Parallel three-phase unbalance optimization method based on spatial and user load time sequence characteristics

Abstract

The application provides a parallel three-phase unbalance optimization method based on space and user load time sequence characteristics i . According to the network structure of a low-voltage distribution network and the density of the access positions of users in a transformer area, the users in the transformer area are divided into a plurality of node sets M. Based on the FCM clustering algorithm, the users in each node set are classified separately to obtain typical user curves of each type of user in each node. The typical user curves are used to replace the actual load of each type of user in each node, an improved genetic algorithm considering the parent gene intergenerational inheritance is created, each node set is optimized separately, the average three-phase unbalance degree in one day is taken as an objective function, a parallel optimization structure is formed, and the three-phase unbalance degree at the whole transformer area and the main feeder can be reduced after optimization. The application can solve the problem of three-phase load unbalance, especially in cooperation with a multi-stage distribution network, reduce line loss, and improve the power supply reliability of the distribution network.

Description

Technical Field

[0001] This invention belongs to the field of power grid technology, and specifically relates to a parallel three-phase imbalance optimization method based on spatial and user load timing characteristics. Background Technology

[0002] Due to their large scale and complex structure, current low-voltage distribution networks face increasingly prominent power consumption problems, particularly three-phase imbalance. Because of the randomness of electricity consumption by users in different distribution areas, the uncertainty of power load, and the spatial differences in the feeders of each user within the entire low-voltage distribution network, a large number of three-phase imbalances exist in actual engineering projects. This negatively impacts power loss reduction, power quality for users in different distribution areas, and the normal operating temperature of transformers.

[0003] Current measures to address three-phase imbalance problems mostly involve using power electronic converters and reactive power compensation devices to compensate for the unbalanced voltage or current on the low-voltage side of the transformer. However, these measures fail to consider the network structure of the low-voltage distribution network in actual engineering projects. In reality, the locations of feeder connections for users in a distribution area are not uniform but vary in density. Previous methods did not take these factors into account, resulting in discrepancies between the actual and anticipated improvement outcomes.

[0004] Besides compensation methods, automatic phase-switching switches are also used to address three-phase imbalance. When the three-phase imbalance of users in a distribution area reaches a certain level, a command is issued to the phase-switching switches to reduce the imbalance. However, phase-switching switches have a maximum number of switching cycles and are relatively expensive; generally, only 5-8 phase-switching switches are installed in a single distribution area. Therefore, designing a superior method for optimizing three-phase imbalance in low-voltage distribution networks has become a research direction. Summary of the Invention

[0005] This invention, based on the network structure of low-voltage distribution networks and the time-series characteristics of user loads in distribution areas, divides the entire low-voltage distribution network into nodes, performs separate cluster analysis on the user set of each node, and forms a typical load database for each node set. It also proposes an improved genetic algorithm that considers the generational inheritance of parent genes, and optimizes the governance with the minimum average three-phase imbalance throughout the day as the optimization objective. This forms a parallel multi-threaded governance optimization structure, which can achieve long-term three-phase balance with only one phase switching operation. Moreover, each node is independent of the others, and their three-phase balance does not affect each other, making it easier to achieve regional individual governance.

