A power distribution network reactive power optimization method based on improved differential evolution algorithm

Abstract

The application discloses a power distribution network reactive power optimization method based on an improved differential evolution algorithm. If there is voltage out-of-limit at any node in the power distribution network, the capacitive reactive power compensation power of a distributed photovoltaic inverter is adjusted according to an optimal objective function value. If there is still voltage out-of-limit at any node in the power distribution network, the active power of the distributed photovoltaic is simultaneously reduced and the capacitive reactive power compensation power of the distributed photovoltaic inverter is adjusted according to the optimal objective function value. The application has the beneficial effect that the voltage deviation and the minimum active loss are taken as the optimization target, the Cauchy disturbance and other methods are integrated into the differential evolution algorithm, the collaborative optimization of the active reduction of the distributed photovoltaic and the capacitive reactive power compensation of the inverter is carried out, compared with the basic differential evolution algorithm, the application can obtain a better result, effectively improves the convergence accuracy and the algorithm robustness, and can solve the voltage out-of-limit problem of the power distribution network with high penetration rate of the distributed photovoltaic.

Description

Technical Field

[0001] This invention relates to the field of reactive power optimization technology for distribution networks, and in particular to a reactive power optimization method for distribution networks based on an improved differential evolution algorithm. Background Technology

[0002] Driven by the "dual carbon" goals, my country has begun large-scale construction of distributed photovoltaic (PV) power generation projects. On the one hand, connecting distributed PV power sources to the distribution network can achieve local energy balance, avoiding the investment and losses associated with long-distance power transmission. However, distributed PV is characterized by numerous points and a wide distribution area, which differs significantly from traditional centralized power generation methods. The large-scale integration of distributed power sources will inevitably change the traditional unidirectional radial power supply mode of the distribution network, leading to bidirectional power flow problems and causing changes in system power flow and voltage distribution, especially voltage exceeding limits. Therefore, the large-scale, contiguous promotion of distributed PV will inevitably bring challenges and profound impacts to the power grid, fundamentally changing the development model of the power system.

[0003] To address the voltage exceedance problem caused by high-penetration distributed photovoltaic (PV) power generation, it is necessary to combine intelligent optimization algorithms for reactive power optimization control of the distribution network. Commonly used optimization algorithms include Particle Swarm Optimization (PSO), Genetic Algorithm (GA), Simulated Annealing (SA), and Differential Evolutionary Algorithm (DE). Currently, there are some studies involving reactive power optimization methods for distribution networks. For example, the patent with publication number CN 104600714 A discloses a reactive power optimization method and device for distribution networks containing distributed power sources, which uses a genetic algorithm to establish a reactive power optimization model for the distribution network; the patent with publication number CN 105870939 A discloses a reactive power optimization method for distribution networks considering multiple wiring methods, which can perform multi-objective reactive power optimization for three-phase unbalanced radial medium-voltage distribution networks based on particle swarm optimization, and can also optimize the capacity of discrete reactive power compensation devices; the patent with publication number CN 113629789 A discloses a reactive power optimization method and system for distribution networks, which uses dynamic smoothness and reactive power optimization degree to construct an objective function, transforming the reactive power optimization problem into a minimization multi-objective optimization problem, thereby establishing a dynamic reactive power optimization model. However, the above methods still suffer from insufficient calculation accuracy and poor robustness when dealing with reactive power optimization in complex distribution networks with high penetration distributed photovoltaic access; in addition, the premature convergence problem of intelligent optimization algorithms is particularly significant when performing complex multi-parameter optimizations. Summary of the Invention

[0004] To address the aforementioned problems, this invention proposes a reactive power optimization method for distribution networks based on an improved differential evolution algorithm, which mainly solves the problem that existing optimization schemes cannot address voltage exceedance issues in distribution networks with high-penetration distributed photovoltaic access.

[0005] To solve the above-mentioned technical problems, the technical solution of the present invention is as follows:

[0006] A reactive power optimization method for distribution networks based on an improved differential evolution algorithm includes the following steps:

[0007] S1, obtains the electrical parameters of the distribution network lines, load information, and distributed photovoltaic information;

[0008] S2, perform power flow calculation of the distribution network based on the line electrical parameters, the load information and the distributed photovoltaic information to obtain the voltage value of each node and the active power loss of each branch under steady state;

[0009] S3, if any node in the distribution network has a voltage limit exceeding the limit, the voltage values ​​of each node and the active power loss of each branch are input into the improved differential evolution algorithm of Cauchy perturbation operation and central solution cross operation for solving, and the optimal objective function value is output. The capacitive reactive power compensation power of the distributed photovoltaic inverter is adjusted according to the optimal objective function value. The optimization objective function of the improved differential evolution algorithm is composed of the voltage deviation of each node and the active power loss of each branch.

