Power distribution network three-phase imbalance dynamic regulation and control method based on hybrid intelligent optimization algorithm

Abstract

The invention discloses a power distribution network three-phase imbalance dynamic regulation and control method based on a hybrid intelligent optimization algorithm, and belongs to the field of power system automation. Establishing a dynamic regulation and control model taking minimization of three-phase current unbalance degree and minimization of switch action times as double targets; solving the model based on an improved genetic algorithm fusing an experience knowledge guidance mechanism, a Pareto optimal selection mechanism and a global and local collaborative search mechanism; and generating a commutation control instruction according to the obtained optimal commutation scheme, and issuing the commutation control instruction to a load commutation switch for execution. By introducing a gene pool to guide a breeding process, a multi-target selection mechanism based on non-dominated sorting and a local optimization strategy combined with tabu search, comprehensive improvement of the algorithm in the aspects of convergence speed, solution diversity and global optimality is realized.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of power system automation, and particularly relates to a power distribution network three-phase imbalance dynamic regulation method based on a hybrid intelligent optimization algorithm. BACKGROUND

[0002] In a low-voltage distribution network, due to the randomness and unevenness of single-phase loads (such as residential electricity and commercial electricity) in space and time, and the disordered access of distributed power sources (such as photovoltaic), three-phase load imbalance is widespread. Three-phase imbalance can cause a series of hazards, including: increasing the additional loss of lines and transformers, reducing the economy of power grid operation; causing excessive neutral line current, causing safety hazards; affecting transformer output, reducing equipment utilization; causing end-user voltage deviation, worsening power quality. In order to solve this problem, load phase-changing switch technology is widely used, which realizes the rebalancing of three-phase load by dynamically adjusting the access phase sequence of single-phase load.

[0003] Existing load phase-changing regulation strategies mostly use conventional heuristic algorithms, such as standard genetic algorithm (GA), simulated annealing algorithm (SA), etc. However, these methods have several deficiencies in practical application: 1) the algorithm is prone to premature convergence, i.e. converging to a local optimal solution too early, and cannot find a globally optimal phase-changing strategy; 2) the stability of the solution result is poor, and multiple runs may obtain regulation schemes with large quality differences, which cannot guarantee the reliability of control; 3) the two conflicting objectives of "minimum imbalance degree" and "minimum switch action times" are not well balanced, which may cause frequent switch action and shorten the service life of equipment; 4) the convergence speed and computational efficiency of the algorithm need to be improved, and it is difficult to meet the stringent requirements of decision timeliness in some severe imbalance scenarios.

[0004] Therefore, there is an urgent need for a new dynamic regulation method that is fast in solving speed, strong in optimization ability, stable and reliable in result, and can effectively coordinate multi-objective optimization. SUMMARY

[0005] To solve the above technical problems, the present application provides a power distribution network three-phase imbalance dynamic regulation method based on a hybrid intelligent optimization algorithm, comprising:

[0006] According to the current value and initial phase sequence state of each single-phase load in the distribution area, a dynamic regulation model is established with the dual objectives of minimizing three-phase current imbalance degree and minimizing switch action times;

[0007] According to the dynamic regulation model, an improved genetic algorithm is used for solving, which integrates an experience knowledge guiding mechanism, a Pareto optimal selection mechanism and a global and local collaborative search mechanism, to generate an optimal phase-changing scheme;

[0008] The optimal commutation scheme is parsed into commutation control commands and the commutation operation is executed to complete the dynamic control of three-phase imbalance.

[0009] Optionally, the experience-guided mechanism includes establishing and dynamically updating a gene pool, and updating the corresponding probability information in the gene pool based on the frequency of occurrence of genes of different phase sequences at each load position in individuals with a Pareto front level of 1 in the current population.

[0010] Optionally, the gene bank is an N×3 probability matrix, where N is the total number of loads involved in regulation, and each column of the matrix records the probability of the corresponding load accessing phases A, B, and C;

[0011] Based on the updated gene pool, offspring individuals are generated with a bias during population crossover operations to increase the probability of inheriting superior genes.

[0012] Optionally, the crossover operation includes two strategies for random selection:

[0013] Genes are retained based on the same gene loci of the parent individuals. For different gene loci, new genes are randomly generated according to the probability of the gene pool, or an arithmetic crossover operation is performed on the phase sequence vectors of the two parent individuals. Non-integer genes generated are rounded and constrained to maintain the validity of gene values.

[0014] Optionally, the improved genetic algorithm employs a Pareto-optimal population selection mechanism in each generation of evolution. It divides the merged population of parents and offspring into multiple front sets through non-dominated sorting, and selects individuals that are both superior and maintain diversity to enter the next generation based on their Pareto rank and crowding distance.