[0006] A parallel three-phase imbalance optimization method based on spatial and user load temporal characteristics includes the following steps: S1. Based on the density of user connections in the distribution area, divide the N distribution area users of the low-voltage distribution network into multiple node sets {M}. i |i=1,2,...,m}, where N and m are positive integers; In this set, the wiring locations of the users in each node area are closely spaced and have the same electrical distance, and each node set M i It includes There are n+n+n users in each distribution area, i.e., n1+n2+...+n i +...+n m =N, where, It is a positive integer; S2. Obtain the average daily load data matrix P of N transformer substations in the calibrated time period of the low-voltage distribution network. (1) in, This represents the average daily load matrix composed of load power data from N users; This represents the average daily load power matrix for the k-th user, where k = 1, 2, ..., N; S3, for each node set M i For users in the affected area, the optimal number of classifications is determined through separate analysis using clustering validity functions. And based on the optimal number of classifications Fuzzy C-means clustering algorithm is used to cluster P, and the cluster center of each node user is used as the typical load curve of each node user, specifically including: S31. Determine the clustering validity function , Here is the membership matrix, and c is the number of categories; S32. For each set of nodes, use the FCM clustering algorithm to calculate c. i From 2 to n i Membership matrix U i and cluster center V i , where c i The number of categories for each node set; S33. Using clustering validity functions To obtain the optimal number of categories for each node set ; S34. Based on the optimal number of classifications for each node. For each set of nodes M i Users are categorized, and the cluster center of each user category is used as a typical load curve. S4. Create an improved genetic algorithm that considers the inheritance of parental genes across generations; An improved genetic algorithm that considers the skipped-generation inheritance of parental genes involves selecting four individuals (parent and offspring) after each iteration's crossover process. These four individuals then compete, and the two with the highest fitness are selected to enter a new population for mutation. Specifically: S41. Determine the population size, and initialize the 0th generation population using the fixed phase sequence combination of the given user as the gene code. S42. Using the average three-phase imbalance of a given user throughout the day as the fitness, calculate the fitness of each individual in the population and save the best individual. S43. Remove phase sequence combinations with high three-phase imbalance; S44. Perform pairwise crossover on the population after the selection operation. After the crossover operation, the four individuals (parent and offspring) compete to select the two individuals with the lowest three-phase imbalance throughout the day to enter the new population. S45. Perform mutation operations on the new population to expand the gene search space; determine whether the algorithm termination condition is met. If not, return to S42. If it is met, output the optimal solution, which is the user phase sequence combination of the substation area with the lowest average three-phase imbalance throughout the day. S5. An improved genetic algorithm that considers the intergenerational inheritance of parental genes is applied to optimize and manage each node in parallel with the objective function of minimizing the average three-phase imbalance throughout the day. Finally, the optimal phase sequence for all users in the transformer area is determined, so that the phase sequence combination of the entire transformer area reaches a three-phase balance state. S51. Based on the results of user classification for each node in step S3, the typical load curves of each node are used as the actual load curves of users in the transformer area. S52. Calculate the three-phase unbalance power index based on the load power of each node user. The calculation expression is as follows:

[0007] in, It consists of three phases: A, B, and C. For node i Phase instantaneous power value; Let be the instantaneous three-phase average power at node i. Represents node i Phase imbalance Let be the instantaneous power value of phase A at node i; Let be the instantaneous power value of phase B at node i; Let be the instantaneous C-phase power value of node i; S53. Based on the established three-phase imbalance power index, and using the improved genetic algorithm that considers the generational inheritance of parental genes, node optimization is performed for each node with the minimum 24-hour average three-phase imbalance as the objective function. Specifically: S531. For node 1, only input the average load matrix of users in the transformer area after replacing the typical load curve of node 1. Based on the improved genetic algorithm created in S4 that considers the inheritance of parent genes across generations, optimize the solution with the minimum three-phase imbalance degree throughout the day as the objective function, and use the optimal solution as the best access phase sequence for each user in node 1. S532. Continue with node 2, only input the average load matrix of the transformer area users after replacing the typical load curve of node 2, and continue to optimize and solve to determine the optimal access phase sequence of each transformer area user in node 2; repeat this optimization operation for each node until the optimal phase sequence of all transformer area users is determined. At this time, the three-phase imbalance of the entire transformer area and the main feeder nodes is the lowest.

[0008] Furthermore, step S1 is specifically as follows: S11. Investigate the wiring and connection of the entire transformer substation area, identify the users connected to each power pole distribution box, and establish a power topology map of the entire transformer substation area. S12. Based on the established transformer substation network topology and the differences in electrical distance between the substation users, the N users in the substation are divided into nodes, resulting in multiple node sets {M}. i |i=1,2,...,m}, where each node set M i The wiring locations of the included transformer substation users are close together, and the electrical distances are the same.

[0009] Further, step S2 specifically includes: S21. Obtain load data D for N users in a calibrated time period of the low-voltage distribution network through a data acquisition and monitoring control system with a calibrated time limit as the sampling interval; S22. Average the daily load power data D within the specified time period to obtain the average daily load data matrix P for the users in the transformer area:

[0010] in, This represents the average daily load matrix composed of load power data from N users; This represents the average daily load power matrix for the k-th user, where k = 1, 2, ..., N. Let represent the average load power of the k-th user at the t-th time sampling point, where t = 1, 2, ..., 24.

