Urban substation site selection optimization method for different power consumption scenarios at micro scale

Abstract

The application provides a city substation site selection optimization method for different power consumption scenarios under micro scale, extracts time sequence power consumption load sample data set under different power consumption scenarios from power data in initial data set; obtains power driving factor text data set by pre-processing point, line and surface multi-source space data; constructs multi-scenario power consumption power fitting model based on random forest method; divides power supply range of each substation based on micro scale power consumption load space distribution obtained by fitting multi-scenario power consumption power fitting model and region growing algorithm; introduces LSCP covering model, establishes substation site selection optimization model, and obtains optimal value of the substation site selection optimization model by using genetic algorithm, that is, optimal substation site selection result is obtained. The beneficial effect of the application is that the substation position under different power consumption scenarios is effectively selected, the city power grid coverage and transformer utilization are improved, and an efficient scheme is provided for city power grid planning.

Description

Methods for optimizing the location of urban substations under different power consumption scenarios at the micro scale Technical Field

[0001] This invention relates to the field of urban substation site selection, and in particular to a method for optimizing urban substation site selection for different power consumption scenarios at a micro scale. Background Technology

[0002] Electricity is an essential resource for human production and daily life, playing a vital role in national economic and social development. To address the power shortage, it is necessary to consider the overall power system, taking into account the spatiotemporal variations of urban power grid load in different seasons, and optimizing the supply structure of power facilities.

[0003] To meet the ever-increasing demand for electricity, the power supply and distribution network needs to consider the spatiotemporal distribution of power load and expand beyond its existing scale. Among these, the site selection of distribution substations is the core of the distribution network structure, directly affecting the power quality of its operation. Calculating the spatial distribution of urban residential electricity demand is crucial for the construction of distribution substations and power system planning at the microscale. Summary of the Invention

[0004] To address the aforementioned issues, this invention combines multi-source spatial data and power big data, utilizing the random forest method in machine learning and the genetic algorithm in heuristic algorithms. It provides a method for optimizing urban substation site selection in different power consumption scenarios at a microscale. At the plot scale, it analyzes the spatial distribution of urban electricity demand and identifies the driving factors of power load. The power supply range of each substation is defined, enabling site optimization for urban substations in different scenarios. The optimized results are then compared with those of existing substations.

[0005] First, the power data in the initial dataset needs to be cleaned and aggregated to extract time-series power load sample datasets under different power consumption scenarios. Then, the multi-source spatial data of points, lines, and surfaces is preprocessed to obtain a text dataset of power driving factors for model input. Next, based on the random forest method, combined with the time-series power load sample dataset and the power driving factor text dataset, a multi-scenario power consumption fitting model is constructed to obtain the model's fitting accuracy. Subsequently, based on the micro-scale power load spatial distribution obtained by fitting the multi-scenario power consumption model, the power supply range of each substation is divided based on the region growing algorithm. Finally, the Location Set Coverage Model (LSCP) is introduced to establish a substation site selection model based on LSCP. The optimal value of the substation site selection model is solved using a genetic algorithm, thus obtaining the optimal substation site selection result.

[0006] Power big data includes at least the station number, start time, end time, current, voltage, and power. The data processing involves removing erroneous power data and calculating transformer power values ​​every half hour. It also calculates the average daily power consumption for each township / administrative region. Distribution transformer address data is also part of the power grid data; the Baidu API is used to match the address of each transformer to obtain its precise latitude and longitude. The address data includes at least the city company name, transformer substation number, substation name, line name, and transformer name.

[0007] To quantitatively represent the impact of Point of Interest (POI) elements on the electrical load of the surrounding space in point-source spatial data, a kernel density method is employed. The mathematical principle is as follows:

[0008]

[0009] Where h is the search bandwidth, K0 is the given point data field, N is the number of two-dimensional points, and dist i Let f(x,y) be the distance between point i and point (x,y), and f(x,y) be the kernel density. Assume that a smooth surface similar to a normal distribution function is built around each point. The density calculated at the center point is the highest. As the distance between the point and the center point gradually increases, the function value continuously decreases until the distance between the point to be calculated and the center point reaches the maximum, at which point the function value is 0.

