Active power distribution network adaptive voltage treatment method based on improved wave search algorithm

Abstract

The invention discloses an active power distribution network adaptive voltage treatment method based on an improved wave search algorithm, and the method comprises the steps: firstly, adding scene parameter simulation on the basis of a power system model, and constructing a random scene power flow model; secondly, the algorithm model is improved through Logistic-Tent chaotic mapping and a differential variation strategy; thirdly, the initial solution set is preferentially selected, and differential variation and adaptive convergence factor optimization are carried out on optimal particles to obtain individuals with higher fitness; and finally, comparing whether the solution set can meet the constraint condition of an actual problem, and re-initializing the solution set which does not meet the constraint condition to form an optimal configuration scheme. The method provided by the invention has the characteristics of strong distributed search and global search capabilities, can adapt to the dynamic change of the active power distribution network, and realizes the real-time and accurate regulation and control of the voltage.

Description

Technical Field

[0001] The present invention belongs to the technical field of power system planning methods, and in particular relates to an active distribution network adaptive voltage management method based on an improved wave search algorithm. Background Art

[0002] As the power system transforms towards intelligence and sustainability, new power systems with large-scale distributed power sources connected to the grid have attracted more and more attention. Active distribution networks have achieved efficient utilization and management of distributed energy by integrating advanced power electronic devices and communication technologies. However, when a large number of distributed energy sources are included in the distribution network system, voltage fluctuations and imbalances become more prominent, sometimes even damaging the insulation of equipment.

[0003] Existing voltage management methods are often based on expected grid operating conditions and load levels, using static voltage control equipment, which lacks the ability to respond to dynamic changes and is unable to effectively deal with the rapid access and exit of distributed energy in active distribution networks. Summary of the invention

[0004] The purpose of the present invention is to provide an adaptive voltage management method for an active distribution network based on an improved wave search algorithm (Improvement wave search algorithm, IWSA), which has the characteristics of strong distributed search and global search capabilities, has certain advantages in dealing with population particle over-limit problems, can adapt to the dynamic changes of the active distribution network, and realize real-time and precise voltage regulation.

[0005] The technical solution adopted by the present invention is an active distribution network adaptive voltage management method based on an improved wave search algorithm, comprising the following steps:

[0006] Step 1: Based on the power system model combining reactive power optimization and distribution network reconstruction, scenario parameter simulation is introduced into the reactive power optimization module to construct a random scenario power flow model;

[0007] Step 2: Improve the wave search algorithm used in the random scenario power flow model by Logistic-Tent chaotic mapping, use the chaotic sequence generated by Logistic-Tent chaotic mapping to initialize the population, and obtain the initial solution set;

[0008] Step 3: Optimize the initial solution set to obtain the optimal individual, improve the adaptive convergence factor and perform differential mutation perturbation on the optimal individual, and retain the optimal solution with the highest population fitness;

[0009] Step 4: The set of optimal solutions with the highest fitness obtained in step 3 is subjected to constraint condition judgment. After the iteration termination condition is met, the results output from step 3 are statistically analyzed to obtain the optimal reactive capacity compensation configuration and network loss solution for the system.

[0010] The present invention is also characterized in that:

[0011] The random scenario power flow model in step 1 is specifically:

[0012]

[0013] Among them, F(·) is the objective function of the upper model; N(·) is the objective function of the lower model; G(·) represents the upper constraint function; g(·) represents the lower constraint function; Ks is the coefficient under different scenarios.

[0014] The conditions for scenario parameter simulation include system load demand and changes in meteorological conditions.

[0015] In the random scenario power flow model, the voltage, branch current and transmission power of each node in the distribution network are constrained.