[0011] Furthermore, the calibration period in step S2 is 30 days.

[0012] Furthermore, the clustering validity function in step S31 The definition is as follows:

[0013] In the formula, The total number of samples, The ambiguity is set to 2. For the k-th sample data, Represents the membership matrix. Represents the membership matrix The Middle The data of the first sample relative to the first Membership degree of a class Indicates the first Cluster centers of classes Represents the number of categories, It is the center of all samples. Clustering validity function It is an extremely large indicator.

[0014] Further, step S33 specifically includes: Each node has a different c i The membership matrix U corresponding to the value i and cluster center V i Substituting these values ​​into the clustering effectiveness function formula (3), when... At this time This is the optimal number of categories for node i, at which point... and For node i, the optimal number of categories The membership matrix and cluster centers at that time.

[0015] Furthermore, the node optimization in step S53 specifically involves: The load matrix data of each node is used as the input of the improved genetic algorithm, and the minimum average three-phase imbalance over the whole day is used as the objective function. The algorithm is solved iteratively. When optimizing the nodes, each node is independent of the others and does not affect them. That is, when optimizing the governance, only the three-phase balance of a single node is considered.

[0016] Furthermore, the calibration time limit in step S21 is 1 hour.

[0017] Compared with the prior art, the present invention has the following beneficial effects: 1. When optimizing three-phase imbalance, this invention takes into account the network structure of the low-voltage distribution network, the density of user access locations in the distribution area, and the timing characteristics of user loads. Compared with traditional solutions, this invention is more practical and has better governance results.

[0018] 2. This invention employs a method where each node user is optimized and managed independently, ultimately determining the optimal phase sequence for all users in all transformer areas. This approach isolates each node user, forming a parallel, multi-threaded optimization structure. Since the typical load databases of each node differ and are not correlated during optimization and management, it facilitates easier implementation of individual area-specific management.

[0019] 3. A genetic algorithm that considers the generational inheritance of parental genes is adopted, which has better performance than traditional genetic algorithms. Attached Figure Description

[0020] Figure 1 This is a flowchart illustrating the steps of the parallel three-phase imbalance mitigation method based on spatial and user load timing characteristics of the present invention. Figure 2 This is a node topology diagram of a low-voltage distribution network after node division according to an embodiment of the present invention; Figure 3 This is a flowchart of an improved genetic algorithm that considers the intergenerational inheritance of parental genes according to an embodiment of the present invention. Figure 4 The variation of the three-phase imbalance of example node 4 in an embodiment of the present invention with the number of iterations of the improved genetic algorithm; Figure 5 This is an embodiment of the present invention showing the real-time three-phase imbalance of a low-voltage distribution network before optimization. Figure 6 This is an optimized real-time three-phase imbalance of a low-voltage distribution network according to an embodiment of the present invention. Detailed Implementation

[0021] To address the shortcomings of existing technologies, this invention provides a parallel three-phase imbalance optimization method based on spatial and user load temporal characteristics. Starting from the network structure of the low-voltage distribution network, it divides the users in each distribution area into nodes, and then uses a genetic algorithm with parent-child inheritance to optimize each node individually, forming a parallel optimization structure. This reduces the three-phase imbalance while weakening the correlation between different areas of the entire low-voltage distribution network, making it easier to achieve individual area management, and ultimately determining the optimal phase sequence access scheme for all distribution area users. Three-phase imbalance problems exist in all parts of the power system, but their impact is particularly significant and prominent in low-voltage distribution networks. Low-voltage distribution networks are generally 0.4kV transmission lines, similar to most rural areas or communities in my country, where a single transformer supplies power to a village or community. Users supplied by the same transformer are collectively referred to as distribution area users.

