovering eliminated solutions with a certain probability during iteration, increasing population diversity. Therefore, the mutation operation is not only related to the initial solution set `X`, but also to the Cauchy perturbation set `dX` and the eliminated solutions from the previous iteration. This strategy changes the direction of algorithm optimization, preventing the differential evolution process from prematurely falling into local optima.

[0105] Furthermore, the population size constraint in the Differential Evolution (DE) algorithm reflects the amount of information contained in the population. Increasing NP will increase the computational load, while decreasing it will improve computational speed but limit population diversity, which is detrimental to the algorithm obtaining the global optimum. In addition, the value of NP should be related to the number of distribution network nodes n to reflect the scale of the distribution network.

[0106] When the active power reduction of distributed photovoltaic power is not considered, the optimization independent variable is the reactive power compensation of the photovoltaic inverter at each node, and the NP expression is:

[0107]

[0108] When considering the active power reduction of distributed photovoltaic (PV) power, the optimization variables include the reactive power compensation of the PV inverter and the active power reduction of the PV power. These two variables have a synergistic relationship, requiring a larger population size to ensure the algorithm's collaborative optimization capability. This is expressed as:

[0109]

[0110] When no cluster partitioning is performed, n represents the total number of nodes in the distribution network; otherwise, it represents the number of nodes in each cluster.

[0111] In a preferred embodiment, constructing the cluster voltage control model includes:

[0112] The objective function of voltage optimization control is set with the weighted sum of node voltage deviation, network active power loss and photovoltaic active power reduction as the control objective. Newton-Raphson power flow calculation is performed to obtain the active power loss of the branch.

[0113] Constraints are set for the active power reduction of photovoltaic power generation and the reactive power compensation when the remaining capacity of the inverter is called, and the photovoltaic reduction rate is used as an indicator of the distributed photovoltaic absorption capacity.

[0114] In a preferred embodiment, the objective function for voltage optimization control is expressed as:

[0115] minF=α1F dv +α2F Ploss +α3F Pcu +F pt

[0116]

[0117] Among them, F dv U represents the sum of squares of voltage deviations at each node of the distribution network. i F represents the voltage at each node; Ploss This represents the sum of active power losses in each branch of the distribution network. Indicates the active power loss of each branch; F Pcu This indicates the total reduction in active power from photovoltaic power generation. Indicates the active power reduction at each photovoltaic node; F pt U represents the voltage over-limit penalty. vu,i and U vl,j These represent the node voltage values ​​when the voltage exceeds the upper and lower limits, respectively; U up and U low α1 represents the set upper and lower voltage thresholds; α2 and α3 represent the weighting coefficients for node voltage deviation, active power loss, and photovoltaic active power reduction, respectively; α4 is the voltage over-limit penalty factor.

[0118] It should be noted that the purpose of the voltage over-limit penalty term is to prevent voltage over-limits. Its control capability is related to the value of α4. When there is no over-limit, F pt =0, let α4 >> α1, α2, α3. Once a voltage over-limit occurs, the objective function value will increase significantly, thereby avoiding node voltage over-limit to a certain extent. Take α4 = 100.

[0119] In this embodiment, the Newton-Raphson power flow calculation is performed as follows:

[0120] The power balance equation of the Newton-Raphson power flow algorithm in Cartesian coordinates is expressed as:

[0121]

[0122]

[0123] U i =e i +jf i ,

[0124] Among them, e i and f i Let e ​​be the real and imaginary parts of the voltage at node i; j and f j Let P be the real and imaginary parts of the voltage at node j. cv For the maximum active power output of photovoltaic node i, P cv -P cu This represents the actual amount of electricity generated; Q C P is the reactive power compensation of the photovoltaic inverter. L and Q L G represents the active and reactive power of the load at node i; ij and B ij For the admittance matrix element Y ij The real and imaginary parts (conductance and susceptance).

[0125] Based on the Newton-Raphson power flow calculation results, the active power loss of branch ij is expressed as:

[0126]

[0127] Among them, U i U j Let θ be the voltage across the branch. ij This represents the phase difference.

[0128] Furthermore, when setting operational constraints for distributed photovoltaic systems, and when reducing active power generation from photovoltaic power generation and using the remaining capacity of inverters for reactive power compensation, the following constraints need to be considered:

[0129]

[0130] Among them, Si Capacity of the photovoltaic inverter at node i; This is the inverter power factor angle.