[0015] Optionally, the Pareto optimal selection mechanism dynamically determines the evolutionary retention scale based on the number of individuals at the first frontier level. When the number of individuals exceeds the preset scale, individuals with larger crowding distances are retained first to ensure a diverse distribution of the solution space.

[0016] Optionally, the global and local collaborative search mechanism includes, during the algorithm iteration process, when the number of iterations reaches half of the total number of iterations, randomly selecting an individual from the first Pareto front of the current population as the initial solution of the tabu search algorithm, and performing a neighborhood search on the individual to improve the quality of the local optimum.

[0017] Optionally, the objective function of the dynamic control model includes:

[0018] Calculate the three-phase current imbalance based on the phase sequence state of each load before and after commutation and minimize this value;

[0019] Calculate the total number of commutation actions based on the difference in load conditions before and after commutation, and minimize this number.

[0020] On the other hand, the present invention also provides an electronic device including a memory, a processor, and a computing program stored in the memory and executable on the processor, wherein the processor implements the method when executing the computing program.

[0021] On the other hand, the present invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the method.

[0022] Compared with the prior art, the present invention has the following advantages and technical effects:

[0023] This invention constructs a hybrid intelligent optimization framework with both global exploration and local exploitation capabilities by organically combining an improved genetic algorithm with a tabu search algorithm. During dynamic regulation, the gene frequency information of historically optimal individuals is recorded in a gene pool, enabling empirical knowledge to guide subsequent searches, making the population evolution direction more targeted and reducing blind randomness. The multi-objective selection mechanism combining non-dominated sorting and crowding distance allows the algorithm to achieve a balance between the two mutually constraining objectives of reducing three-phase current imbalance and reducing the number of commutation actions, thereby obtaining a series of Pareto optimal solutions for system selection.

[0024] Meanwhile, the global and local collaborative search mechanism introduced in the algorithm, by triggering tabu search in the middle of the evolution, optimizes the neighborhood of high-quality individuals, effectively avoiding the defect of standard genetic algorithms being prone to getting trapped in local optima, and improving the global optimality and stability of the final solution.

[0025] Comprehensive experimental results show that the algorithm of this invention is superior to traditional heuristic methods in terms of convergence speed, control accuracy and solution stability. It can achieve higher decision-making efficiency and better control effect in the treatment of three-phase current imbalance, and has strong engineering practicality and promotion value. Attached Figure Description

[0026] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings:

[0027] Figure 1 This is a schematic diagram of a distribution substation with a load switching switch installed according to an embodiment of the present invention;

[0028] Figure 2 This is a flowchart of the NSGA-TS implementation of the present invention;

[0029] Figure 3 This is a schematic diagram of the average convergence curves of each algorithm in 100 tests according to an embodiment of the present invention.

[0030] Figure 4 This is a schematic diagram showing the changes in the three-phase current on the low-voltage side of the distribution transformer according to an embodiment of the present invention. Detailed Implementation

[0031] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0032] It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0033] Example 1

[0034] This embodiment provides a dynamic control method for three-phase imbalance in a distribution network based on a hybrid intelligent optimization algorithm, including:

[0035] To address the aforementioned technical problems, this invention provides a dynamic control method for three-phase imbalance in distribution networks based on a hybrid intelligent optimization algorithm, comprising:

[0036] Based on the current values ​​and initial phase sequence status of each single-phase load in the distribution substation, a dynamic control model is established with the dual objectives of minimizing the three-phase current imbalance and minimizing the number of switching actions.

[0037] Based on the dynamic control model, an improved genetic algorithm that integrates empirical knowledge guidance, Pareto optimal selection, and global and local collaborative search mechanisms is used to solve the problem and generate the optimal commutation scheme.

[0038] The optimal commutation scheme is parsed into commutation control commands and the commutation operation is executed to complete the dynamic control of three-phase imbalance.

[0039] This method is based on a dynamic control model with dual objectives of minimizing three-phase current imbalance and minimizing the number of switching actions, and is solved using an improved genetic algorithm. The execution process of the improved genetic algorithm includes:

[0040] An empirical knowledge-guided population reproduction mechanism is adopted, the mechanism comprising: establishing and dynamically updating a gene pool for storing the frequencies of desirable phasing genes; and, during crossover operations, selectively generating phasing genes for offspring individuals based on the frequency information recorded in the gene pool.

[0041] A Pareto-optimal population selection mechanism is adopted. In each generation of evolution, a non-dominated sorting is performed on the population after the merging of the parent and offspring, and individuals with both superiority and diversity are selected to enter the next generation based on their Pareto front rank and crowding distance.