[0010] S4. If any node in the distribution network still has voltage exceeding the limit, then adjust the active power of the distributed photovoltaic system and the capacitive reactive power compensation power of the distributed photovoltaic inverter according to the optimal objective function value.

[0011] The beneficial effects of this invention are as follows: taking the minimum voltage deviation and active power loss as the optimization objective, it integrates methods such as Cauchy perturbation into the differential evolution algorithm to perform synergistic optimization of active power reduction of distributed photovoltaic and capacitive reactive power compensation of inverter. Compared with the basic differential evolution algorithm, it can obtain better results and effectively improve the convergence accuracy and robustness of the algorithm, and can solve the problem of voltage over-limit in distribution networks with high penetration of distributed photovoltaic access. Attached Figure Description

[0012] Figure 1 This is a flowchart of the reactive power optimization method for distribution networks based on the improved differential evolution algorithm disclosed in Embodiment 1 of the present invention;

[0013] Figure 2 This is a detailed flowchart of step S3 disclosed in Embodiment 1 of the present invention;

[0014] Figure 3 This is a detailed flowchart of step S302 disclosed in Embodiment 1 of the present invention;

[0015] Figure 4 This is a schematic diagram of the voltage distribution of each node after reactive power optimization in scenario 1 of embodiment 2 of the present invention;

[0016] Figure 5This is a schematic diagram comparing the optimization values ​​of the objective function after performing 10 reactive power optimizations using two different algorithms in scenario 2 of embodiment 2 of the present invention.

[0017] Figure 6 This is a schematic diagram of the voltage distribution of each node after reactive power optimization in scenario 2 of embodiment 2 of the present invention;

[0018] Figure 7 This is a schematic diagram illustrating the convergence process of the objective function after reactive power optimization using two algorithms in scenario 2 of embodiment 2 of the present invention. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of this invention clearer and more explicit, the content of this invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and not intended to limit it. Furthermore, it should be noted that, for ease of description, only the parts relevant to this invention are shown in the accompanying drawings, not all of them.

[0020] Example 1

[0021] This embodiment proposes a reactive power optimization method for distribution networks based on an improved differential evolution algorithm. Using the minimization of voltage deviation and active power loss as optimization objectives, it incorporates methods such as Cauchy perturbation into the differential evolution algorithm to perform synergistic optimization of active power reduction by distributed photovoltaic (PV) systems and capacitive reactive power compensation by inverters. Compared to the basic differential evolution algorithm, it achieves better results and effectively improves the algorithm's convergence accuracy and robustness, thus addressing the voltage limit exceedance problem in distribution networks with high-penetration distributed PV systems. Figure 1 As shown, it includes the following steps S1-S4:

[0022] S1 obtains the electrical parameters of the distribution network lines, load information, and distributed photovoltaic information.

[0023] In this embodiment, the line electrical parameters include: the impedance and ground admittance of each branch, the capacity and ground impedance of the reactive power compensation equipment at each node, and the network voltage reference value U. B and network power reference value S B Impedance, admittance, and reactive power compensation equipment parameters are used for power flow calculations to obtain network voltage distribution; reference values ​​are used for per-unit value standardization of network parameters.

[0024] S2 calculates the power flow of the distribution network based on the line electrical parameters, load information, and distributed photovoltaic information, and obtains the voltage values ​​of each node and the active power loss of each branch under steady state.

[0025] In this embodiment, the Newton-Raphson method is used for power flow calculation in the distribution network. The power flow calculation is performed using per-unit values, with the upper and lower voltage limits set to 1.05 pu and 0.95 pu, respectively.

[0026] S3. If any node in the distribution network experiences voltage exceedance, the voltage values ​​of each node and the active power loss of each branch are input into an improved Differential Evolution (MDE) algorithm that incorporates Cauchy perturbation operation and central de-crossing operation. The algorithm solves this problem and outputs the optimal objective function value. Based on this optimal objective function value, the capacitive reactive power compensation power of the distributed photovoltaic inverter is adjusted, utilizing the capacitive reactive power compensation capability of the distributed photovoltaic inverter for voltage control. The optimization objective function of the improved differential evolution algorithm is composed of the voltage deviation of each node and the active power loss of each branch.