[0010] For linear multi-source spatial data, to quantify the impact of various line features on spatial electrical load, the Euclidean distance from a grid point to its nearest line feature is calculated. The calculation formula is as follows:

[0011]

[0012] Where x i With y i The coordinates of point i are represented by n, which represents the number of source pixels. This is used to determine the straight-line distance between the center of each pixel and the nearest source pixel. If the calculated straight-line distance is the maximum distance in terms of influence, the result is assigned to the pixel center; otherwise, it is assigned to 0.

[0013] For planar multi-source spatial data, bilinear interpolation is used for interpolation processing. The principle of bilinear interpolation is as follows:

[0014]

[0015] Q 11 Q 12 Q 21 Q22 Let x1, x2, y1, and y2 represent the coordinates of the four adjacent points of the point to be calculated, x and y be the coordinates of the point to be calculated, and f(x,y) be the numerical value of the point to be calculated. 11 ) represents Q 11 The value of the point, f(Q) 12 ) represents Q 12 The value of the point, f(Q) 21 ) represents Q 21 The value of the point, f(Q) 22 ) represents Q 22 The numerical value of the point.

[0016] Urban electricity consumption was divided into four scenarios: peak electricity consumption, winter peak electricity consumption, summer peak electricity consumption, and off-peak electricity consumption. The peak electricity consumption scenario was calculated by summing the maximum power values ​​of all distribution transformers in each township at different times. The winter peak electricity consumption scenario used the maximum power values ​​obtained from December to February each year. The summer peak electricity consumption scenario used the maximum power values ​​obtained from June to August each year. The off-peak electricity consumption scenario used the maximum power values ​​obtained from March to May and September to November each year. The power consumption fitting models for these four different scenarios were constructed using the random forest method in machine learning, employing 18 categories of power driving factors from multi-source spatial data as features and multi-period electricity loads from each plot as labels. The goodness-of-fit R-squared was used. 2 The Pearson correlation coefficient (Pearson R) and root mean square error (RMSE) were used to evaluate the model's fitting accuracy. This model was used to fit the spatial distribution of electricity load in multiple scenarios within the study area at a spatial resolution of 100m, ultimately determining the electricity demand of urban residents.

[0017] Based on multi-scenario fine-scale power mapping, in order to simulate the process of substations allocating power to power demand points, the idea of ​​region growing algorithm is used to divide the reasonable power supply area of ​​each substation. The set region growing conditions are: (1) The Euclidean distance between the power demand point and the substation is less than or equal to the power supply radius. (2) The load rate of the substation is less than the maximum load rate. The specific steps of dividing the power supply range of the substation are as follows: First, the grid point where the substation is located is designated as the seed point. The growth condition judgment is performed on the power demand point closest to the substation. If the condition is met, the power demand point is allocated to this substation. Second, the growth condition judgment is performed on the grid of the 8 neighboring power demand points around the seed point. The demand points that meet the conditions are allocated. When judging the growth condition of the power demand point, the Euclidean distance from all substations to the demand point is calculated. If the nearest substation meets the condition for continued growth, the demand point is allocated to the nearest substation. Otherwise, the demand point is allocated to other substations step by step from near to far. The process continues until one of the following conditions is met: the substation reaches its maximum load rate, all power demand points are allocated, or the demand exceeds the substation's power supply radius: at which point the area growth terminates and the substation's power supply range is defined.

[0018] The substation location optimization model mainly consists of two parts: problem formulation and optimization solution. It includes urban substation site selection based on LSCP and optimal location solution based on genetic algorithm. The construction process of the substation site selection optimization model is as follows:

[0019] (1) The LSCP coverage model is introduced, which aims to cover all service demand points within a certain area with the minimum number of facilities. Based on the characteristics of urban power grid multi-scenario power load spatial distribution and distribution network: transformer capacity, number of transformers, power factor, power supply radius, and maximum load rate, and considering the actual substation power supply range division, while maximizing urban power grid coverage and power grid equipment utilization, an LSCP-based urban substation site selection problem is established. This problem can be expressed as: within the urban area, covering all power demand points with the minimum number of 110kV terminal substations, with the goal of maximizing power grid coverage and distribution transformer utilization. The power grid coverage rate is the ratio of the number of power demand points that receive power to the total number of power demand points, and the distribution transformer utilization rate is defined as the ratio of the current load rate of the substation to the maximum load rate. The specific mathematical expression is as follows:

