Three-phase unbalance optimization method based on gated recurrent unit and hyper-heuristic algorithm

Abstract

The application discloses a three-phase imbalance optimization method based on a gating cycle unit and a super heuristic algorithm, and has the characteristics that: circuit data and resource data of a transformer area are acquired, and a three-phase imbalance circuit optimization model is constructed based on the data; an initial population is generated based on a Tent chaotic mapping strategy, and an improved numerical optimization algorithm is utilized by using an elite selection strategy and a differential evolution operator to obtain accurate prediction results; based on the prediction results, the selection strategy of the super heuristic algorithm is enriched by using the idea of reinforcement learning and the greedy mechanism, so that the low-level heuristic algorithm in the problem domain intelligently selects a heuristic algorithm for optimization, and then the three-phase imbalance circuit optimization model is solved. The application improves the prediction accuracy of the gating cycle unit by improving the numerical optimization algorithm, realizes accurate prediction of the transformer area load in different periods, and at the same time, the selection strategy of the super heuristic algorithm is enriched by using the idea of reinforcement learning and the greedy mechanism, and then the optimization effect of the three-phase imbalance circuit is improved.

Description

Technical Field

[0001] This invention relates to the field of three-phase imbalance optimization technology in transformer substations, and in particular to a three-phase imbalance optimization method based on improved gated cyclic unit prediction and hyperheuristic algorithm. Background Technology

[0002] The development of science and technology has led to an increasing demand for electricity. The types of loads in power distribution networks are becoming more and more complex, and problems such as three-phase imbalance, insufficient reactive power, and harmonics are becoming increasingly prominent. Deep learning and heuristic algorithms are widely used to solve these problems and have effectively addressed them.

[0003] Three-phase unbalanced circuits not only affect the lifespan of electrical appliances but also the communication quality of communication systems. Existing research mainly uses deep learning and heuristic algorithms to effectively identify users in low-voltage distribution areas, or uses deep learning to predict load and then employs heuristic algorithms to optimize three-phase unbalanced circuits based on the predicted data. However, different heuristic algorithms have different search strategies, resulting in variations in the results of each run. Existing research also only uses a single heuristic algorithm to solve the problem, failing to find a more effective heuristic algorithm.

[0004] Therefore, a three-phase imbalance optimization method based on improved gated cyclic unit prediction and hyperheuristic algorithm is proposed. Summary of the Invention

[0005] The purpose of this section is to outline some aspects of embodiments of the present invention and to briefly describe some preferred embodiments. Simplifications or omissions may be made in this section, as well as in the abstract and title of this application, to avoid obscuring the purpose of these documents; however, such simplifications or omissions should not be construed as limiting the scope of the invention.

[0006] In view of the above-mentioned problems, the present invention is proposed.

[0007] Therefore, the technical problems solved by this invention are: the load prediction results have errors with the actual situation, the solution efficiency is low, the power supply duration is unstable, and the phase sequence adjustment of the transformer area is insufficient.

[0008] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a three-phase imbalance optimization method based on gated cyclic units and hyperheuristic algorithms, comprising: Obtain circuit and resource data for the transformer area, and construct a three-phase unbalanced circuit optimization model based on the data; The initial population is generated based on the Tent chaotic mapping strategy, and the arithmetic optimization algorithm is improved by using the elite selection strategy and differential evolution operator to obtain accurate prediction results. Based on the prediction results, the superheuristic algorithm selection strategy is enriched by using reinforcement learning and greedy mechanisms, so that the low-level heuristic algorithm in the problem domain can be intelligently selected for optimization, thereby solving the optimization model of the three-phase unbalanced circuit.

[0009] As a preferred embodiment of the three-phase imbalance optimization method based on gated loop units and hyperheuristic algorithms described in this invention, wherein: the construction of the three-phase imbalance circuit optimization model includes: An optimization model for a three-phase unbalanced circuit is constructed with the objectives of three-phase unbalance, commutation frequency, and losses in each branch.

[0010] As a preferred embodiment of the three-phase imbalance optimization method based on gated loop units and hyperheuristic algorithms described in this invention, the construction of the three-phase imbalance circuit optimization model further includes: The area has A side road, One load, The numbers 1, 2, and 3 represent the phase sequence, with 1, 2, and 3 indicating the U, V, and W phase circuits, respectively. Indicates a branch exist Phase current, Indicates a branch impedance, Indicates a branch The three-phase imbalance Indicates the commutation period. Indicates load Commutation frequency within period T Indicates a branch neutral line current, Indicates a branch At the neutral line loss Indicates load On the side road of Phase current This represents the minimum satisfactory duration of the phase.