[0016] Step 2 specifically includes the following steps:

[0017] Step 2.1: Initialize the parameters of the random scenario power flow model and generate the initial electromagnetic wave particle positions;

[0018] Step 2.2: Use the Logistic-Tent mapping function to generate the chaotic sequence of the Logistic-Tent mixed mapping of formula (6), and use formula (6) to update the initial electromagnetic wave particle position;

[0019]

[0020] Among them, r1 is the parameter of Logistic mapping; r2 is the control parameter for adjusting the nonlinearity of the system, and the range of r2 is (0,4);

[0021] Step 2.3: Substitute the electromagnetic wave particle position population updated in step 2.2 into the power flow calculation equation, and use the Newton-Raphson iteration method to perform iterative calculation to obtain the fitness value of each electromagnetic wave particle;

[0022] Step 2.4: Sort the set of electromagnetic wave particle fitness values ​​obtained in step 2.3 according to the fitness value size, and use formula (8) to gradually narrow the search range to generate a population In the process of gradually narrowing the search range, the improved greedy mechanism of formula (9) is used to control the population near the global optimal position;

[0023] W1 new =Wmin +r3(W max -W min ) (8);

[0024]

[0025] Among them, W min is a vector composed of the minimum values ​​of each dimension of W, W max is a vector of the maximum values ​​of each dimension of W, f mean is the average fitness value of all particles, and r3 is a random number between 0 and 1.

[0026] Step 3 specifically includes the following steps:

[0027] Step 3.1: Use formula (10) to calculate the population Update and generate population And by improving the greedy mechanism, Location not inferior to Location, population Perform boundary processing to obtain the processed population

[0028]

[0029] in, t is the current iteration number, T is the total iteration number, σ is the waveform size control coefficient, m is the column vector of the population particles that obey the normal distribution and are arranged in order, W l For W press and W best The position matrix after rearrangement of the proximity;

[0030] Step 3.2: Population Perform correction processing to obtain the optimal individual;

[0031] Step 3.3: Improve the adaptive convergence factor and perform differential mutation perturbation on the optimal individual in step 3.2 to obtain the optimal solution with the highest fitness;

[0032] Step 3.4: Use the central difference method to optimize the objective function of the reactive power optimization upper model, and use the optimized objective function in the next round of iteration.

[0033] The specific strategy of adaptive convergence factor improvement and differential mutation perturbation in step 3.3 is to introduce two strategies, adaptive step size and differential mutation, to enhance the performance of the algorithm. The adaptive step size adjustment formula is:

[0034]

[0035] Wherein, α2 is the step size coefficient, and the initial value of α2 is set to α0=0.4; t is the current number of iterations; T is the total number of iterations;

[0036] Differential mutation is to use formula (14) and formula (15) to find the best individual again. The individual generation formula is:

[0037]

[0038] Among them, G is the convergence criterion factor; are different individuals in the population.

[0039] The beneficial effects of the present invention are:

[0040] 1) The method of the present invention adopts the IWSA algorithm which combines deterministic and uncertain improvement strategies, so that the iterative process can obtain the optimal solution with fewer iterations within the voltage allowable range, thereby avoiding the high computational cost and running memory generated by the existing WSA algorithm; at the same time, the Logistic-Tent chaotic mapping strategy is introduced to re-evaluate the fitness ranking of the initial population, thereby improving the efficiency and accuracy of the algorithm and enhancing the randomness of the population;

[0041] 2) The IWSA algorithm randomly crosses the elements of different solution sets, and uses an adaptive convergence factor to adjust the intensity and scope of the crossover operation during the crossover process to improve the ability to mine the best individual. It also uses a differential mutation strategy to iterate the optimization of the population optimization individuals, which improves the global search ability of the algorithm and avoids falling into the local optimum.

[0042] 3) The method of the present invention optimizes the static parameters given in voltage management into a dynamic adaptive optimization strategy, and establishes the connection and data drive between the control quantity and the state quantity, thereby better solving the voltage over-limit problem in the grid-connected operation of distributed power sources. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 It is a flow chart of specific implementation steps of the method of the present invention. DETAILED DESCRIPTION

[0044] The present invention is described in detail below with reference to the accompanying drawings and specific embodiments.