[0020] Objective function:

[0021]

[0022] Constraints:

[0023] E j ≤C j Lj (1)

[0024] d ij ≤R (2)

[0025] W1 + W2 = 1 (3)

[0026] Where W1 represents the coverage weight, W2 represents the utilization weight, and if substation j is selected, then Y... j =1; otherwise, Y j =0, if the electricity demand point i is allocated, then Y i =1, otherwise, Y i =0, i and j represent the electricity demand point index and substation index respectively, E j T represents the actual active power output of substation j. i T represents the total number of electricity demand points. j L represents the total number of substations selected. j C represents the maximum load factor of substation j. j d represents the total capacity of the main transformer in substation j. ij R represents the distance between demand point i and substation j, and R represents the power supply radius of the substation.

[0027] The constraints (1) in the substation location optimization model are used to ensure that the substation outputs active power to the demand point within the power supply safety range and does not cause overload. Constraint (2) is used to limit the Euclidean distance between the power demand point and the substation to be less than or equal to the power supply radius of the substation. Constraint (3) is used to ensure that the weight sum of the grid coverage rate and transformer utilization rate is 1. (2) For the objective function of the substation location model proposed in (1), a heuristic optimization algorithm is used to find the optimal solution. The specific solution steps include seven steps: encoding, initialization, determining the fitness function, selection, crossover, mutation, and termination of iteration. The main methods include binary encoding, tournament selection, single-point crossover, and multi-point mutation. First, the encoding part adopts the binary encoding method, using n binary bits as the chromosome structure, where n represents the number of candidate facilities. If the candidate facility is selected, the corresponding bit value is set to 1; otherwise, the bit value is set to 0. Second, the population size, initial iteration number, and the initial candidate points and power demand points of different scenarios are initialized. The fitness function is designed according to the objective function. A tournament selection method is used, where K individuals are randomly selected from the population each time. The individual with the highest fitness value is selected to enter the next generation. A single-point crossover method is used, where a crossover point is randomly set in the individual's encoding, and parts of the chromosomes of two individuals are exchanged. Genes are randomly exchanged between two individuals in the population according to the crossover rate. A multi-point mutation method is used, where gene values ​​at certain loci in the individual's chromosome encoding string are replaced with alleles at those loci, thus forming new individuals. Finally, the genetic algorithm terminates after reaching its maximum number of iterations, yielding the location latitude and longitude, coverage rate, and transformer utilization rate of each substation to be newly built.

[0028] The beneficial effects of the technical solution provided by this invention are: the substation site selection analysis for different power consumption scenarios at the micro scale effectively evaluates the rationality of the existing urban substation distribution, and compares and analyzes the optimized site selection results with the existing results, providing an efficient solution for urban power grid planning. Attached Figure Description

[0029] The present invention will be further described below with reference to the accompanying drawings and embodiments. In the accompanying drawings:

[0030] Figure 1 is a flowchart of a method for optimizing the location of urban substations for different power consumption scenarios at the microscale, according to an embodiment of the present invention.

[0031] Figure 2 is a flowchart of data preprocessing in an embodiment of the present invention. Detailed Implementation

[0032] To provide a clearer understanding of the technical features, objectives, and effects of the present invention, specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0033] Please refer to Figures 1-2. Figure 1 is a flowchart of a method for optimizing the location of urban substations under different power consumption scenarios at a microscale according to an embodiment of the present invention. Figure 2 is a flowchart of data preprocessing according to an embodiment of the present invention, specifically including:

[0034] (1) Remove the erroneous data values ​​in the power big data (including station number, start time, end time, current, voltage, and power), count the transformer power value every half hour, and calculate the average daily power consumption in each township administrative region by region.

[0035] (2) The multi-source spatial data is processed to obtain a text dataset of power driving factors, which includes 18 categories of power driving factors from multi-source spatial data.

[0036] For point-source spatial data, kernel density analysis is used to quantitatively represent the impact of POI elements on the electrical load of the surrounding space. The specific formula for calculating kernel density is as follows:

[0037]

[0038] Where h is the search bandwidth, K0 is the given point data field, N is the number of two-dimensional points, and dist i Let f(x,y) be the distance between point i and position (x,y), and f(x,y) be the kernel density. Assume that a smooth surface similar to a normal distribution function is established with each point as the center. The density calculated at the center point is the highest. As the distance between the point and the center point gradually increases, the function value continuously decreases until the distance between the point to be calculated and the center point reaches the maximum, at which point the function value is 0.