[0022] As an example, the present invention provides a parallel three-phase imbalance optimization method based on spatial and user load timing characteristics, such as... Figure 1 As shown, it includes the following steps: S1. Based on the density of user connections in the distribution area, divide the N distribution area users of the low-voltage distribution network into multiple node sets {M}. i |i=1,2,...,m}; For each node set M i The wiring locations of the users in the included transformer substations are closely spaced, and the electrical distances are the same. It includes a total of There are n+n+n users in each distribution area, i.e., n1+n2+...+n i +...+n m =N. Specifically, this includes: S11. Investigate the wiring and connection of the entire transformer substation area, identify the users connected to each power pole distribution box, and establish a power topology map of the entire transformer substation area. S12. Based on the established distribution network topology, and according to the different electrical distances between the distribution users, divide the N users in the distribution area into m node sets. S2. Obtain the load data D of N users in the low-voltage distribution network within the past month through the SCADA system.

[0023] The daily load power of N users in the low-voltage distribution network is collected at 1-hour sampling intervals. The daily load power over nearly thirty days is averaged to obtain the average daily load matrix of the users in the distribution area. :

[0024] in, This represents the average daily load matrix composed of load power data from N users; This represents the average daily load power matrix for the k-th user, where k = 1, 2, ..., N. Let represent the average load power of the k-th user at the t-th time sampling point, where t = 1, 2, ..., 24.

[0025] S3. A clustering validity function is used to perform parallel, separate analysis of the average daily load data of users in each node cluster area to determine the optimal number of clusters. The load data is clustered using the fuzzy C-means clustering algorithm to form a typical load database for each node set, specifically including: S31. Determine the clustering validity function ;

[0026] In the formula, The total number of samples, The ambiguity is set to 2. For the k-th sample data, Represents the membership matrix. Represents the membership matrix The Middle The data of the first sample relative to the first Membership degree of a class Indicates the first Cluster centers of classes Represents the number of categories, It is the center of all samples. Clustering validity function It is a very large indicator; the larger the value, the better the clustering result.

[0027] S32. For each set of nodes, use the FCM clustering algorithm to calculate c. i From 2 to n i Membership matrix U i and cluster center V i， c i , where c i The number of categories for each node set; S321. For node 1, traverse c1 from 2 to n1 to obtain the membership matrix U1 and cluster center V1 for each c1. S322. Continue performing cluster validity analysis on the next node 2 until all nodes have completed the cluster validity analysis. Finally, the different c values ​​for each node can be obtained. i The corresponding U i and V i This is to obtain the optimal number of classifications for each node based on the clustering validity function.

[0028] S33. Using clustering validity functions Obtain the optimal number of categories for each node; Each node has a different c i The membership matrix U corresponding to the value i and cluster center V i Substituting these values ​​into the clustering effectiveness function formula (3), when... At this time This is the optimal number of categories for node i, at which point... and For node i, the optimal number of categories The membership matrix and cluster centers at that time.

[0029] S34. Based on the optimal number of classifications for each node. For each set of nodes n i The users are classified, and the cluster center of each user group is used as the typical load curve; S4. Create an improved genetic algorithm that considers the skipped-generation inheritance of parental genes. This improved genetic algorithm involves selecting four individuals (parent and offspring) after each iteration's crossover process to compete. The two individuals with the highest fitness are then added to a new population, where mutation is performed. S41. Determine the population size and initialize the 0th generation population using a fixed phase sequence combination given for a specific user as the gene encoding.

[0030] S42. Using the average three-phase imbalance of a given user throughout the day as the fitness, calculate the fitness of each individual in the population and save the best individual.

[0031] S43. Use roulette wheel selection on the population with good fitness to eliminate the phase sequence combination with high three-phase imbalance.

[0032] S44. Perform pairwise crossover on the selected population. After the crossover, the parent and its offspring (a total of 4 individuals) compete to select the two individuals with higher fitness to enter the new population. This ensures that the superior parent genes can be inherited across generations and that the superior gene segments are not lost due to the crossover.

[0033] S45. Perform mutation operations on the new population to expand the gene search space. Determine if the algorithm termination condition is met. If not, return to S42; otherwise, the algorithm terminates and outputs the optimal solution.

[0034] S5. An improved genetic algorithm that considers the intergenerational inheritance of parental genes is applied to optimize and manage each node in parallel with the objective function of minimizing the average three-phase imbalance throughout the day. Finally, the optimal phase sequence for all users in the transformer area is determined, so that the phase sequence combination of the entire transformer area reaches a three-phase balance state.