[0011] branch road neutral line current The calculation formula is: (1) in, Indicates phase sequence, Indicates a branch exist Phase current; branch road neutral line current With load On the side road of Phase current Closely related: (2) in, (3), Indicates load On the side road of Phase current This represents the minimum satisfactory duration of the phase. Use branch roads The ratio of the difference between the maximum and average values ​​of three-phase currents to the average value represents the branch current. Three-phase imbalance The calculation formula is as follows: (4) in, Indicates a branch The current in phase U, Indicates a branch exist Phase current, Indicates a branch exist Phase current; The loss on the neutral line of a branch is the product of the resistance and the square of the corresponding neutral line current. Therefore, the branch... The loss at the neutral line is: (5) in, Indicates a branch impedance, Indicates a branch Neutral current; Calculate the load for each day based on the commutation cycle. commutation number : (6) in, Indicates the commutation period; By comparison and Whether the phases at different times are the same determines the load. Whether or not commutation occurs, the formula for calculating the number of load commutations is: = (7) in, (8) After determining the target weights using the analytic hierarchy process (AHP), the objective function is constructed as follows: (9) in, Indicates the branch road number, Indicates the number of branches. This indicates the load designation, and M indicates the load quantity. , , These represent the proportions of three-phase imbalance, center line loss, and commutation frequency in the objective function, respectively. , Indicates a branch The three-phase imbalance Indicates a branch At the neutral line loss This indicates the number of load commutations.

[0012] As a preferred embodiment of the three-phase imbalance optimization method based on gated cyclic units and hyperheuristic algorithms described in this invention, wherein: the generation of the initial population based on the Tent chaotic mapping strategy includes: The arithmetic optimization algorithm parameters are initialized, and the Tent chaotic mapping strategy is introduced to generate an initial population, which is continuously updated. The initial population is then merged with the updated population, and the fitness function value is calculated, along with updates to the acceleration coefficient MOA and probability coefficient MOP. Based on random numbers... The relationship with the acceleration coefficient MOA can be used to determine whether the stage is the exploration or development stage. The formula used to update the initial population is expressed as: (10) in, (11), Indicates the first Generations of populations Indicates the first Generational population; The formula for calculating the fitness function value is expressed as follows: (12) in, Represents the fitness function value The number of test data. For the first The true value of each test data point. For the first Predicted values ​​for each test data set; The formula for updating the acceleration coefficient MOA is expressed as follows: (13) Where n represents the number of iterations, Min represents the maximum number of iterations, Max represents the minimum value of the acceleration function, and Min represents the maximum value of the acceleration function. The formula for updating the probability coefficient MOP is expressed as follows: (14) Where n represents the number of iterations, Indicates the maximum number of iterations. Indicates a sensitive parameter.

[0013] As a preferred embodiment of the three-phase imbalance optimization method based on gated cyclic units and hyperheuristic algorithms described in this invention, wherein: the method based on random numbers The relationship with the acceleration coefficient MOA determines whether to enter the exploration or development phase, including: When entering the exploration phase, random numbers are generated. Update the position according to the following formula: (15) Where μ is the control parameter for adjusting the search process, with a value of 0.499, and ε is the minimum value. Indicates the first Generations of populations Indicates the first Generations of populations Represents the probability coefficient. upper bound of the population Population lower bound; when When entering the development phase, random numbers are generated. Update the position according to the following formula: (16) Where μ is the control parameter for adjusting the search process, with a value of 0.499, and ε is the minimum value. Indicates the first Generations of populations Indicates the first Generations of populations Represents the probability coefficient. upper limit of population, Lower bound of the population.

[0014] As a preferred embodiment of the three-phase imbalance optimization method based on gated cyclic units and hyperheuristic algorithms described in this invention, wherein: the improvement of the arithmetic optimization algorithm using an elite selection strategy and differential evolution operators includes: The elite selection strategy is used to find the individuals with the best fitness in the initial and new populations, and then the differential evolution operator is used to perform selection, crossover and mutation operations on the initial and new populations.