[0045] The invention provides an adaptive voltage management method for active distribution network based on improved wave search algorithm, such as Figure 1 As shown, the specific steps include:

[0046] Step 1: Based on the power system model combining reactive power optimization and distribution network reconstruction, scenario parameter simulation is introduced into the reactive power optimization module to construct a random scenario power flow model;

[0047] Based on the improved power system model combining reactive power optimization with distribution network reconstruction, scenario parameter simulation is added to reactive power optimization. By considering the uncertainty of load demand and meteorological conditions, random scenarios for analysis are generated, and scenario parameter variables representing different environmental significance are introduced into reactive power optimization calculations. In the upper-level module of reactive power optimization, the switch operation will be transmitted from the reconstruction module, and there is no need to consider the network radial constraints and switch operation-related constraints. Based on the results of the reactive power optimization module, the lower-level reconstruction module takes the distribution network topology as the optimization object, determines the output of distributed power sources and the gear position of reactive power compensation equipment according to the reactive power optimization module, and optimizes network losses through network reconstruction. The random scenario power flow model is specifically as follows:

[0048]

[0049] Among them, F(·) is the objective function of the upper model; N(·) is the objective function of the lower model; G(·) and g(·) represent the constraint functions of the upper and lower layers respectively; Ks is the coefficient under different scenarios.

[0050] On the basis of the above-mentioned random scenario power flow model, the safe and stable operation of the system is considered during operation, and constraints are imposed on the voltage, branch current and transmission power of each node in the distribution network.

[0051] Step 2: Improve the wave search algorithm used in the random scenario power flow model by Logistic-Tent chaotic mapping, use the chaotic sequence generated by Logistic-Tent chaotic mapping to initialize the population, and obtain the initial solution set;

[0052] The improved wave search algorithm IWSA receives the current distribution network node voltage, active and reactive power parameters as input, and outputs the fitness value distribution of each device in the distribution network; in each iteration, the electromagnetic wave particles collect experience through interaction with the environment to evaluate the current particle state. Therefore, the process of solving the adaptive voltage management problem of active distribution network based on the IWSA algorithm can be understood as the process of training the WSA algorithm, where step 2 specifically includes the following steps:

[0053] Step 2.1: Initialize the parameters of the random scenario power flow model to generate the initial electromagnetic wave particle position W; the initialization algorithm requires a series of parameter settings. Assume that the number of particles is n, the dimension of the problem to be optimized is d, and the position of the electromagnetic wave particle W is simulated by a matrix. Use f([W n1 ,W n2 , ···W nd ]) represents the fitness value of the nth individual, and the fitness value of the population can be represented by the following vector W. Finally, prepare n random numbers k1, k2...K nInitialize the particle position from 0 to 1. Where lb and ub are the upper and lower boundaries of the search space, that is, the limits of the node configuration reactive capacity. Use formula (2) to randomly generate the initial reactive compensation capacity x of the active distribution network node i i , and the electromagnetic wave particle population is simulated by the matrix W in formula (4); where W nd represents the nth particle in the dth dimension.

[0054]

[0055] W i =lb+x * (ub-lb) (3);

[0056]

[0057] Step 2.2: Improve the wave search algorithm with Logistic-Tent chaotic mapping. Use the Logistic-Tent mapping function in formula (5) to generate a chaotic sequence of Logistic-Tent mixed mapping. In the process of updating the position of the electromagnetic wave, use the Logistic-Tent chaotic sequence in formula (6) to update the position of the electromagnetic wave particles, that is, the voltage deviation and network loss of the distribution network.

[0058]

[0059] Among them, x n is the current state; r1 is the parameter of the Logistic mapping; μ is the parameter of the Tent mapping; r2 is the control parameter for adjusting the degree of nonlinearity of the system, and its range is (0,4).