[0131] In this embodiment, the Power Curtailment Rate (PCR) is used as an indicator of distributed photovoltaic (PV) grid integration capacity. It is defined as the ratio of PV power generation reduced due to grid inability to absorb it to the total available power generation, expressed as:

[0132]

[0133] The smaller the PCR value, the stronger the distribution network's ability to absorb distributed photovoltaic power.

[0134] In a preferred embodiment, voltage optimization control of the distribution network includes:

[0135] Based on the objective function of voltage optimization control, independent voltage control is performed on each independent cluster;

[0136] Power flow calculation is performed from the front-end cluster to the back-end cluster with branch connections. After the front-end cluster completes power flow calculation and voltage control, the node connected to the back-end cluster is found in the front-end cluster and used as the balancing node of the back-end cluster to obtain the reactive power compensation and active power reduction of each photovoltaic node. The effectiveness of the cluster voltage control strategy is verified by comparing the differences between cluster control and centralized control.

[0137] When performing voltage control, if the voltage exceeds the limit, the remaining reactive capacity of each photovoltaic inverter is first used for reactive power compensation. If the voltage still exceeds the limit, active power reduction is then performed according to sensitivity ranking until the voltage of each node is limited to the set range.

[0138] It should be noted that in the cluster voltage control model adopted, voltage control is based on virtual balance nodes. Each cluster exchanges voltage information through terminal nodes, which effectively avoids the problem of excessive calculation error caused by independent control ignoring global coordination.

[0139] Specifically, independent voltage control is performed in each independent cluster. Due to the radial structure of the distribution network, power flow calculations are typically performed from the front-end clusters to the back-end clusters, and there must be a branch connection between adjacent clusters. Therefore, after the front-end cluster completes power flow calculations and voltage control, it identifies the node connected to the back-end cluster from the front-end cluster and copies this node and its voltage information to the back-end cluster as its balancing node. This node is called the "virtual balancing node" of the back-end cluster. The power flow calculations and voltage control of the back-end cluster are based on the voltage of this virtual balancing node. A schematic diagram of this operation is shown below. Figure 4 As shown.

[0140] After the cluster partitioning is completed, the balancing node of cluster b is denoted as b0, and its voltage can be expressed as:

[0141]

[0142] Among them, a j This represents the upstream node in cluster a that is connected to cluster b.

[0143] Furthermore, by ignoring power flow between clusters, the calculation process for inter-cluster coupling is simplified. This significantly reduces information exchange between clusters, mitigating to some extent the decrease in global coordination caused by cluster control. The reactive power compensation and active power reduction of each photovoltaic node are obtained through the above calculations. These data are then input into the global power flow model for overall calculation. The effectiveness of the cluster voltage control method is verified by comparing the differences between cluster control and centralized control.

[0144] This embodiment adopts a "compensation first, reduction later" control method. When performing voltage control, the remaining reactive capacity of each photovoltaic inverter is first used for reactive compensation. When the program determines that the maximum reactive compensation still cannot solve the voltage over-limit problem, active power reduction is further performed until the voltage of each node is limited to the allowable range.

[0145] In this voltage control scenario, the allowable voltage deviation under global conditions is set to 0.05 pu. When using a cluster voltage control strategy, considering that it ignores global coordination to some extent, the deviation may increase when verifying the cluster control results under global conditions. Therefore, when performing voltage control within a single cluster, the allowable voltage deviation must be even smaller; in this embodiment, it is set to 0.025 pu, which is half of that for centralized control.

[0146] It should be noted that in this invention, cluster division is performed, the voltage sensitivity between each node is analyzed, the "electrical distance" between nodes is quantified, and nodes with strong coupling are grouped into the same cluster. The number and boundaries of clusters are automatically determined through modularity optimization to ensure strong coupling between nodes within a cluster and weak coupling between clusters. Hierarchical control is adopted, and the IDE algorithm based on Cauchy perturbation is used in each cluster. The reactive power compensation of the remaining capacity of the photovoltaic inverter and the active power reduction of distributed photovoltaic power are used as control variables ("reactive power compensation takes priority, active power reduction comes later") to achieve voltage optimization control.