[0027] In step S3, the objective function is optimized as follows:

[0028] min F=α1F1+α2F2

[0029]

[0030]

[0031] Among them, v i Let i be the voltage value of the distribution network node. Let α1 and α2 represent the active power loss of distribution network branch ij, and α1 and α2 be preset weighting coefficients, predetermined according to actual conditions. It should be noted that, in the distribution network reactive power optimization method based on the improved differential evolution algorithm described in this invention, i and j represent specific distribution network node numbers.

[0032] More specifically, in step S3, such as Figure 2 As shown, the improved differential evolution algorithm includes S301-S307:

[0033] S301, Initialize parameters and solution space X. The parameters include the voltage value of each node and the active power loss of each branch. The solution space X includes NP solutions (i.e., the size of the differential evolution population).

[0034] S302, the solution space X is divided into the optimal solution set bX and the inferior solution set wX, and the Cauchy perturbation operation is performed on the optimal solution set bX to generate the Cauchy perturbation population;

[0035] In step S302, as Figure 3 As shown, the Cauchy perturbation operations include S30201-S30203:

[0036] S30201, Arrange the solutions in the solution space X in descending order of quality, select the best bNP solutions, and form the optimal solution set bX;

[0037] S30202, Cauchy perturbations are applied to the optimal bNP solutions to generate a Cauchy perturbation population dX for the optimal solution set bX, where,

[0038] The Cauchy distribution function can be expressed as:

[0039]

[0040] The expression for the Cauchy perturbation operator is:

[0041] dX=bX+etatan(π*(rand(bNP,n)-0.5))

[0042] In the formula, dX is the Cauchy perturbation population, bX is the optimal bNP solutions, η is the Cauchy perturbation coefficient, and n is the dimension of the solution space X, which should be understood as the number of nodes in the distribution network in this invention, i.e., n is the number of nodes in the distribution network.

[0043] The expression for the Cauchy perturbation coefficient is:

[0044] η = 0.005*(gen / G+δ)

[0045] In the formula, gen is the current iteration number, G is the maximum iteration number, and δ is the adjustment parameter;

[0046] The expression for the adjustment parameter is:

[0047]

[0048] In the formula, F worst For the worst objective function value, F best For the optimal objective function value, F mean ζ is the average value of the objective function, and ζ is a small perturbation parameter. The purpose of the small perturbation parameter is to avoid the denominator of δ being equal to 0. Its value is predetermined by the experimenter. Finally, the Cauchy perturbation population dX for the optimal solution set is generated.

[0049] S30203, Generate a Cauchy variant population cX based on the Cauchy perturbation population dX. The expression for the Cauchy variant population cX is:

[0050] cX=(X-bX)∪dX∪wX (-1) That is, in the solution space, the optimal solution set bX is replaced with the Cauchy perturbation population dX, and then merged with the inferior solution set generated in the previous iteration to form a population.

[0051] S303, Based on Cauchy perturbation and low-probability perturbation, perform mutation operation on the Cauchy perturbation population to generate a mutated population;

[0052] In step S303, the expression for the mutation operation is:

[0053] m i =X i +M(X best -X i +X r1 -cX r2 )

[0054] In the formula, m i X is the variable, M is the coefficient of variation, which is set by the experimenter, and X is the variable of variation. best For the optimal solution in the solution space X or (X-bX)∪dX, r1 and r2 are different integers different from i. Thus, the mutation is related to the original solution set X, as well as the Cauchy perturbation population and the inferior solutions eliminated in the previous round. It not only searches the region near the original optimal solution set, but also re-examines the inferior solutions that have been eliminated locally. This can, to some extent, prevent the differential evolution process from getting trapped in local optima too early.

[0055] It should be noted that the coefficient of variation M is generally taken as [0,2]. Its function is to determine the amplification ratio of the deviation vector. If M is too small, it may cause premature convergence of the algorithm; if M is too large, it will lead to poor convergence of the algorithm. In this embodiment of the invention, M is adaptively adjusted and taken as M = M0 * 2. λ ,in Thus, as the iteration process proceeds, the value of M gradually approaches M0 from 2M0. That is, in the early stage of the iteration, M is larger, which is conducive to the diversified development of the population and to searching for the best solution. In the later stage of the iteration, M decreases, which is conducive to preserving the good information of the population and avoiding destroying the optimal solution.