[0035] S51. Based on the user classification results of each node, the typical load curves of each node are used as the actual load curves of users in the transformer area. S52. Calculate the three-phase unbalance power index based on the load power of each node user. The calculation expression is as follows:

[0036] in, It consists of three phases: A, B, and C. For node i Phase instantaneous power value; Let be the instantaneous three-phase average power at node i. Represents node i Phase imbalance Let be the instantaneous power value of phase A at node i; Let be the instantaneous power value of phase B at node i; Let be the instantaneous C-phase power value of node i; S53. Based on the established three-phase imbalance power index, and using an improved genetic algorithm that considers generational inheritance of parental genes, parallel node optimization is performed on each node with the minimum daily average three-phase imbalance as the objective function. Specifically, node optimization involves using the load matrix data of each node as input to the improved genetic algorithm, with the minimum daily average three-phase imbalance as the objective function, and iteratively solving for the solution. During node optimization, each node is independent and does not affect the others; that is, the optimization focuses only on the three-phase balance of a single node. S531. For node 1, only input the average load matrix of users in the transformer area after replacing the typical load curve of node 1. Based on the improved genetic algorithm created in S4 that considers the inheritance of parent genes across generations, optimize the solution with the minimum three-phase imbalance throughout the day as the objective function. The solution result of the algorithm is used as the optimal access phase sequence for each user in node 1.

[0037] S532. Continuing with node 2, input only the average load matrix of the transformer area users after replacing the typical load curve of node 2, and continue to optimize and solve to determine the optimal access phase sequence for each transformer area user in node 2. Repeat this optimization operation for each node until the optimal phase sequence for all transformer area users is determined. At this point, the three-phase imbalance of the entire transformer area and the main feeder is at its lowest.

[0038] The following is a specific implementation of the present invention in a low-voltage distribution network. In this embodiment, the low-voltage distribution network has a total of 543 users.

[0039] Step 1: Based on the different access locations and density of users in each distribution area, the entire distribution area is divided into 11 nodes. The distribution network topology after node division is as follows: Figure 2 As shown in Table 1, the number of users on each node after node partitioning is displayed.

[0040] Table 1

[0041] Step 2: Obtain the average daily load matrix D of the 543 users in the power distribution network over the past month through the SCADA system.

[0042] Step 3: Cluster analysis was performed on the 11 nodes of this distribution network using a clustering validity function to determine the optimal number of clusters for each node. Each node was then classified individually, and a database of typical load curves for each node was created. Table 2 shows the optimal number of clusters for each node.

[0043] Table 2

[0044] Step four: An improved genetic algorithm that considers generational inheritance of parental genes is used to optimize and manage each node individually. The algorithm flow is as follows: Figure 3 As shown. Taking node 4 as an example, Figure 4 It is the objective function, i.e., the change of the three-phase imbalance with the number of algorithm iterations.

[0045] Step 5: Perform parallel optimization and management on each node individually, without considering the impact of three-phase balance between nodes, to determine the optimal phase sequence for the 543 users in this distribution network.

[0046] like Figure 5 As shown, this represents the real-time three-phase imbalance of the low-voltage side of the transformer and each node in the low-voltage distribution network before the treatment. The three-phase imbalance on the low-voltage side of the transformer is as high as 16.0083%, and the three-phase imbalance at other major feeder nodes is also as high as 15% or more. In particular, the three-phase imbalance at node 6 reaches 41.6851%.

[0047] The objective function is to minimize the average three-phase imbalance over a 24-hour period. The solution is obtained using a genetic algorithm that considers the generational inheritance of parental genes, following a parallel optimization structure.

[0048] Optimization results are as follows Figure 6 As shown, after considering the serial optimization of the distribution network structure, the three-phase imbalance on the low-voltage side of the transformer is reduced to 0.84895%, and the three-phase imbalance at most other major feeder nodes is below 3%. Figure 5 and Figure 6 The comparison shows that the present invention not only reduces the three-phase imbalance on the low-voltage side of the transformer, but also takes into account the three-phase imbalance at the main feeder nodes in the distribution network, and has a good effect on the treatment of three-phase imbalance.