[0039] For linear multi-source spatial data, to facilitate the quantification of the impact of various line features on spatial power load, the Euclidean distance processing method is adopted. The Euclidean distance dist(X,Y) from a grid point to its nearest line feature is calculated using the following formula:

[0040]

[0041] Where x i With y i Represents the coordinates of point i. Determine the straight-line distance between the center of each pixel and the nearest source pixel. If the calculated straight-line distance is the maximum influential distance, then assign the result to the pixel center; otherwise, assign it to 0.

[0042] To unify the format and resolution of planar raster data, the following bilinear interpolation method is used for interpolation processing:

[0043]

[0044] Q 11 Q 12 Q 21 Q 22 Let x1, x2, y1, and y2 represent the coordinates of the four adjacent points to be calculated, x and y represent the coordinates of the point to be calculated, and f(x,y) represent the value of the point to be calculated.

[0045] After data preprocessing, the overall framework is constructed, mainly consisting of three parts: micro-scale power consumption mapping, substation power supply range delineation, and substation site selection optimization.

[0046] (3) The process of plotting electrical power at the microscale is as follows:

[0047] Urban electricity consumption is divided into four scenarios: peak electricity consumption in winter, peak electricity consumption in summer, and off-peak electricity consumption in normal times. A random forest method is used to construct a multi-scenario electricity consumption power fitting model. Eighteen categories of power driving factors from multi-source spatial data (specifically including community density, density of residential facilities, distance to highways, distance to railways, distance to main roads, distance to train stations, density of educational institutions, density of companies and enterprises, density of entertainment venues, density of financial institutions, density of medical institutions, density of hotels, nighttime light index, population density, density of restaurants, density and slope of shopping districts, and density of transportation facilities) are used as features, and the electricity load of each plot is used as a label input into this multi-scenario electricity consumption power fitting model. The goodness-of-fit R-squared is used. 2 The Pearson correlation coefficient (Pearson R) and root mean square error (RMSE) were used to evaluate the model's fitting accuracy. This multi-scenario power consumption fitting model was used to obtain the spatial distribution of power load in the study area under multiple scenarios with a spatial resolution of 100m, ultimately determining the electricity demand of urban residents.

[0048] (4) The process of dividing the power supply range of the substation is as follows:

[0049] The idea of ​​region growth algorithm is used to divide the reasonable power supply area of ​​each substation. The set region growth conditions are: (1) the Euclidean distance between the power demand point and the substation is less than or equal to the power supply radius. (2) the load rate of the substation is less than the maximum load rate. The specific steps of dividing the power supply range of the substation are as follows: First, the grid point where the substation is located is designated as the seed point. The growth condition judgment is performed on the power demand point closest to the substation. If the condition is met, the power demand point is allocated to this substation. Second, the growth condition judgment is performed on the grid of the 8 neighboring power demand points around the seed point. The demand points that meet the conditions are allocated. When judging the growth condition of the power demand point, the Euclidean distance from all substations to the demand point is calculated. If the nearest substation meets the condition for continued growth, the demand point is allocated to the nearest substation. Otherwise, the demand point is allocated to other substations step by step from near to far. The judgment is repeated until the substation reaches the maximum load rate, all power demand points are allocated, or the power supply radius of the substation is exceeded. The region growth terminates and the division of the power supply range of the substation is completed.

[0050] (5) The process of substation site selection optimization is as follows:

[0051] It mainly includes urban substation site selection based on the Location Set Coverage Model (LSCP) and optimal location solution based on genetic algorithm.

[0052] a. First, a substation location model based on LSCP is established. This model addresses the urban substation location problem, which can be formulated as follows: Within an urban area, the goal is to cover all electricity demand points with the minimum number of 110kV terminal substations, maximizing grid coverage and transformer utilization. Grid coverage is defined as the ratio of the number of electricity demand points receiving power to the total number of electricity demand points. Transformer utilization is defined as the ratio of the current load rate to the maximum load rate of the substation. The objective function and constraints of this substation location model are shown below:

[0053] Objective function:

[0054]

[0055] Constraints:

[0056] E j ≤C j L j (1)

[0057] d ij ≤R (2)

[0058] W1 + W2 = 1 (3)

[0059] The parameters in the above formula and their corresponding meanings are shown in the table below:

[0060]

[0061]

[0062] Constraint (1) Ensure that the substation outputs active power to the demand point within the safe power supply range and does not cause overload. Constraint (2) Limit the Euclidean distance between the power demand point and the substation to be less than or equal to the power supply radius of the substation. Constraint (3) Ensure that the weighted sum of the grid coverage rate and transformer utilization rate is 1.