[0056] Meanwhile, considering low-probability perturbations, the mutation process in this embodiment is changed to: if rand < Pr, then mutation is performed according to the strategy above; otherwise, m i =X L +rand*(X U -X L ), where X U X L These are the upper and lower bounds of the solution, respectively. Here, Pr is a constant greater than 0.9 in the interval (0,1).

[0057] S304, based on the central solution, performs a crossover operation on the Cauchy perturbation population and the variant population to generate an evolutionary population;

[0058] In step S304, the expression for the crossover operation is:

[0059]

[0060] In the formula, ui X is the crossover variable. mean Let CR be the average value of the solution space X or (X-bX)∪dX, and let CR be the cross coefficient.

[0061] S305 performs boundary condition processing on solutions in the evolutionary population to prevent solutions from going out of bounds;

[0062] S306, In the natural selection process, the superior solutions in the evolutionary population replace the inferior solutions, and the eliminated inferior solutions are stored in the inferior solution set wX;

[0063] S307, determine whether the solution in the evolutionary population has converged. If it has not converged, return to step S302. If it has converged or the maximum number of iterations has been reached, terminate the iteration and output the optimal objective function value.

[0064] According to the "Technical Regulations for Photovoltaic Power Plant Grid Connection of State Grid Corporation", in this embodiment of the invention, the output power factor threshold of the distributed photovoltaic inverter is set to 0.95 (leading or lagging); and the range of active power reduction of distributed photovoltaic is 0 (no reduction) to 1 (full reduction).

[0065] S4. If any node in the distribution network still has voltage exceeding the limit, the active power of distributed photovoltaic power generation is reduced simultaneously according to the optimal objective function value, and the capacitive reactive power compensation power of the distributed photovoltaic inverter is also adjusted. The MDE is used to perform coordinated optimization of distributed photovoltaic active power reduction and inverter capacitive reactive power compensation to control the node voltage within a reasonable range.

[0066] In step S4, since there is a synergistic relationship between photovoltaic active power reduction and inverter capacitive reactive power compensation, the population size of the improved differential evolution algorithm should be at least n. 2 In step S3, the population size of the improved differential evolution algorithm is at least n. It should be noted that, to ensure sufficient diversity in the differential evolution population and thus accurate iteration results, the population size should have a lower threshold; in this embodiment, NP ≥ 100.

[0067] Example 2

[0068] The following study takes the IEEE 33-bus distribution system as the research object, sets up two scenarios to verify the reactive power optimization method of the distribution network based on the improved differential evolution algorithm proposed in the first embodiment above, and compares the performance difference between the improved algorithm and the basic algorithm.

[0069] Scenario 1: Low-penetration distributed photovoltaic power is connected to the distribution network. Before reactive power optimization, only a few nodes in the distribution network experience slight voltage overruns.

[0070] Scenario 2: High-penetration distributed photovoltaic power is connected to the distribution network. Before reactive power optimization, most nodes in the distribution network experienced severe voltage overruns.

[0071] The following assumptions are made for the above scenarios: all distributed photovoltaic systems are operating at full capacity before reactive power optimization. Furthermore, to obtain a sufficient number of samples and ensure the accuracy of the results, each scenario is processed 10 times independently using two algorithms. It should also be noted that if the objective function value no longer changes or changes less than a set threshold during the iteration process, the iteration is considered to have converged. To make this judgment more accurate and ensure that the algorithm iteration does not converge further, the objective function change must be less than the threshold for 100 consecutive iterations before convergence can be considered achieved and the algorithm iteration can be terminated.

[0072] In Scenario 1, the distributed photovoltaic penetration rate is low, and the number of nodes exceeding voltage limits is small. Only inverter reactive power compensation is needed to ensure the voltage of each node is within acceptable limits; therefore, active power reduction is not performed. The entire process involves few variables, has a small solution space, requires few algorithm iterations, and has low time complexity. The results obtained by the two different algorithms in this scenario are shown in Table 1.

[0073] Table 1 Comparison of Calculation Results of Two Algorithms in Scenario 1

[0074]

[0075]

[0076] As shown in the table, both algorithms achieved the same optimal result in 10 calculations. The number of iterations showed no clear pattern, which is due to the randomness of the differential evolution algorithm. Furthermore, because the improved algorithm incorporates Cauchy perturbations, it takes slightly longer than the basic algorithm under similar iteration counts.