[0049] This invention divides the users in a distribution area into multiple node sets M based on the network structure of the low-voltage distribution network and the density of user access locations. i， Each set has a total of Each node set is individually classified using the FCM clustering algorithm, resulting in a typical user curve for each node. Replacing the actual load of each user class within each node with this typical user curve, an improved genetic algorithm considering generational inheritance is proposed. This algorithm optimizes each node set individually, using the minimum daily average three-phase imbalance as the objective function, forming a parallel optimization structure to ultimately determine the optimal phase sequence for all transformer substations. This invention is applicable to solving three-phase load imbalance problems in distribution network planning, particularly in conjunction with multi-level distribution networks, reducing line losses and improving power supply reliability.

[0050] The embodiments described above are merely preferred embodiments of the present invention and are not intended to limit the scope of the present invention. For example, the number of samples in the load matrix can also be in units of weeks or months. Therefore, without departing from the spirit of the present invention, all modifications and improvements made by those skilled in the art to the technical solutions of the present invention should fall within the protection scope defined by the claims of the present invention.

Claims

1. A parallel three-phase imbalance optimization method based on spatial and user load temporal characteristics, characterized in that, Specifically, the following steps are included: S1. Based on the density of user connections in the distribution area, divide the N distribution area users of the low-voltage distribution network into multiple node sets {M}. i |i=1,2,...,m}, where N and m are positive integers; In this set, the wiring locations of the users in each node area are closely spaced and have the same electrical distance, and each node set M i It includes There are n+n+n users in each distribution area, i.e., n1+n2+...+n i +...+n m =N, where, It is a positive integer; S2. Obtain the average daily load data matrix P of N transformer substations in the calibrated time period of the low-voltage distribution network. (1) ； in, This represents the average daily load matrix composed of load power data from N users; This represents the average daily load power matrix for the k-th user, where k = 1, 2, ..., N; S3, for each node set M i For users in the affected area, the optimal number of classifications is determined through separate analysis using clustering validity functions. And based on the optimal number of classifications Fuzzy C-means clustering algorithm is used to cluster P, and the cluster center of each node user is used as the typical load curve of each node user, specifically including: S31. Determine the clustering validity function , Here is the membership matrix, and c is the number of categories; S32. For each set of nodes, use the FCM clustering algorithm to calculate c. i From 2 to n i Membership matrix U i and cluster center V i , where c i The number of categories for each node set; S33. Using clustering validity functions To obtain the optimal number of categories for each node set ; S34. Based on the optimal number of classifications for each node. For each set of nodes M i Users are categorized, and the cluster center of each user category is used as a typical load curve. S4. Create an improved genetic algorithm that considers the inheritance of parental genes across generations; An improved genetic algorithm that considers the skipped-generation inheritance of parental genes involves selecting four individuals (parent and offspring) after each iteration's crossover process. These four individuals then compete, and the two with the highest fitness are selected to enter a new population for mutation. Specifically: S41. Determine the population size, and initialize the 0th generation population using the fixed phase sequence combination of the given user as the gene code. S42. Using the average three-phase imbalance of a given user throughout the day as the fitness, calculate the fitness of each individual in the population and save the best individual. S43. Remove phase sequence combinations with high three-phase imbalance; S44. Perform pairwise crossover on the population after the selection operation. After the crossover operation, the four individuals (parent and offspring) compete to select the two individuals with the lowest three-phase imbalance throughout the day to enter the new population. S45. Perform mutation operations on the new population to expand the gene search space; determine whether the algorithm termination condition is met. If not, return to S42. If it is met, output the optimal solution, which is the user phase sequence combination of the substation area with the lowest average three-phase imbalance throughout the day. S5. An improved genetic algorithm that considers the intergenerational inheritance of parental genes is applied to optimize and manage each node in parallel with the objective function of minimizing the average three-phase imbalance throughout the day. Finally, the optimal phase sequence for all users in the transformer area is determined, so that the phase sequence combination of the entire transformer area reaches a three-phase balance state. S51. Based on the results of user classification for each node in step S3, the typical load curves of each node are used as the actual load curves of users in the transformer area. S52. Calculate the three-phase unbalance power index based on the load power of each node user. The calculation expression is as follows: in, It consists of three phases: A, B, and C. For node i Phase instantaneous power value; Let be the instantaneous three-phase average power at node i. Represents node i Phase imbalance Let be the instantaneous power value of phase A at node i; Let be the instantaneous power value of phase B at node i; Let be the instantaneous C-phase power value of node i; S53. Based on the established three-phase imbalance power index, and using the improved genetic algorithm that considers the generational inheritance of parental genes, node optimization is performed for each node with the minimum 24-hour average three-phase imbalance as the objective function. Specifically: S531. For node 1, only input the average load matrix of users in the transformer area after replacing the typical load curve of node 1. Based on the improved genetic algorithm created in S4 that considers the inheritance of parent genes across generations, optimize the solution with the minimum three-phase imbalance degree throughout the day as the objective function, and use the optimal solution as the best access phase sequence for each user in node 1. S532. Continue with node 2, only input the average load matrix of the transformer area users after replacing the typical load curve of node 2, and continue to optimize and solve to determine the optimal access phase sequence of each transformer area user in node 2; repeat this optimization operation for each node until the optimal phase sequence of all transformer area users is determined. At this time, the three-phase imbalance of the entire transformer area and the main feeder nodes is the lowest.