[0063] b. For the substation site selection problem function proposed in step a, a genetic algorithm is used to find the optimal solution. The specific solution steps include seven steps: encoding, initialization, determining the fitness function, selection, crossover, mutation, and termination of iteration. These mainly include binary encoding, tournament selection, single-point crossover, and multi-point mutation. Finally, after the genetic algorithm reaches its maximum number of iterations, it terminates, obtaining the latitude and longitude coordinates, coverage rate, and transformer utilization rate of each new substation, thus obtaining the final substation site selection result.

[0064] The beneficial effects of this invention are: by conducting substation site selection analysis for different power consumption scenarios at the micro scale, the rationality of the distribution of existing urban substations is effectively evaluated, and the results of the optimized site selection are compared and analyzed with the existing results, providing an efficient solution for urban power grid planning.

[0065] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for optimizing the site selection of urban substations under different power consumption scenarios at the microscale, characterized in that: include: S1: Perform data cleaning and aggregation on the power big data in the initial dataset to extract time-series power load sample datasets under different power consumption scenarios; S2: Perform data preprocessing on multi-source spatial data of points, lines, and surfaces to obtain a text dataset of power driving factors; S3: Based on the random forest method, combine the time-series power load sample dataset and the text dataset of power driving factors to construct a multi-scenario power consumption fitting model; S4: Based on the micro-scale power load spatial distribution obtained by fitting the multi-scenario power consumption fitting model, divide the power supply range of each substation based on the region growing algorithm. S5: Introducing the LSCP coverage model, a substation location optimization model based on LSCP is established. A genetic algorithm is used to solve for the optimal value of this substation location optimization model, thus obtaining the optimal substation location result. The construction process of the substation location optimization model is as follows: The LSCP coverage model is introduced. Based on the spatial distribution of power loads in various scenarios of the urban power grid and the characteristics of the distribution network: transformer capacity, number of transformers, power factor, power supply radius, and maximum load rate, considering the actual substation power supply range division, and maximizing the coverage of the urban power grid and the utilization rate of power grid equipment, an LSCP-based substation location model is established. This substation location model can be expressed as: Within the urban area, the minimum number of 110kV terminal substations is used to cover all power demand points. The goal is to maximize the power grid coverage rate and the distribution transformer utilization rate. The power grid coverage rate is the ratio of the number of power demand points that receive power to the total number of power demand points. The distribution transformer utilization rate is defined as the ratio of the current load rate of the substation to the maximum load rate. The mathematical expression of this substation location model is as follows: Objective function: Constraints: (1) (2) (3) Where W1 represents the coverage weight and W2 represents the utilization weight. If substation j is selected, then Y j =1; otherwise, Y j =0, if electricity demand point i is allocated, then Y i =1, otherwise, Y i =0, where i and j represent the electricity demand point index and substation index, respectively. T represents the actual active power output of substation j. i T represents the total number of electricity demand points. j L represents the total number of substations selected. j C represents the maximum load factor of substation j. j d represents the total capacity of the main transformer in substation j. ij R represents the distance between demand point i and substation j, and R represents the power supply radius of the substation.

2. The method for optimizing the location of urban substations for different power consumption scenarios at a microscale, as described in claim 1, is characterized in that: The aforementioned power big data includes station number, start time of data collection, end time of data collection, current, voltage, and power.

3. The method for optimizing the location of urban substations for different power consumption scenarios at a microscale, as described in claim 1, is characterized in that: The data processing procedure involves removing erroneous power data, calculating the transformer power value every half hour, and calculating the average daily power consumption in each township administrative region.