[0060] Step 2.3: Substitute the electromagnetic wave particle population W1 whose position has been updated in step 2.2 into the power flow calculation equation of formula (7). The power flow calculation equation optimizes the equation model according to the updated electromagnetic wave particle position parameters, and uses the Newton-Raphson iteration method to perform iterative calculations. The fitness value of each electromagnetic wave particle is calculated according to formula (7);

[0061] Obj(F min (W1)) = α1F V +β1F PL (7);

[0062] Among them, α1 and β1 represent weight coefficients; F v is the target function related to voltage; F PL represents the objective function related to power loss.

[0063] Step 2.4: Arrange the set of electromagnetic wave particle fitness values ​​obtained in step 2.3 from small to large according to the obtained fitness values ​​to form a new particle group after arrangement, and record the electromagnetic wave particle with the smallest fitness value as W min ; Use formula (8) to gradually narrow the search range and generate new population individuals The search efficiency is improved; and the greedy mechanism is improved by using formula (9) to adjust the search space in the process of narrowing the search range, and control the position of the population near the global optimal position.

[0064] W1 new =W min +r3(W max -W min ) (8);

[0065]

[0066] Among them, W min is a vector composed of the minimum values ​​of each dimension of W, W max is a vector of the maximum values ​​of each dimension of W, f mean is the average fitness value of all particles, and r3 is a random number between 0 and 1.

[0067] Step 3: Optimize the initial solution set, perform global development and local survey phases at the same time, and retain the optimal solution of this generation to obtain the optimal individual. Perform adaptive convergence factor improvement and differential variation perturbation on the optimal individual, use the decision coefficient to select between the current individual position and the optimal individual position of the current iteration, retain the optimal solution with the highest population fitness, and ensure that the iterative individual voltage at the current position is within the limit. Step 3 specifically includes the following steps:

[0068] Step 3.1: Use formula (10) to calculate the population Update, that is, update the voltage deviation of the nodes in each electromagnetic wave particle to generate a population And with the help of the improved greedy mechanism of formula (9), the updated group position is not inferior to the current group position.

[0069]

[0070] in, t is the current iteration number, T is the total iteration number, σ is the waveform size control coefficient, m is the column vector of the population particles that obey the normal distribution and are arranged in order, W l For W press and W best The position matrix after rearrangement of the proximity.

[0071] Then the newly generated electromagnetic wave particles Boundary processing is performed and equation (11) is used to simulate the situation where some particles with lower fitness values ​​exceed the reactive power compensation capacity limit and move to W best Reflection, while particles with higher fitness values ​​stay away from W best Continue to spread outward.

[0072]

[0073] Where, β2=0.75+ei / n w2 , β2 is the reflection intensity coefficient, and r3 is a random value between 0 and 1. fi It is the position matrix after W is rearranged in order from small to large fitness values.

[0074] Step 3.2: In order to correctly predict the electromagnetic wave particles near the capacity limit, the obtained population needs to be Correction and processing are performed to make the particle group search in the current optimal direction. At the same time, there is a certain probability that it will deflect in the direction of W. Here, a periodic factor and a standard normal distribution random number are introduced to strengthen the individual's more extensive exploration in the solution space, and the particle is updated according to formula (12) to obtain the optimal individual

[0075]

[0076] Among them, δ is the receiving coefficient, δ = 0.6 + (1.2-0.5) sin (tπ / 2T), η is a random number that obeys the normal distribution, n w3 is the number of particles that simulate receiving electromagnetic waves. best * is the historical optimal position obtained by convolution.

[0077] Step 3.3: Improve the adaptive convergence factor and perform differential mutation perturbation on the optimal individuals obtained, and retain the optimal solution with higher population fitness;

[0078] In order to improve the wave search algorithm and combine it with the voltage and reactive power optimization of the distribution network, the performance of the algorithm can be enhanced by introducing two strategies: adaptive step size and differential mutation. The adaptive step size strategy dynamically adjusts the step size according to the progress of the current iteration, so that the algorithm has a larger search range in the early stage and searches the local area more finely in the later stage. The step size adjustment formula can be shown as formula (13):

[0079]

[0080] Among them, α2 is the step size coefficient, and the initial value of α2 is set to α0=0.4; t is the current number of iterations; and T is the total number of iterations.