2. The parallel three-phase imbalance optimization method based on spatial and user load temporal characteristics according to claim 1, characterized in that, The specific steps of S1 are as follows: S11. Investigate the wiring and connection of the entire transformer substation area, identify the users connected to each power pole distribution box, and establish a power topology map of the entire transformer substation area. S12. Based on the established transformer substation network topology and the differences in electrical distance between the substation users, the N users in the substation are divided into nodes, resulting in multiple node sets {M}. i |i=1,2,...,m}, where each node set M i The wiring locations of the included transformer substation users are close together, and the electrical distances are the same.

3. The parallel three-phase imbalance optimization method based on spatial and user load temporal characteristics according to claim 1, characterized in that, Step S2 specifically involves: S21. Obtain load data D for N users in a calibrated time period of the low-voltage distribution network through a data acquisition and monitoring control system with a calibrated time limit as the sampling interval; S22. Average the daily load power data D within the specified time period to obtain the average daily load data matrix P for the users in the transformer area: in, This represents the average daily load matrix composed of load power data from N users; This represents the average daily load power matrix for the k-th user, where k = 1, 2, ..., N. Let represent the average load power of the k-th user at the t-th time sampling point, where t = 1, 2, ..., 24.

4. The parallel three-phase imbalance optimization method based on spatial and user load temporal characteristics according to claim 1, characterized in that, The calibration period in step S2 is 30 days.

5. The parallel three-phase imbalance optimization method based on spatial and user load timing characteristics according to claim 1, characterized in that, The clustering validity function in step S31 The definition is as follows: In the formula, The total number of samples, The ambiguity is set to 2. For the k-th sample data, Represents the membership matrix. Represents the membership matrix The Middle The data of the first sample relative to the first Membership degree of a class Indicates the first Cluster centers of classes Represents the number of categories, It is the center of all samples. Clustering validity function It is an extremely large indicator.

6. The parallel three-phase imbalance optimization method based on spatial and user load timing characteristics according to claim 5, characterized in that, Step S33 specifically involves: Each node has a different c i The membership matrix U corresponding to the value i and cluster center V i Substituting these values ​​into the clustering effectiveness function formula (3), when... At this time This is the optimal number of categories for node i, at which point... and For node i, the optimal number of categories The membership matrix and cluster centers at that time.

7. The parallel three-phase imbalance optimization method based on spatial and user load timing characteristics according to claim 1, characterized in that, The node optimization in step S53 specifically involves: The load matrix data of each node is used as the input of the improved genetic algorithm, and the minimum average three-phase imbalance over the whole day is used as the objective function. The algorithm is solved iteratively. When optimizing the nodes, each node is independent of the others and does not affect them. That is, when optimizing the governance, only the three-phase balance of a single node is considered.

8. The parallel three-phase imbalance optimization method based on spatial and user load timing characteristics according to claim 3, characterized in that, The calibration time limit in step S21 is 1 hour.

Images