[0042] A global and local collaborative search mechanism is adopted. During the algorithm iteration process, when the preset triggering conditions are met, the tabu search algorithm is called to perform a neighborhood search on the Pareto optimal individuals in the current population, so as to obtain higher quality local optimal solutions and enhance the quality of the global population.

[0043] In the aforementioned knowledge-guided population reproduction mechanism, the gene pool is a The matrix, where N is the total number of loads participating in the regulation, and the matrix is... Column record number The probability of a load being connected to phases A, B, and C is calculated and updated based on the frequency of different phase sequence genes at each load position in individuals with a Pareto front level of 1 in the current population.

[0044] Furthermore, the gene bank is an N×3 probability matrix, where N is the total number of loads involved in regulation, and each column of the matrix records the probability of the corresponding load accessing phases A, B, and C.

[0045] Based on the updated gene pool, offspring individuals are generated with a bias during population crossover operations to increase the probability of inheriting superior genes.

[0046] Furthermore, the crossover operation includes two strategies for random selection:

[0047] Genes are retained based on the same gene loci of the parent individuals. For different gene loci, new genes are randomly generated according to the probability of the gene pool, or an arithmetic crossover operation is performed on the phase sequence vectors of the two parent individuals. Non-integer genes generated are rounded and constrained to maintain the validity of gene values.

[0048] Furthermore, the improved genetic algorithm employs a Pareto-optimal population selection mechanism in each generation of evolution. It divides the merged population of parents and offspring into multiple front sets through non-dominated sorting, and selects individuals that are both superior and maintain diversity to enter the next generation based on their Pareto rank and crowding distance.

[0049] Furthermore, the Pareto optimal selection mechanism dynamically determines the evolutionary retention scale based on the number of individuals at the first frontier level. When the number of individuals exceeds the preset scale, individuals with larger crowding distances are retained first to ensure a diverse distribution of the solution space.

[0050] In the global and local collaborative search mechanism, the triggering condition is: the current iteration number of the algorithm reaches half of the preset total iteration number; the initial solution of the tabu search algorithm is randomly selected from individuals with the Pareto front level of the current population as level 1, so as to ensure that the starting point of the local search has high quality.

[0051] The objective function of the dynamic control model is specifically:

[0052] Minimize three-phase current imbalance ,in ;

[0053] And, minimize the total number of switching actions. ,in This represents the total number of loads whose states differ before and after commutation.

[0054] Example 2

[0055] This embodiment provides a dynamic control method for three-phase imbalance in a distribution network based on a hybrid intelligent optimization algorithm, including:

[0056] Based on the current values ​​and initial phase sequence status of each single-phase load in the distribution substation, a dynamic control model is established with the dual objectives of minimizing the three-phase current imbalance and minimizing the number of switching actions.

[0057] Based on the dynamic control model, an improved genetic algorithm that integrates empirical knowledge guidance, Pareto optimal selection, and global and local collaborative search mechanisms is used to solve the problem and generate the optimal commutation scheme.

[0058] The optimal commutation scheme is parsed into commutation control commands and the commutation operation is executed to complete the dynamic control of three-phase imbalance.

[0059] Reference Figure 1 The dynamic control system of this invention is applied to a low-voltage distribution substation that includes an intelligent phase-switching terminal and several distributed load phase-switching switches. The intelligent phase-switching terminal, as a central processing module, is responsible for executing the method of this invention and for exchanging data and issuing commands with each load phase-switching switch via wireless communication (such as GPRS).

[0060] The first step is to establish a dynamic control model;

[0061] First, the intelligent commutation terminal collects the effective current values ​​of the N single-phase loads that need to participate in regulation at the current moment, forming a load current column vector:

[0062] ;

[0063] in, This is the column vector of load currents. Let T be the effective current value of the N single-phase loads that need to be regulated at the current moment, and let T be the time.

[0064] Simultaneously, the initial phase sequence state of these loads before commutation is acquired and encoded into a... vector , where the vector's th element , respectively representing the load Initially connect phases A, B, and C.

[0065] The dynamic control strategy model in this embodiment aims to find an optimal post-commutation phase sequence state vector. To achieve this, the model contains two optimization objectives:

[0066] Objective 1: Minimize the three-phase current imbalance.

[0067] ;

[0068] in, It is based on the phase sequence state vector and load current vector The calculated total three-phase current after commutation The objective function value for the three-phase current imbalance is... As an intermediate variable for calculating the three-phase current imbalance, This is the average value of the three-phase current.

[0069] Objective 2: Minimize the number of times the load switching switch operates.