[0081] Formulas (14) and (15) use the differential mutation strategy to search for the best individual again. That is, a differential operation is performed on the basis of the current individual, and a random disturbance is added to generate a new individual to avoid falling into the local optimal solution too early. The specific steps of differential mutation are: select at least 3 individuals, that is, reactive compensation capacity, and then calculate the difference of the selected individuals. Then add the difference result to a random vector to generate a new individual. The generation formula of the new optimal individual obtained by optimizing the wave search algorithm with the differential mutation strategy is:

[0082]

[0083] Among them, G is the convergence criterion factor; are different individuals in the population.

[0084] Step 3.4: Use the central difference method to optimize the objective function of the reactive power optimization upper model, and use the optimized objective function in the next round of iteration.

[0085] Formula (16) uses the central difference method to perform reverse difference gradient fitting on the objective function of the upper module of reactive power optimization to search for the optimal solution to improve the search efficiency and accuracy. Formulas (17) and (18) randomly set the particles beyond the boundary within the search range by selecting appropriate finite difference values ​​and combining random search and gradient guidance to improve the optimization direction of the algorithm.

[0086] g i =(f(W +∈i )-f(W -∈i )) / 2∈ (16);

[0087] W' best =W best -α3×g i (17);

[0088] W” best =lb+r4×(ub-lb)ifW' best >uborW' best <lb (18);

[0089] Among them, g i is the gradient; α3 is the scaling factor. lb and ub are the lower and upper boundaries of the search range respectively; r4 is a d-dimensional randomized column vector whose element values ​​range from 0 to 1.

[0090] Step 4: The set of optimal solutions with the highest fitness obtained in step 3 is subjected to constraint condition judgment. After the iteration termination condition is met, the results output from step 3 are statistically analyzed to obtain the optimal reactive capacity compensation configuration and network loss solution for the system.

[0091] The method for adaptive voltage management of active distribution network based on improved wave search algorithm provided by the present invention adopts IWSA algorithm which combines deterministic and uncertain improvement strategies, so that the iterative process can obtain the optimal solution with fewer iterations within the allowable voltage range, improve the efficiency and accuracy of the algorithm, enhance the randomness of the population, and thus better solve the voltage over-limit problem in the grid-connected operation of distributed power sources.

[0092] Example 1

[0093] The active distribution network adaptive voltage management method based on the improved wave search algorithm provided in this embodiment 1 includes the following steps:

[0094] Step 1: Based on the power system model combining reactive power optimization and distribution network reconstruction, scenario parameter simulation is introduced in the reactive power optimization module to construct a random scenario power flow model. The conditions of scenario parameter simulation include system load demand and meteorological condition changes. The specific random scenario power flow model is as follows:

[0095]

[0096] Among them, F(·) is the objective function of the upper model; N(·) is the objective function of the lower model; G(·) represents the upper constraint function; g(·) represents the lower constraint function; Ks is the coefficient under different scenarios.

[0097] In the random scenario power flow model, the voltage, branch current and transmission power of each node in the distribution network are constrained.

[0098] Step 2: Improve the wave search algorithm used in the random scenario power flow model by Logistic-Tent chaotic mapping, use the chaotic sequence generated by Logistic-Tent chaotic mapping to initialize the population, and obtain the initial solution set;

[0099] Step 3: Optimize the initial solution set to obtain the optimal individual, improve the adaptive convergence factor and perform differential mutation perturbation on the optimal individual, and retain the optimal solution with the highest population fitness;

[0100] Step 4: The set of optimal solutions with the highest fitness obtained in step 3 is subjected to constraint condition judgment. After the iteration termination condition is met, the results output from step 3 are statistically analyzed to obtain the optimal reactive capacity compensation configuration and network loss solution for the system.