[0015] As a preferred embodiment of the three-phase imbalance optimization method based on gated cyclic units and hyperheuristic algorithms described in this invention, wherein the selection, crossover, and mutation operations include: New individuals are generated after the exploration or development phase. Then, the fitness function value of the new individuals is calculated, and a new set of individuals is generated. And recalculate based on the fitness function value Sort the elements within the specified range. .turn up The individual with the best fitness At the same time, three individuals are randomly selected from the remaining individuals. , and The corresponding fitness function values ​​are respectively , and ,at the same time ; Generate mutation vectors: (17) in, This represents the scaling factor, and its typical value range is... ; scaling factor The calculation formula is expressed as: (18) in, , , , and Individuals , and The fitness function value; right Except The remaining individuals are crossovered according to the following formula: (19) Where cr is the crossover probability between [0,1], and the mean of the current population fitness function. Maximum value and minimum value Related, generally , The specific formula is as follows: (20) in, For the first The true value of each test data point; This represents the mean of the current population fitness function; For the maximum value of the fitness function, This represents the minimum value of the fitness function.

[0016] As a preferred embodiment of the three-phase imbalance optimization method based on gated recurrent units and hyperheuristic algorithms described in this invention, wherein: the enrichment of the hyperheuristic algorithm selection strategy using reinforcement learning ideas and greedy mechanisms includes: If the dataset is sufficient, a reinforcement learning approach is adopted, and the optimal low-level heuristic algorithm is selected based on the effective information accumulated in the early stages. If the dataset is insufficient, a greedy mechanism is used to select the low-level heuristic algorithm that can maximize the improvement of the current solution and construct a new algorithm.

[0017] As a preferred embodiment of the three-phase imbalance optimization method based on gated cyclic units and hyperheuristic algorithms described in this invention, wherein: determining whether the dataset is sufficient includes: Minimum sample set required for computational reinforcement learning When the number of samples is greater than or equal to At that time, reinforcement learning methods are used; the calculation formula is expressed as: (twenty one) in, This represents the minimum number of samples required for reinforcement learning. The complexity of a sample set is generally expressed as... .

[0018] As a preferred embodiment of the three-phase imbalance optimization method based on gated loop units and hyperheuristic algorithms described in this invention, wherein: solving the three-phase imbalance circuit optimization model includes: The Metropolis acceptance criterion is used to determine whether to accept the newly generated individual. If accepted, the individual is updated and the process proceeds to determine whether the termination calculation condition has been met. If not accepted, return to the step of determining whether to accept; If the conditions for terminating the calculation are met, output and record the optimization result; otherwise, return to the strategy selection step.

[0019] The beneficial effects of this invention are as follows: The method described in this invention uses an improved arithmetic optimization algorithm to optimize the initial parameters of the gated loop unit, thereby increasing the prediction accuracy and making the load prediction results of the transformer area more consistent with the actual situation; it utilizes various heuristic algorithms in the problem domain of the hyperheuristic algorithm to optimize the three-phase unbalanced circuit according to the established model, and makes full use of the heuristic algorithms in the problem domain through the selection strategy of the control domain and the solution acceptance strategy, and continuously trains the relevant parameters of each heuristic algorithm based on the idea of ​​reinforcement learning to improve the solution speed, thereby increasing the stability of power supply duration and solving problems such as insufficient phase sequence adjustment in the transformer area. Attached Figure Description

[0020] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. Wherein: Figure 1 The overall flowchart of a three-phase imbalance optimization method based on gated cyclic units and hyperheuristic algorithms provided in an embodiment of the present invention is shown below. Figure 2 The flowchart shows the improved arithmetic optimization algorithm in the three-phase imbalance optimization method based on gated cyclic units and hyperheuristic algorithms provided in the first embodiment of the present invention. Figure 3 A diagram illustrating a hyperheuristic algorithm framework based on a greedy strategy and reinforcement learning, provided for the second embodiment of the present invention. Figure 4 A comparison of the load balancing rate of the three-phase imbalance optimization method based on gated loop units and hyperheuristic algorithms provided in the second embodiment of the present invention under different numbers of tasks; Figure 5 A comparison diagram of the U-phase bus load prediction algorithm provided in the second embodiment of the present invention. Detailed Implementation

[0021] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0022] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0023] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0024] This invention is described in detail with reference to the schematic diagrams. When detailing the embodiments of this invention, for ease of explanation, the cross-sectional views illustrating the device structure may be partially enlarged, not adhering to the usual scale. Furthermore, the schematic diagrams are merely examples and should not be construed as limiting the scope of protection of this invention. In actual fabrication, the three-dimensional spatial dimensions of length, width, and depth should be included.