[0070] ;

[0071] in, The objective function value for the number of load commutator switches actuated in the dynamic control model. This represents the number of times the master switch has been activated. Let be the variable used to determine whether the phase sequence of the i-th load changes before and after commutation. The total number of loads participating in the regulation, i ranges from 1 to N.

[0072] Among them, when hour, ;otherwise . This represents the number of elements in the state vector that differ before and after commutation, i.e., the total number of switching actions.

[0073] Step 2: Hybrid intelligent optimization solution method based on NSGA-TS;

[0074] Reference Figure 2This invention uses the NSGA-TS algorithm to solve the above-mentioned bi-objective model. Its core lies in integrating multiple innovative mechanisms into a unified genetic algorithm framework.

[0075] First, population initialization and encoding: a certain number of individuals (e.g., 50) are randomly generated, each individual representing a possible commutation scheme, i.e., one... Phase sequence state vector Each gene locus (element) can have a value of 1, 2, or 3.

[0076] Secondly, there is the population reproduction mechanism guided by experiential knowledge;

[0077] This mechanism is one of the key differences between this invention and standard genetic algorithms. It improves efficiency by establishing a gene pool and using the "experience" accumulated during the search process to guide subsequent searches.

[0078] The establishment and updating of gene banks: A gene bank is a... The probability matrix. After each generation of evolution, the system counts all individuals in the current population that are located at the first Pareto front (i.e., the current optimal solution set). For the... For each load position, count the number of individuals with gene 1 (phase A), 2 (phase B), and 3 (phase C) among these optimal individuals, and calculate the corresponding frequencies. This frequency is then updated to the th load position in the gene pool. The probability values ​​of the sequence. This allows the gene pool to dynamically reflect which sequence combinations appear to be "better" at the moment.

[0079] Experience-guided crossover: When generating offspring, this invention employs two crossover strategies and randomly selects one to balance "exploration" and "utilization." Strategy one retains identical genes from the parents; for different gene loci, generation is no longer completely random but follows the probability distribution of corresponding positions in the gene pool, making offspring more likely to inherit superior gene fragments. Strategy two uses arithmetic crossover to explore new solution spaces with a degree of randomness, compensating for the potential over-conservatism of Strategy one. This combined strategy utilizes historical experience while ensuring the breadth of the search.

[0080] Mutation operation: A random three-point mutation strategy is adopted to perform small-amplitude random perturbations on individuals in order to maintain population diversity. This is a key step to prevent premature convergence of the algorithm.

[0081] Then there is the population selection mechanism based on Pareto optimality;

[0082] After reproduction produces offspring, the parent and offspring are merged. A fast non-dominated sorting is then performed on this large population. First, all individuals are assigned to different fronts based on Pareto dominance. Individuals in Front 1 are optimal in both objectives (imbalance and number of actions) and are not dominated by any other individual. Then, for individuals within the same front, their sparse distribution in the solution space is measured by calculating their crowding distance. Finally, all individuals are sorted according to front rank (lower rank is better) and crowding distance (greater distance is better), and a predetermined number of high-performing individuals are selected for the next generation. This mechanism ensures that the algorithm's evolutionary direction moves towards the Pareto optimal front and maintains solution diversity, providing decision-makers with a range of high-quality alternatives with different focuses.

[0083] Finally, there is the global and local collaborative search mechanism;

[0084] To overcome the tendency of standard genetic algorithms to get trapped in local optima, this invention introduces the Tabu Search (TS) algorithm for local depth optimization. When the algorithm has iterated halfway through (at which point the global search has been sufficiently performed), the system randomly selects a high-quality solution from the current population's Front 1 as the initial solution for TS. The TS algorithm refines this solution through its unique neighborhood search and "tabu list" mechanism (avoiding repeated searches of already visited solutions in the short term). Simultaneously, its "disregard criterion" allows breaking taboos under certain circumstances, ensuring that the algorithm eventually converges to a high-quality local optimum. Replacing this TS-optimized solution back into the population effectively improves the overall quality of the population, achieving synergy between "global exploration" and "local development," significantly enhancing the algorithm's stability and the quality of the final solution.

[0085] On the other hand, this embodiment also provides an electronic device, including a memory, a processor, and a computing program stored in the memory and executable on the processor, wherein the processor implements the method when executing the computing program.

[0086] On the other hand, this embodiment also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the method.

[0087] Example 3

[0088] This embodiment provides a dynamic control method for three-phase imbalance in a distribution network based on a hybrid intelligent optimization algorithm, including:

[0089] This embodiment takes a distribution substation containing 15 controllable single-phase loads as an example, whose initial three-phase currents are as follows: The three-phase current imbalance is as high as 80%.