[0147] To address the limitation of existing cluster partitioning methods that rarely consider the impact of active power-voltage sensitivity, this invention proposes a cluster partitioning index that comprehensively considers the influence of both active and reactive power. Based on this, a normalized distribution network edge weight matrix is ​​established, which not only enables adaptive clustering of the distribution network but also has a faster computation speed. The differential evolution algorithm is improved and applied to distribution network cluster control, resulting in better optimization accuracy and computational stability, effectively solving the problem of premature convergence in traditional optimization algorithms. Based on cluster voltage control using virtual balance nodes, each cluster exchanges voltage information through terminal nodes, which can, to some extent, avoid the problem of excessive computational errors caused by independent control neglecting global coordination.

[0148] The above is a schematic scheme of the county-wide photovoltaic cluster voltage control method based on the improved differential evolution algorithm in this embodiment. It should be noted that the technical solution of the county-wide photovoltaic cluster voltage control system based on the improved differential evolution algorithm is based on the same concept as the technical solution of the county-wide photovoltaic cluster voltage control method based on the improved differential evolution algorithm described above. Details not described in detail in the technical solution of the county-wide photovoltaic cluster voltage control system based on the improved differential evolution algorithm in this embodiment can be found in the description of the technical solution of the county-wide photovoltaic cluster voltage control method based on the improved differential evolution algorithm described above.

[0149] Example 2

[0150] This embodiment provides a county-wide photovoltaic cluster voltage control system based on an improved differential evolution algorithm, including:

[0151] The calculation module is used to calculate the electrical distance between nodes in the distribution network and to preprocess the electrical distance data.

[0152] The partitioning module is used to partition the cluster based on the preprocessed electrical data between nodes through modularity optimization. The objective function for cluster partitioning is set to obtain the power grid cluster partitioning result.

[0153] The optimized control module is used to construct a cluster voltage control model within each cluster based on the grid cluster division results. This model uses an improved differential evolution algorithm with reactive power compensation of the remaining capacity of photovoltaic inverters and active power reduction of distributed photovoltaics as control variables, and performs voltage optimization control of the distribution network.

[0154] This embodiment also provides a computer device suitable for county-wide photovoltaic cluster voltage control based on an improved differential evolution algorithm, including:

[0155] The system includes a memory and a processor. The memory stores computer-executable instructions, and the processor executes these instructions to implement the county-wide photovoltaic cluster voltage control method based on the improved differential evolution algorithm proposed in the above embodiments.

[0156] This embodiment also provides a storage medium storing a computer program that, when executed by a processor, implements the county-wide photovoltaic cluster voltage control method based on the improved differential evolution algorithm proposed in the above embodiments.

[0157] The storage medium proposed in this embodiment and the county-wide photovoltaic cluster voltage control method based on the improved differential evolution algorithm proposed in the above embodiments belong to the same inventive concept. Technical details not described in detail in this embodiment can be found in the above embodiments, and this embodiment has the same beneficial effects as the above embodiments.

[0158] Based on the above description of the implementation methods, those skilled in the art can clearly understand that the present invention can be implemented using software and necessary general-purpose hardware, and of course, it can also be implemented using hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk, or optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of the various embodiments of the present invention.

[0159] Example 3

[0160] Refer to Tables 1-2 and Figures 5-9 As an embodiment of the present invention, a county-wide photovoltaic cluster voltage control method based on an improved differential evolution algorithm is provided. To verify its beneficial effects, it is scientifically demonstrated through economic benefit calculations and simulation experiments.

[0161] In this embodiment, the voltage control method based on clustering partitioning and the voltage control method based on integral-differential evolution algorithm were verified on the standard IEEE 33-node distribution system and the distribution network platform of a county in Guangdong Province, China. All algorithms were implemented and run in the MATLAB environment.

[0162] According to the clustering partitioning method of the present invention, node partitioning is performed on the IEEE 33-node distribution system containing distributed photovoltaic power, and the results are as follows. Figure 5 As shown, good zoning effects can be achieved in both low (ρ = 0.716) and high (ρ = 0.7202) distributed photovoltaic (DPV) penetration rates. Figure 5Black nodes represent existing nodes that have not yet connected to DPV; blue nodes represent nodes that have connected to DPV in a low-penetration scenario; and red nodes represent newly added nodes that have connected to DPV in a high-penetration scenario (relative to the low-penetration state).