[0025] Furthermore, in the description of this invention, it should be noted that the terms "upper," "lower," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. These terms are used solely for the convenience of describing the invention and for simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the invention. In addition, the terms "first," "second," or "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0026] Unless otherwise explicitly specified and limited, the terms "installation," "connection," and "joining" in this invention should be interpreted broadly. For example, they can refer to fixed connections, detachable connections, or integral connections; similarly, they can refer to mechanical connections, electrical connections, or direct connections, or indirect connections through an intermediate medium, or internal connections between two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0027] Example 1 Reference Figures 1-4 As an embodiment of the present invention, a three-phase imbalance optimization method based on gated cyclic units and hyperheuristic algorithms is provided, comprising: S1: Obtain circuit data and resource data for the transformer area, and construct a three-phase unbalanced circuit optimization model based on the data; It should be noted that the circuit data of the transformer area includes the connection method of the circuit within the transformer area, the number of loads, the number of branches, the relationship between branches and loads, the phase and current of each load in different branches, etc.; the resource data includes the loads within the transformer area at different times.

[0028] Furthermore, based on the circuit data and resource data of the transformer area, a three-phase imbalance optimization model is constructed with the three-phase imbalance degree, commutation frequency and loss of each branch as the objectives. Furthermore, the area has A side road, One load, The numbers 1, 2, and 3 represent the phase sequence, with 1, 2, and 3 indicating the U, V, and W phase circuits, respectively. Indicates a branch exist Phase current, Indicates a branch impedance, Indicates a branch The three-phase imbalance Indicates the commutation period. Indicates load Commutation frequency within period T Indicates a branch neutral line current, Indicates a branch At the neutral line loss Indicates load On the side road of Phase current This represents the minimum satisfactory duration of the phase.

[0029] branch road neutral line current The calculation formula is: (1) branch road neutral line current With load On the side road of Phase current Closely related: (2) in, (3) Use branch roads The ratio of the difference between the maximum and average values ​​of three-phase currents to the average value represents the branch current. Three-phase imbalance The calculation formula is as follows: (4) The loss on the neutral line of a branch is the product of the resistance and the square of the corresponding neutral line current. Therefore, the branch... The loss at the neutral line is: (5) Calculate the load for each day based on the commutation cycle. commutation number : (6) By comparison and Whether the phases at different times are the same determines the load. Whether or not commutation occurs, the formula for calculating the number of load commutations is: = (7) in, (8) After determining the target weights using the Analytic Hierarchy Process (AHP), the objective function is constructed as follows: (9) S2: The initial population is generated based on the Tent chaotic mapping strategy, and the improved arithmetic optimization algorithm (IAOA) is improved by using the elite selection strategy and differential evolution operator to obtain accurate prediction results; Furthermore, the original data is input and decomposed into a finite set of IMFs based on the frequency components of the original signal using variational mode decomposition (VMD) technology; Furthermore, a search space of size N and dimension dim is defined, and MOA and MOP are input simultaneously. A population is randomly generated, and the initial population is updated according to the following formula: (10) in, (11) Furthermore, the mean squared error during GRU training is used as a formula to calculate the fitness function value, expressed as follows: (12) in, Represents the fitness function value The number of test data. For the first The true value of each test data point. For the first Predicted values ​​for each test data point.

[0030] Furthermore, after the fitness function value is calculated, based on the iteration number n and the maximum iteration number... The minimum and maximum values ​​of the acceleration function (Min and Max) are used to update the acceleration factor (MOA). The minimum value (Min) is typically 0.2, and the maximum value (Max) is typically 1. The MOA calculation formula is as follows: (13) Based on the iteration number n and the maximum iteration number Sensitive parameters The update probability coefficient (MathOptimizer probability, MOP) is calculated using the following formula: (14) Furthermore, based on random numbers The relationship with MOA determines whether to enter the exploration or development phase; the criterion is: when When entering the exploration phase, random numbers are generated. μ is a control parameter for adjusting the search process, with a value of 0.499, and ε is the minimum value. The position is updated according to the following formula: (15) when When entering the development phase, random numbers are generated. Update the position according to the following formula: (16) Furthermore, new individuals are generated after the exploration or development phase. Then, the fitness function value of the new individuals is calculated, and a new set of individuals is generated. And recalculate based on the fitness function value Sort the elements within the specified range. Searching based on elite selection strategies The individual with the best fitness At the same time, three individuals were randomly selected. , and The corresponding fitness function values ​​are respectively , and ,at the same time .