[0077] In scenario 1, voltage regulation needs can be met by only performing reactive power capacity compensation of the inverter, without performing active power reduction. Therefore, there are fewer variables involved, the solution space is small, the calculation speed is fast, and the accuracy is high. Regardless of whether the algorithm is improved, the optimal solution can be obtained.

[0078] The result of the 6th iteration, which has the fastest iteration speed, is used to verify the effectiveness of the algorithm in reactive power optimization. The voltages at each node are shown below. Figure 4 As shown in the figure, before reactive power optimization, some nodes experienced voltage exceeding the upper limit; after optimization, the voltage of all nodes met the requirements. The optimization curves obtained by the two algorithms basically overlapped.

[0079] In Scenario 2, due to the large-scale integration of distributed photovoltaic systems, there are many nodes exceeding voltage limits. Even with maximum utilization of inverter reactive power capacity for voltage regulation, it is impossible to bring the entire grid voltage back to the acceptable range. Therefore, active power reduction is necessary, involving more parameters than in Scenario 1, and these parameters have interrelationships. The calculation process is lengthy, and the number of algorithm iterations and time complexity increase significantly compared to Scenario 1. The results obtained by the two different algorithms are shown in Table 2.

[0080] Table 2 Comparison of Calculation Results of Two Algorithms in Scenario 2

[0081]

[0082] The line graph of the results of 10 calculations is as follows Figure 5 As shown in the figure, the calculation results of the basic algorithm are unstable and exhibit some fluctuations, while the improved algorithm not only yields better results but also shows stable results across multiple calculations, almost forming a straight line. In 10 calculations, the overall standard deviation of the objective function value of the basic algorithm is σ1 = 2.11 × 10⁻⁶. -4 The standard deviation of the improved algorithm is σ² = 2.99 × 10⁻⁶. -6 The values ​​σ2σ1 indicate that the improved algorithm exhibits significant stability, with similar results across multiple calculations, while the basic algorithm shows greater fluctuations, is prone to premature convergence and getting trapped in local optima, and struggles to find the optimal objective function value. However, due to operations such as Cauchy perturbation, the improved algorithm often takes longer.

[0083] The result of the 6th iteration is used to plot and verify the application effect of the improved differential evolution algorithm in reactive power optimization of the distribution network in this embodiment. The voltage of each node is as follows: Figure 6 As shown in the figure, before reactive power optimization, the voltage of most nodes was severely out of range. Even after using the inverter's reactive power capacity for compensation, it was still impossible to bring all node voltages back to the acceptable range. Only after active power reduction could the voltage of each node meet the requirements. The convergence process of the improved algorithm is as follows: Figure 7 As shown, by Figure 7 It can be seen that during the algorithm iteration process, only the reactive capacity of the inverter is compensated first. When the algorithm determines that the reactive capacity alone cannot meet the optimization target, active power is reduced. Therefore, the iteration curve shows a relatively large numerical change in the middle section.

[0084] In summary, when the solution space is large, there are many variables involved, or there is a certain synergistic relationship between the variables, the improved differential evolution algorithm described in this invention has significant algorithm robustness and can find better optimization results. After optimization, the voltage deviation and network loss in the distribution network are smaller.

[0085] The above embodiments are merely illustrative of the technical concept and features of the present invention, and are intended to enable those skilled in the art to understand the content of the present invention and implement it accordingly. They should not be construed as limiting the scope of protection of the present invention. All equivalent changes or modifications made based on the essence of the content of the present invention should be covered within the scope of protection of the present invention.