[0163] As can be seen, after high-penetration DPV access, the present invention not only significantly increases the number of DPV access points in the system, but also increases the installed capacity of inverters at some nodes, thereby posing higher voltage and stability control requirements for the normal operation of the power distribution network.

[0164] Figure 6 This diagram illustrates the convergence path of the objective function value F of DPVs under low and high penetration conditions when using IDE and centralized control methods. The black curve corresponds to the low penetration condition. The algorithm converges after 137 iterations, and the horizontal segment in the latter half of the black curve represents the process of determining whether the convergence condition is met. Conversely, the red curve corresponds to the high penetration condition. The algorithm converges after a total of 313 iterations, and the initial part of the curve corresponds to the case where only inverter capacity is used for reactive power compensation. However, even maximum reactive power compensation cannot solve the voltage overrun problem in the distribution network, and active power reduction is further implemented to adjust the voltage of each node back to the threshold range. The sharp drop in the middle of the curve indicates that the calculation program determines that reactive power compensation alone cannot achieve the control objective, thus requiring active power reduction. The change in the curve also confirms the voltage overrun penalty term F. pt The necessity of this is that, in order to obtain a smaller F, the algorithm must be optimized in the direction of avoiding voltage overshoot.

[0165] 1. Performance comparison of IDE with other algorithms

[0166] The performance of IDE, basic DE, and two classic optimization algorithms (Particle Swarm Optimization (PSO) and Ant Colony Optimization (ACO)) was compared under centralized and clustered voltage control modes. To obtain sufficient samples, each of the four algorithms was run independently 10 times under different modes. After counting the number of occurrences of each mode in each set of results, the objective function values ​​were averaged, and the standard deviation and coefficient of variation (the ratio of standard deviation to mean) were further calculated. The weighting coefficients of the objective function were all set to the default values: α1 = α2 = α3 = 1.

[0167] In addition, since only reactive power compensation is performed under low penetration conditions, the calculation process is simple and the results of each algorithm are not significantly different. Therefore, the case of DPVs being integrated into the distribution network is set as high penetration.

[0168] First, the performance of the four algorithms under centralized voltage control mode is compared, and the results are shown in Table 1. According to Table 1, under centralized control mode, ACO has the slowest convergence speed and poor optimization ability, reaching the maximum number of iterations (1000) in each calculation. Although PSO converges faster, its optimization ability is poor, its robustness is low, and it is prone to getting trapped in local optima. Basic DE has high optimization accuracy, but its calculation speed is slow. Compared with the first three, IDE has significant performance advantages: it can not only obtain more accurate function values, but also has more stable results and good algorithm robustness; in addition, IDE has stronger optimization ability, fewer iterations, and shorter time overhead.

[0169] Table 1: Performance Comparison of Four Algorithms under Centralized Voltage Control Mode

[0170]

[0171] The performance of the four algorithms was then compared under cluster voltage control mode, and the results are shown in Table 2. According to Table 2, under cluster control mode, ACO struggles to achieve optimal results, exhibiting poor robustness, long optimization time, and the worst performance. PSO performs relatively better, but still suffers from low optimization accuracy and poor robustness. Compared to the basic DE, IDE demonstrates superior optimization accuracy and robustness, with more stable results. However, in this mode, IDE no longer holds the efficiency advantage over the basic DE, and its computation time even exceeds that of the basic DE. This is because the number of nodes per cluster is significantly reduced in cluster control mode compared to centralized control mode, greatly reducing optimization complexity. Therefore, even the basic DE can reach the optimum with fewer iterations. Simultaneously, IDE introduces Cauchy perturbation operations, slightly increasing the time consumption per iteration.

[0172] Table 2: Performance Comparison of Four Algorithms under Cluster Voltage Control Mode

[0173]

[0174] In summary, IDE outperforms PSO, ACO, and basic DE in terms of optimization accuracy and result stability. Although IDE takes slightly longer to compute in cluster mode than basic DE, the overall difference is only about 9.8%, so the following discussion is based on IDE.