[0031] Furthermore, differential evolution operators are used to perform selection, crossover, and mutation operations on the initial and new populations; the resulting mutation vector is represented as: (17) in, This represents the scaling factor, and its typical value range is... , , and Three individuals were randomly selected during the representation process; scaling factor The calculation formula is: (18) in, , , , and Three individuals were randomly selected. , and The fitness function value; right Individuals in the data are crossovered according to the following formula: (19) Where cr is the crossover probability between [0,1], and the mean of the current population fitness function. Maximum value and minimum value Related, generally , The specific formula is as follows: (20) Furthermore, comparing the worst individual among the new individuals with the individual... The fitness function value of an individual If the fitness function value of an individual is less than the fitness function value of the worst individual, then use... Replace the worst individual; otherwise, do not perform the replacement operation.

[0032] Furthermore, it checks whether the number of iterations is greater than the maximum number of iterations. If it is, it exits the loop; otherwise, it returns to the step of calculating the fitness function.

[0033] Furthermore, after improving the GRU by setting the initial weights of the GRU through IOA, the improved GRU is then used to predict the IMF and the prediction results are output. Furthermore, by improving GRU to aggregate the forecast results of each IMF, an overall forecast result of the original load sequence is generated; S3: Based on the prediction results, the superheuristic algorithm selection strategy is enriched by using reinforcement learning and greedy mechanisms, so that the low-level heuristic algorithm in the problem domain can intelligently select the heuristic algorithm for optimization, thereby solving the optimization model of the three-phase unbalanced circuit.

[0034] Furthermore, the population is randomly initialized, and a low-level heuristic algorithm is used to solve the optimization problem; It should be noted that the low-level heuristic algorithms include traditional algorithms for solving three-phase imbalance problems as well as advanced algorithms from the last three years.

[0035] Furthermore, determine whether the dataset is sufficient, and select a strategy based on the determination result; Minimum number of samples required for reinforcement learning The calculation formula is expressed as follows: (twenty one) The minimum sample set required for reinforcement learning is calculated using formula (21). When the number of samples is greater than or equal to In such cases, reinforcement learning methods can be employed.

[0036] Otherwise, a greedy mechanism is used, that is, a low-level heuristic algorithm that maximizes the improvement of the current solution is selected to construct a new algorithm; Furthermore, the Metropolis acceptance criterion is used to determine whether to accept newly generated individuals. The specific criteria for this decision are as follows: (twenty two) in, This represents the fitness function value of the new population. This represents the function value of the individual before the adaptation was updated. Indicates the current temperature; Meanwhile, from formula (9), we can see that the fitness function of the hyperheuristic algorithm is calculated as follows: (twenty three) in, Indicates the branch road number, Indicates the number of branches. This indicates the load designation, and M indicates the load quantity. , , These represent the proportions of three-phase imbalance, center line loss, and commutation frequency in the objective function, respectively. , Indicates a branch The three-phase imbalance Indicates a branch At the neutral line loss This indicates the number of load commutations.

[0037] Generate random numbers in the range [0,1]. ,when Then receive Otherwise, do not accept, and return to the acceptance / rejection step for re-evaluation.

[0038] It should be noted that adopting the Metropolis acceptance criterion allows the next generation of the population to maintain good fitness.

[0039] Furthermore, if a new individual is accepted, it is determined whether the termination calculation condition is met. If it is met, the optimization result is output and recorded. If it is not met, the process returns to the step of determining whether the dataset is sufficient for re-evaluation.

[0040] It should be noted that the criterion for terminating the calculation is that the maximum number of iterations has been reached.

[0041] Example 2 Reference Figure 5 As an embodiment of the present invention, a three-phase imbalance optimization method based on gated cyclic units and hyperheuristic algorithms is provided. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments.