[0101] Example 2

[0102] The active distribution network adaptive voltage management method based on the improved wave search algorithm provided in this embodiment 2 is based on the embodiment 1, and step 2 specifically includes the following steps:

[0103] Step 2.1: Initialize the parameters of the random scenario power flow model and generate the initial electromagnetic wave particle positions;

[0104] Step 2.2: Use the Logistic-Tent mapping function to generate the chaotic sequence of the Logistic-Tent mixed mapping of formula (6), and use formula (6) to update the initial electromagnetic wave particle position;

[0105]

[0106] Among them, r1 is the parameter of Logistic mapping; r2 is the control parameter for adjusting the nonlinearity of the system, and the range of r2 is (0,4);

[0107] Step 2.3: Substitute the electromagnetic wave particle position population updated in step 2.2 into the power flow calculation equation, and use the Newton-Raphson iteration method to perform iterative calculation to obtain the fitness value of each electromagnetic wave particle;

[0108] Step 2.4: Sort the set of electromagnetic wave particle fitness values ​​obtained in step 2.3 according to the fitness value size, and use formula (8) to gradually narrow the search range to generate a population In the process of gradually narrowing the search range, the improved greedy mechanism of formula (9) is used to control the population near the global optimal position;

[0109] W1 new =W min +r3(W max -W min ) (8);

[0110]

[0111] Among them, W min is a vector composed of the minimum values ​​of each dimension of W, W max is a vector of the maximum values ​​of each dimension of W, f mean is the average fitness value of all particles, and r3 is a random number between 0 and 1.

[0112] Step 3 specifically includes the following steps:

[0113] Step 3.1: Use formula (10) to calculate the population Update and generate population And by improving the greedy mechanism, Location not inferior to Location, population Perform boundary processing to obtain the processed population

[0114]

[0115] in, t is the current iteration number, T is the total iteration number, σ is the waveform size control coefficient, m is the column vector of the population particles that obey the normal distribution and are arranged in order, W l For W press and W best The position matrix after rearrangement of the proximity;

[0116] Step 3.2: Population Perform correction processing to obtain the optimal individual;

[0117] Step 3.3: Improve the adaptive convergence factor and perform differential mutation perturbation on the optimal individual in step 3.2 to obtain the optimal solution with the highest fitness;

[0118] Step 3.4: Use the central difference method to optimize the objective function of the reactive power optimization upper model, and use the optimized objective function in the next round of iteration.

[0119] Example 3

[0120] The active distribution network adaptive voltage management method based on the improved wave search algorithm provided in this embodiment 3 is based on embodiment 2. The specific strategy of adaptive convergence factor improvement and differential variation disturbance in step 3.3 is to introduce two strategies, adaptive step size and differential variation, to enhance the performance of the algorithm. The adaptive step size adjustment formula is:

[0121]

[0122] Wherein, α2 is the step size coefficient, and the initial value of α2 is set to α0=0.4; t is the current number of iterations; T is the total number of iterations;

[0123] Differential mutation is to use formula (14) and formula (15) to find the best individual again. The individual generation formula is:

[0124]

[0125] Among them, G is the convergence criterion factor; are different individuals in the population.

Claims

1. An adaptive voltage management method for active distribution network based on improved wave search algorithm, characterized in that: The following steps are involved: Step 1: Based on the power system model combining reactive power optimization and distribution network reconstruction, scenario parameter simulation is introduced into the reactive power optimization module to construct a random scenario power flow model; Step 2: Performing Logistic-Tent chaotic mapping improvement on the wave search algorithm used in the random scenario power flow model, initializing the population using the chaotic sequence generated by the Logistic-Tent chaotic mapping, and obtaining an initial solution set; Step 3: Optimize the initial solution set to obtain the optimal individual, perform adaptive convergence factor improvement and differential mutation perturbation on the optimal individual, and retain the optimal solution with the highest population fitness; Step 4: The set of optimal solutions with the highest fitness obtained in step 3 is subjected to constraint condition judgment. After the iteration termination condition is met, the results output from step 3 are statistically analyzed to obtain the optimal reactive capacity compensation configuration and network loss solution for the system.