[0175] 2. Comparison of cluster voltage control and centralized voltage control

[0176] First, the results of different voltage control methods under grid-connected conditions for low-penetration DPVs were compared, such as... Figure 7As shown, before voltage control was implemented, only node 16 in the distribution network experienced over-limit conditions, and both centralized and cluster control methods could bring the voltage back to the set range. The difference lies in the fact that the voltage obtained by the cluster control method is generally lower than that obtained by the centralized control method. On the one hand, this is because low-penetration grid connection has little impact on the overall voltage rise of the distribution network; on the other hand, cluster control ignores the line voltage drop of the entire distribution network, controlling only the nodes within each cluster, resulting in more inductive reactive power input to reduce the line voltage drop.

[0177] Secondly, the control results of different voltage control methods under grid-connected conditions of high-penetration DPVs were compared. Tables 1 and 2 show that the cluster voltage control method sacrifices some accuracy in the optimization results, resulting in a 14.3% increase in the objective function value F compared to the centralized control method. However, it significantly improves computational efficiency, with a runtime only 4% of that of centralized control.

[0178] Finally, the solution parameters of the two control methods under high permeability conditions were compared, including the sum of squared node voltage deviations F. dv Active power loss F in distribution network loss And the total amount of active reduction F pcu The result is as follows Figure 8 As shown, compared to centralized control, cluster control increases active power reduction, but correspondingly slightly reduces node voltage deviation and distribution network active power losses. In this case, the distribution network active power reduction rate (PCR) is approximately 5.98% under centralized control and approximately 7.55% under cluster control.

[0179] In summary, cluster control can effectively keep the voltage of each node within the threshold range, preventing voltage exceedances and improving power quality. However, because it only considers the interaction within a single cluster and ignores global coordination, it increases active power reduction compared to centralized control, thereby reducing the distribution network's capacity to absorb DPVs. Nevertheless, cluster control has the advantages of short solution time and high computational efficiency, and therefore has better application potential in the context of large-scale, stochastic, and highly volatile DPV grid connection.

[0180] 3. The impact of weighting coefficients on the absorption capacity of DPVs

[0181] Given the high penetration rate of DPVs, this paper discusses the impact of changing the weighting coefficient α3 of the active power reduction term in DPVs on the photovoltaic absorption capacity of the distribution network using the IDE and cluster voltage control methods. Furthermore, to obtain lower voltage deviation and active power loss, α1 = α2 = 1 is set.

[0182] Figure 9The node voltage distributions are given when α3 is 0, 0.01, 0.1, and 1. As α3 increases, the node voltages first gradually rise and then tend to stabilize; the two curves corresponding to α3 = 0.1 and α3 = 1 almost overlap.

[0183] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A county-wide photovoltaic cluster voltage control method based on an improved differential evolution algorithm, characterized in that, include: Calculate the electrical data between nodes in the distribution network and preprocess the electrical data; Based on the preprocessed electrical data between nodes, cluster partitioning is performed through modularity optimization. The objective function for cluster partitioning is set to obtain the power grid cluster partitioning results. Based on the grid cluster division results, an improved differential evolution algorithm is used within each cluster to construct a cluster voltage control model with reactive power compensation of remaining capacity of photovoltaic inverters and active power reduction of distributed photovoltaics as control variables, so as to optimize the voltage control of the distribution network.

2. The county-wide photovoltaic cluster voltage control method based on the improved differential evolution algorithm as described in claim 1, characterized in that, Calculating electrical data between nodes in a distribution network and preprocessing the electrical data includes: obtaining the electrical distance between nodes through the sensitivity relationship between voltage amplitude and power, normalizing the electrical distance, and establishing a side weight matrix.

3. The county-wide photovoltaic cluster voltage control method based on the improved differential evolution algorithm as described in claim 2, characterized in that, Cluster partitioning is performed through modularity optimization, and the objective function for cluster partitioning includes: Based on the edge weight matrix, with the goal of minimizing the change in modularity, the objective function for cluster partitioning is set to partition the clusters, and the number and boundaries of the clusters are automatically determined. Each node is sequentially assigned to all adjacent clusters, and the change in modularity is calculated. If the change in modularity is greater than zero, the current node is assigned to the cluster with the largest change in modularity; otherwise, it remains unchanged. Determine whether the cluster dependencies and cluster structure are constant. If they are constant, complete the cluster partitioning. If they are not constant, repeat the above steps until they are constant.