Claims

1. A reactive power optimization method for distribution networks based on an improved differential evolution algorithm, characterized in that, Includes the following steps: S1, obtains the electrical parameters of the distribution network lines, load information, and distributed photovoltaic information; S2, perform power flow calculation of the distribution network based on the line electrical parameters, the load information and the distributed photovoltaic information to obtain the voltage value of each node and the active power loss of each branch under steady state; S3, if any node in the distribution network experiences voltage exceedance, the voltage values ​​of each node and the active power loss of each branch are incorporated into an improved differential evolution algorithm that combines Cauchy perturbation operation and central solution crossover operation for solution, and the optimal objective function value is output. The capacitive reactive power compensation power of the distributed photovoltaic inverter is adjusted according to the optimal objective function value. The optimization objective function of the improved differential evolution algorithm is composed of the voltage deviation of each node and the active power loss of each branch. In step S3, the improved differential evolution algorithm includes: S301, initializing parameters and solution space. The parameters include the voltage values ​​of each node and the active power loss of each branch, and the solution space. include A solution; S302, the solution space Divided into the optimal solution set and inferior solution set and for the optimal solution set Perform a Cauchy perturbation operation to generate a Cauchy perturbation population; S303, perform a mutation operation on the Cauchy perturbation population based on the Cauchy perturbation and the low-probability perturbation to generate a mutated population; S304, Perform a crossover operation on the Cauchy perturbation population and the mutant population based on the central solution to generate an evolutionary population; S305, Perform boundary condition processing on the solutions in the evolutionary population; S306, During the natural selection operation, replace inferior solutions with superior solutions in the evolutionary population, and store the eliminated inferior solutions in the inferior solution set. S307, determine whether the solution in the evolutionary population has converged. If it has not converged, return to step S302. If it has converged or the maximum number of iterations has been reached, terminate the iteration and output the optimal objective function value. S4. If any node in the distribution network still has voltage exceeding the limit, then adjust the active power of the distributed photovoltaic system and the capacitive reactive power compensation power of the distributed photovoltaic inverter according to the optimal objective function value.

2. The reactive power optimization method for distribution networks based on the improved differential evolution algorithm as described in claim 1, characterized in that, The electrical parameters of the line include: the impedance and ground admittance of each branch, the capacity and ground impedance of the reactive power compensation equipment at each node, and the network voltage reference value. and network power reference value .

3. The reactive power optimization method for distribution networks based on the improved differential evolution algorithm as described in claim 1, characterized in that, The power flow calculation of the distribution network adopts the Newton-Raphson method.

4. The reactive power optimization method for distribution networks based on the improved differential evolution algorithm as described in claim 1, characterized in that, The power flow calculation of the distribution network is performed using per-unit values, with the upper and lower voltage limits set at 1.05 pu and 0.95 pu, respectively.

5. The reactive power optimization method for distribution networks based on the improved differential evolution algorithm as described in claim 1, characterized in that, In step S3, the optimization objective function is: in, For distribution network nodes voltage value, For distribution network branches Active power loss, and These are the preset weighting coefficients.

6. The reactive power optimization method for distribution networks based on the improved differential evolution algorithm as described in claim 1, characterized in that, In step S302, the Cauchy perturbation operation includes: S30201, the solution space The solutions are arranged from best to worst, and the optimal one is selected. The solutions form the optimal solution set. ; S30202, for the optimal Each solution is subjected to Cauchy perturbation, generating a set of optimal solutions. Cauchy disturbance population ,in, The expression for the Cauchy perturbation operator is: In the formula, For Cauchy disturbance populations, For optimal One solution. The Cauchy perturbation coefficient is... For solution space dimensionality; The expression for the Cauchy perturbation coefficient is: In the formula, This represents the current iteration number. The maximum number of iterations, To adjust the parameters; The expression for the adjustment parameter is: In the formula, The worst objective function value, The optimal objective function value is... The average value of the objective function. For small perturbation parameters; S30203, based on the aforementioned Cauchy perturbation population Generate Cauchy variant population The Cauchy variant population The expression is: 。 7. The reactive power optimization method for distribution networks based on the improved differential evolution algorithm as described in claim 6, characterized in that, In step S303, the expression for the mutation operation is: In the formula, As a variable, The coefficient of variation is 1. For solution space or The optimal solution in and For different Different integers.

8. The reactive power optimization method for distribution networks based on the improved differential evolution algorithm as described in claim 7, characterized in that, In step S304, the expression for the crossover operation is: In the formula, For cross variables, For solution space or The average value, This is the cross coefficient.

9. The reactive power optimization method for distribution networks based on the improved differential evolution algorithm according to claim 1, characterized in that: If the number of variables in the distribution network is In step S4, considering the synergistic relationship between inverter capacitive reactive power compensation and photovoltaic active power reduction in the input parameters, the population size of the improved differential evolution algorithm is at least [value missing]. In step S3, only the inverter's capacitive reactive power compensation needs to be considered, and the population size of the improved differential evolution algorithm should be at least [value missing]. .

Images