[0090] The method described in this embodiment is used for regulation, with the NSGA-TS algorithm parameters set as follows: population size 50, number of iterations 50. (Refer to...) Figure 3 The convergence curve of the NSGA-TS algorithm (red curve in the figure) shows that its convergence speed and final fitness are superior to standard GA, GA-TS, SA and other algorithms, proving the superiority of the mechanism described in this invention.

[0091] After approximately 3.9 seconds of calculation, the algorithm outputs the optimal control strategy: switch loads 1 and 9, originally connected to phase A, to phase B. After executing this strategy, as follows... Figure 4 As shown, the three-phase current of the transformer area becomes The three-phase current imbalance was reduced to 0%, while the total number of switching operations was only 2. This result demonstrates that the method of this invention can quickly and accurately find the optimal commutation strategy that balances governance effectiveness and economic cost.

[0092] The above are merely preferred embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A dynamic control method for three-phase imbalance in a distribution network based on a hybrid intelligent optimization algorithm, characterized in that, include: Based on the current values ​​and initial phase sequence status of each single-phase load in the distribution substation, a dynamic control model is established with the dual objectives of minimizing the three-phase current imbalance and minimizing the number of switching actions. Based on the dynamic control model, an improved genetic algorithm that integrates empirical knowledge guidance, Pareto optimal selection, and global and local collaborative search mechanisms is used to solve the problem and generate the optimal commutation scheme. The optimal commutation scheme is parsed into commutation control commands and the commutation operation is executed to complete the dynamic control of three-phase imbalance.

2. The method for dynamic control of three-phase imbalance in distribution networks based on a hybrid intelligent optimization algorithm according to claim 1, characterized in that, The experiential knowledge-guided mechanism includes establishing and dynamically updating a gene pool, and updating the corresponding probability information in the gene pool based on the frequency of occurrence of different phase sequence genes at each load position in individuals with a Pareto front level of 1 in the current population.

3. The dynamic control method for three-phase imbalance in distribution networks based on a hybrid intelligent optimization algorithm according to claim 2, characterized in that, The gene bank is an N×3 probability matrix, where N is the total number of loads involved in regulation, and each column of the matrix records the probability of the corresponding load being connected to phases A, B, and C. Based on the updated gene pool, offspring individuals are generated with a bias during population crossover operations to increase the probability of inheriting superior genes.

4. The dynamic control method for three-phase imbalance in distribution networks based on a hybrid intelligent optimization algorithm according to claim 3, characterized in that, The crossover operation includes two strategies for random selection: Genes are retained based on the same gene loci of the parent individuals. For different gene loci, new genes are randomly generated according to the probability of the gene pool, or an arithmetic crossover operation is performed on the phase sequence vectors of the two parent individuals. Non-integer genes generated are rounded and constrained to maintain the validity of gene values.

5. The method for dynamic control of three-phase imbalance in distribution networks based on a hybrid intelligent optimization algorithm according to claim 1, characterized in that, The improved genetic algorithm employs a Pareto-optimal population selection mechanism in each generation of evolution. It divides the merged population of parents and offspring into multiple front sets through non-dominated sorting, and selects individuals that are both superior and maintain diversity to enter the next generation based on their Pareto rank and crowding distance.

6. The dynamic control method for three-phase imbalance in distribution networks based on a hybrid intelligent optimization algorithm according to claim 1, characterized in that, The Pareto optimal selection mechanism dynamically determines the evolutionary retention scale based on the number of individuals at the first frontier level. When the number of individuals exceeds the preset scale, individuals with larger crowding distances are retained first to ensure a diverse distribution of the solution space.

7. The method for dynamic control of three-phase imbalance in distribution networks based on a hybrid intelligent optimization algorithm according to claim 1, characterized in that, The global and local collaborative search mechanism includes, during the algorithm iteration process, when the number of iterations reaches half of the total number of iterations, randomly selecting an individual from the first Pareto front of the current population as the initial solution of the tabu search algorithm, and performing a neighborhood search on the individual to improve the quality of the local optimum.

8. The method for dynamic control of three-phase imbalance in distribution networks based on a hybrid intelligent optimization algorithm according to claim 1, characterized in that, The objective function of the dynamic control model includes: Calculate the three-phase current imbalance based on the phase sequence state of each load before and after commutation and minimize this value; Calculate the total number of commutation actions based on the difference in load conditions before and after commutation, and minimize this number.

9. An electronic device comprising a memory, a processor, and a computing program stored in the memory and executable on the processor, characterized in that, When the processor executes the computing program, it implements the method of any one of claims 1-8.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the method of any one of claims 1-8.

Images