[0042] Input the data at 00:00 on June 2, 2021 as the starting point for the phase sequence optimization algorithm. The algorithm first calls the improved GRU prediction algorithm (IAOA-GRU) to obtain the predicted load value of the transformer area from 0:00 to 06:00, and then compares it with the traditional GUR algorithm. The predicted load of the bus is shown using phase U as an example, and the prediction results are as follows. Figure 5 As shown in Table 1, the load forecast data for each time point are as follows.

[0043] Table 1 Load forecast values ​​for each algorithm

[0044] Based on the load forecast results, the three-phase unbalanced circuit was optimized using the hyperheuristic algorithm proposed in this invention. The optimization results are shown in Table 2.

[0045] Table 2 Relevant Indicators of Optimization Results

[0046] The above optimization results show that the method described in this invention can improve the accuracy of load prediction for users in the transformer area, make the simulation more closely resemble the actual industrial environment, and effectively reduce the three-phase imbalance and commutation frequency of the circuit, thereby increasing circuit stability.

[0047] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A three-phase imbalance optimization method based on gated cyclic units and hyperheuristic algorithms, characterized in that, include: Obtain circuit and resource data for the transformer area, and construct a three-phase unbalanced circuit optimization model based on the data; The initial population is generated based on the Tent chaotic mapping strategy, and the arithmetic optimization algorithm is improved by using the elite selection strategy and differential evolution operator to obtain accurate prediction results. Based on the prediction results, the superheuristic algorithm selection strategy is enriched by using reinforcement learning and greedy mechanisms, so that the low-level heuristic algorithm in the problem domain can be intelligently selected for optimization, thereby solving the optimization model of the three-phase unbalanced circuit. The aforementioned strategy for enriching hyperheuristic algorithm selection using reinforcement learning and greedy mechanisms includes: Determine if the dataset is sufficient. If the dataset is sufficient, use reinforcement learning to select the low-level heuristic algorithm with the best optimization effect based on the effective information accumulated in the early stage. If the dataset is insufficient, use a greedy mechanism to select the low-level heuristic algorithm that can maximize the improvement of the current solution and construct a new algorithm. The determination of whether the dataset is sufficient includes: Minimum sample set required for computational reinforcement learning When the number of samples is greater than or equal to At that time, reinforcement learning methods are used; the calculation formula is expressed as: （21） in, This represents the minimum number of samples required for reinforcement learning. Indicates the complexity of the sample set. .

2. The three-phase imbalance optimization method based on gated cyclic units and hyperheuristic algorithms as described in claim 1, characterized in that: The construction of the three-phase unbalanced circuit optimization model includes: An optimization model for a three-phase unbalanced circuit is constructed with the objectives of three-phase unbalance, commutation frequency, and losses in each branch.

3. The three-phase imbalance optimization method based on gated cyclic units and hyperheuristic algorithms as described in claim 1 or 2, characterized in that: The construction of the three-phase unbalanced circuit optimization model also includes: The area has A side road, One load, The numbers 1, 2, and 3 represent the phase sequence, with 1, 2, and 3 indicating the U, V, and W phase circuits, respectively. Indicates a branch exist Phase current, Indicates a branch impedance, Indicates a branch The three-phase imbalance Indicates the commutation period. Indicates load Commutation frequency within period T Indicates a branch neutral line current, Indicates a branch At the neutral line loss Indicates load On the side road of Phase current This represents the minimum satisfactory duration of the phase. branch road neutral line current The calculation formula is: (1) in, Indicates phase sequence, Indicates a branch exist Phase current; branch road neutral line current With load On the side road of Phase current Closely related: （2） in, (3), Indicates load On the side road of Phase current This represents the minimum satisfactory duration of the phase. Use branch roads The ratio of the difference between the maximum and average values ​​of three-phase currents to the average value represents the branch current. Three-phase imbalance The calculation formula is as follows: （4） in, Indicates a branch The current in phase U, Indicates a branch exist Phase current, Indicates a branch exist Phase current; The loss on the neutral line of a branch is the product of the resistance and the square of the corresponding neutral line current. Therefore, the branch... The loss at the neutral line is: (5) in, Indicates a branch impedance, Indicates a branch Neutral current; Calculate the load for each day based on the commutation cycle. commutation number : （6） in, Indicates the commutation period; By comparison and Whether the phases at different times are the same determines the load. Whether or not commutation occurs, the formula for calculating the number of load commutations is: = （7） in, (8) After determining the target weights using the analytic hierarchy process (AHP), the objective function is constructed as follows: （9） in, Indicates the branch road number, Indicates the number of branches. This indicates the load designation, and M indicates the load quantity. , , These represent the proportions of three-phase imbalance, center line loss, and commutation frequency in the objective function, respectively. , Indicates a branch The three-phase imbalance Indicates a branch At the neutral line loss This indicates the number of load commutations.