4. The method for optimizing the location of urban substations for different power consumption scenarios at a microscale, as described in claim 1, is characterized in that: For point-like multi-source spatial data, a kernel density processing method is adopted. The formula for calculating kernel density is as follows: Where h is the search bandwidth. Given a point data field, where N is the number of two-dimensional points, Let f(x,y) be the distance between point i and point (x,y), and f(x,y) be the kernel density. Assume that a smooth surface similar to a normal distribution function is built around each point. The density calculated at the center point is the highest. As the distance between the point and the center point gradually increases, the function value continuously decreases until the distance between the point to be calculated and the center point reaches the maximum, at which point the function value is 0.

5. The method for optimizing the location of urban substations for different power consumption scenarios at a microscale, as described in claim 1, is characterized in that: For linear multi-source spatial data, the Euclidean distance from a raster point to its nearest line feature is calculated using the following formula. ： in and The coordinates of point i are represented by n, which represents the number of source pixels. This is used to determine the straight-line distance between the center of each pixel and the nearest source pixel. If the calculated straight-line distance is the maximum distance in terms of influence, the result is assigned to the pixel center; otherwise, it is assigned to 0.

6. The method for optimizing the location of urban substations for different power consumption scenarios at a microscale, as described in claim 1, is characterized in that: For planar multi-source spatial data, bilinear interpolation is used for interpolation processing. The principle of bilinear interpolation is as follows: in ， ， ， These represent the four adjacent points of the point to be calculated. ， ， ， Let x and y be the coordinates of four adjacent points, respectively, and let f(x,y) be the coordinates of the point to be calculated. Let f(Q) be the value of the point to be calculated. 11 ) represents Q 11 The value of the point, f(Q) 12 ) represents Q 12 The value of the point, f(Q) 21 ) represents Q 21 The value of the point, f(Q) 22 ) represents Q 22 The numerical value of the point.

7. The method for optimizing the location of urban substations for different power consumption scenarios at a microscale, as described in claim 1, is characterized in that: The steps for dividing the power supply range of a substation are as follows: First, designate the grid point where the substation is located as the seed point. Then, determine the growth conditions for the nearest power demand point to the substation. If the conditions are met, allocate the power demand point to this substation. Next, determine the growth conditions for the 8 neighboring power demand grid points around the seed point. Allocate the demand points that meet the conditions. When determining the growth conditions for a power demand point, calculate the Euclidean distance from all substations to that demand point. If the nearest substation meets the conditions for continued growth, allocate the demand point to the nearest substation. Otherwise, gradually determine the demand point from the nearest to the farthest substation and allocate it to other substations. Repeat this process until one of the following conditions is met: the substation reaches its maximum load rate, all power demand points are allocated, or the demand point exceeds the substation's power supply radius. At this point, the area growth terminates, and the substation's power supply range is divided.

8. The method for optimizing the location of urban substations for different power consumption scenarios at a microscale, as described in claim 7, is characterized in that: The set regional growth conditions are: (1) the Euclidean distance between the electricity demand point and the substation is less than or equal to the power supply radius, and (2) the load rate of the substation is less than the maximum load rate.

9. The method for optimizing the location of urban substations for different power consumption scenarios at a microscale, as described in claim 1, is characterized in that: The specific steps for solving the optimal value using a genetic algorithm include seven parts: encoding, initialization, determining the fitness function, selection, crossover, mutation, and terminating the iteration. First, the encoding part uses binary encoding, employing n binary digits as the chromosome structure, where n represents the number of candidate facilities. If a candidate facility is selected, the corresponding bit value is set to 1; otherwise, the bit value is set to 0. Next, the population size, initial iteration count, and the initial candidate points and electricity demand points for different scenarios are initialized. The fitness function is designed based on the objective function and uses a tournament selection method, randomly selecting K values ​​from the population each time. For each individual, based on its fitness value, the individual with the highest fitness is selected to enter the next generation of the population. A single-point crossover method is used, where a crossover point is randomly set in the individual's encoding, and parts of the chromosomes of two individuals are exchanged. In the population, two individuals randomly exchange genes according to the crossover rate. A multi-point mutation method is used, where the gene values ​​at certain loci in the individual's chromosome encoding string are replaced with alleles at those loci, thus forming new individuals. After multiple iterations, the genetic algorithm terminates when the preset maximum number of iterations is reached, and the location latitude and longitude, coverage rate, and transformer utilization rate of each substation that needs to be newly built are obtained.

Images