2. The method for adaptive voltage management of active distribution network based on improved wave search algorithm according to claim 1 is characterized in that: The random scenario power flow model described in step 1 is specifically: Among them, F(·) is the objective function of the upper model; N(·) is the objective function of the lower model; G(·) represents the upper constraint function; g(·) represents the lower constraint function; Ks is the coefficient under different scenarios.

3. The method for adaptive voltage management of active distribution network based on improved wave search algorithm according to claim 2 is characterized in that: The conditions simulated by the scenario parameters include system load requirements and changes in meteorological conditions.

4. The method for adaptive voltage management of active distribution network based on improved wave search algorithm according to claim 2 is characterized in that: In the random scenario power flow model, the voltage of each node, branch current and transmission power of the distribution network are constrained.

5. The method for adaptive voltage management of active distribution network based on improved wave search algorithm according to claim 2 is characterized in that: Step 2 specifically includes the following steps: Step 2.1: Initialize the parameters of the random scenario power flow model to generate the initial electromagnetic wave particle position; Step 2.2: Using the Logistic-Tent mapping function, generate the chaotic sequence of the Logistic-Tent mixed mapping of formula (6), and use formula (6) to update the position of the initial electromagnetic wave particle; Among them, r1 is the parameter of Logistic mapping; r2 is the control parameter for adjusting the nonlinearity of the system, and the range of r2 is (0,4); Step 2.3: Substitute the electromagnetic wave particle position population updated in step 2.2 into the power flow calculation equation, and use the Newton-Raphson iteration method to perform iterative calculation to obtain the fitness value of each electromagnetic wave particle; Step 2.4: Sort the set of electromagnetic wave particle fitness values ​​obtained in step 2.3 according to the fitness value size, and use formula (8) to gradually narrow the search range to generate a population In the process of gradually narrowing the search range, the improved greedy mechanism of formula (9) is used to control the population near the global optimal position; W1 new =In min +r3(W max -IN min ) (8); Among them, W min is a vector composed of the minimum values ​​of each dimension of W, W max is a vector of the maximum values ​​of each dimension of W, f mean is the average fitness value of all particles, and r3 is a random number between 0 and 1.

6. The method for adaptive voltage management of active distribution network based on improved wave search algorithm according to claim 5 is characterized in that: Step 3 specifically includes the following steps: Step 3.1: Use formula (10) to calculate the population Update and generate population And with the help of the improved greedy mechanism, Location not inferior to Location, population Perform boundary processing to obtain the processed population in, t is the current iteration number, T is the total iteration number, σ is the waveform size control coefficient, m is the column vector of the population particles that obey the normal distribution and are arranged in order, W l For W press and W best The position matrix after rearrangement of the proximity; Step 3.2: Population Perform correction processing to obtain the optimal individual; Step 3.3: The optimal individual described in step 3.2 is subjected to adaptive convergence factor improvement and differential mutation perturbation to obtain the optimal solution with the highest fitness; Step 3.4: Use the central difference method to optimize the objective function of the reactive power optimization upper model, and use the optimized objective function in the next round of iteration.

7. The method for adaptive voltage management of active distribution network based on improved wave search algorithm according to claim 6 is characterized in that: The specific strategy of adaptive convergence factor improvement and differential mutation disturbance in step 3.3 is to introduce two strategies, adaptive step size and differential mutation, to enhance the performance of the algorithm. The adaptive step size adjustment formula is: Wherein, α2 is the step size coefficient, and the initial value of α2 is set to α0=0.4; t is the current number of iterations; T is the total number of iterations; The differential mutation specifically uses formula (14) and formula (15) to optimize the optimal individual again, and the individual generation formula is: Among them, G is the convergence criterion factor; are different individuals in the population.

Images