4. The county-wide photovoltaic cluster voltage control method based on the improved differential evolution algorithm as described in claim 1, characterized in that, The improved differential evolution algorithm includes: The improved differential evolution algorithm optimizes the mutation process by introducing a mutation operation based on Cauchy distribution. Select multiple optimal solution sets from the solution set, and then apply Cauchy perturbation to these solutions according to the Cauchy distribution function to generate the Cauchy perturbation set of the optimal solution set; Based on the Cauchy perturbation set, combined with the worst solution set eliminated in the previous iteration, a Cauchy mutation set is generated for mutation operation.

5. The county-wide photovoltaic cluster voltage control method based on the improved differential evolution algorithm as described in claim 4, characterized in that, The construction of the cluster voltage control model includes: The objective function of voltage optimization control is set with the weighted sum of node voltage deviation, network active power loss and photovoltaic active power reduction as the control objective. Newton-Raphson power flow calculation is performed to obtain the active power loss of the branch. Constraints are set for the active power reduction of photovoltaic power generation and the reactive power compensation when the remaining capacity of the inverter is called, and the photovoltaic reduction rate is used as an indicator of the distributed photovoltaic absorption capacity.

6. The county-wide photovoltaic cluster voltage control method based on the improved differential evolution algorithm as described in claim 5, characterized in that, Voltage optimization control of the distribution network includes: Based on the objective function of the voltage optimization control, independent voltage control is performed on each independent cluster; Power flow calculation is performed from the front-end cluster to the back-end cluster with branch connections. After the front-end cluster completes power flow calculation and voltage control, the node connected to the back-end cluster is found in the front-end cluster and used as the balancing node of the back-end cluster to obtain the reactive power compensation and active power reduction of each photovoltaic node. The effectiveness of the cluster voltage control strategy is verified by comparing the differences between cluster control and centralized control. When performing voltage control, if the voltage exceeds the limit, the remaining reactive capacity of each photovoltaic inverter is first used for reactive power compensation. If the voltage still exceeds the limit, active power reduction is then performed according to sensitivity ranking until the voltage of each node is limited to the set range.

7. The county-wide photovoltaic cluster voltage control method based on the improved differential evolution algorithm as described in claim 5, characterized in that, The objective function for the voltage optimization control is expressed as: minF=α1F dv +α2F Ploss +α3F Pcu +F pt Among them, F dv U represents the sum of squares of voltage deviations at each node of the distribution network. i F represents the voltage at each node; Ploss This represents the sum of active power losses in each branch of the distribution network. Indicates the active power loss of each branch; F Pcu This indicates the total reduction in active power from photovoltaic power generation. Indicates the active power reduction at each photovoltaic node; F pt U represents the voltage over-limit penalty. vu,i and U vl,j These represent the node voltage values ​​when the voltage exceeds the upper and lower limits, respectively; U up and U low α1 represents the set upper and lower voltage thresholds; α2 and α3 represent the weighting coefficients for node voltage deviation, active power loss, and photovoltaic active power reduction, respectively; α4 is the voltage over-limit penalty factor.

8. A county-wide photovoltaic cluster voltage control system based on an improved differential evolution algorithm, employing the county-wide photovoltaic cluster voltage control method based on an improved differential evolution algorithm as described in any one of claims 1 to 7, characterized in that, include: The calculation module is used to calculate the electrical data between nodes in the power distribution network and to preprocess the electrical data. The partitioning module is used to partition the cluster based on the preprocessed electrical data between nodes through modularity optimization. The objective function for cluster partitioning is set to obtain the power grid cluster partitioning result. The optimization control module is used to construct a cluster voltage control model within each cluster based on the grid cluster division results, using an improved differential evolution algorithm and taking the reactive power compensation of the remaining capacity of photovoltaic inverters and the active power reduction of distributed photovoltaics as control variables, to perform voltage optimization control of the distribution network.

9. A computer device, characterized in that, include: Memory and processor; The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the computer-executable instructions are executed by the processor, they implement the steps of the county-wide photovoltaic cluster voltage control method based on the improved differential evolution algorithm as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, It stores computer-executable instructions, which, when executed by a processor, implement the steps of the county-wide photovoltaic cluster voltage control method based on the improved differential evolution algorithm as described in any one of claims 1 to 7.

Images