4. The three-phase imbalance optimization method based on gated cyclic units and hyperheuristic algorithms as described in claim 3, characterized in that: The initial population generation strategy based on the Tent chaotic mapping includes: The arithmetic optimization algorithm parameters are initialized, and the Tent chaotic mapping strategy is introduced to generate an initial population, which is continuously updated. The initial population is then merged with the updated population, and the fitness function value is calculated, along with updates to the acceleration coefficient MOA and probability coefficient MOP. Based on random numbers... The relationship with the acceleration coefficient MOA can be used to determine whether the stage is the exploration or development stage. The formula used to update the initial population is expressed as: （10） in, (11), Indicates the first Generations of populations Indicates the first Generational population; The formula for calculating the fitness function value is expressed as follows: （12） in, Represents the fitness function value The number of test data. For the first The true value of each test data point. For the first Predicted values ​​for each test data set; The formula for updating the acceleration coefficient MOA is expressed as follows: （13） Where n represents the number of iterations, Min represents the maximum number of iterations, Max represents the minimum value of the acceleration function, and Min represents the maximum value of the acceleration function. The formula for updating the probability coefficient MOP is expressed as follows: (14) Where n represents the number of iterations, Indicates the maximum number of iterations. Indicates a sensitive parameter.

5. The three-phase imbalance optimization method based on gated cyclic units and hyperheuristic algorithms as described in claim 4, characterized in that: The one based on random numbers The relationship with the acceleration coefficient MOA determines whether to enter the exploration or development phase, including: When entering the exploration phase, random numbers are generated. Update the position according to the following formula: (15) Where μ is the control parameter for adjusting the search process, with a value of 0.499, and ε is the minimum value. Indicates the first Generations of populations Indicates the first Generations of populations Represents the probability coefficient. upper limit of population, Population lower bound; when When entering the development phase, random numbers are generated. Update the position according to the following formula: (16) Where μ is the control parameter for adjusting the search process, with a value of 0.499, and ε is the minimum value. Indicates the first Generations of populations Indicates the first Generations of populations Represents the probability coefficient. upper limit of population, Lower bound of the population.

6. The three-phase imbalance optimization method based on gated cyclic units and hyperheuristic algorithms as described in claim 5, characterized in that: The improved arithmetic optimization algorithm utilizing elite selection strategy and differential evolution operator includes: The elite selection strategy is used to find the individuals with the best fitness in the initial and new populations, and then the differential evolution operator is used to perform selection, crossover and mutation operations on the initial and new populations.

7. The three-phase imbalance optimization method based on gated cyclic units and hyperheuristic algorithms as described in claim 6, characterized in that: The selection, crossover, and mutation operations include: New individuals are generated after the exploration or development phase. Then, the fitness function value of the new individuals is calculated, and a new set of individuals is generated. And recalculate based on the fitness function value Sort the elements within the specified range. ,turn up The individual with the best fitness At the same time, three individuals are randomly selected from the remaining individuals. , and The corresponding fitness function values ​​are respectively , and ,at the same time ; Generate mutation vectors: （17） in, This represents the scaling factor, with a value range of 100%. ; scaling factor The calculation formula is expressed as: (18) in, , , , and Individuals , and The fitness function value; right Except The remaining individuals are crossovered according to the following formula: (19) Where cr is the crossover probability between [0,1], and the mean of the current population fitness function. Maximum value and minimum value Related, , The specific formula is as follows: (20) in, For the first The true value of each test data point; This represents the mean of the current population fitness function; For the maximum value of the fitness function, This represents the minimum value of the fitness function.

8. The three-phase imbalance optimization method based on gated cyclic units and hyperheuristic algorithms as described in claim 7, characterized in that: The solution to the three-phase unbalanced circuit optimization model includes: The Metropolis acceptance criterion is used to determine whether to accept the newly generated individual. If accepted, the individual is updated and the process proceeds to determine whether the termination calculation condition has been met. If not accepted, return to the step of determining whether to accept; If the conditions for terminating the calculation are met, output and record the optimization result; otherwise, return to the strategy selection step.

Images