An optimization method and system for power distribution based on user requirements

Abstract

The present invention discloses an optimization method and system for power distribution based on demand response, which relates to the technical field of power management. Based on line losses, converter losses, and transformer losses, and considering the over-limit behavior of nodes, the present invention formulates a corresponding objective function to more comprehensively describe the power loss situation in the power system, assist in optimizing the power distribution instruction and achieving the expected effect, and precisely control the power distribution to the corresponding user terminals; a power consumption prediction model is constructed by combining the GRU gated unit and the SVR support vector regression model. The GRU gated unit is used to extract deep time features to improve data quality, and the SVR support vector regression model is used for power consumption prediction, which can improve the accuracy of power consumption prediction; based on the power consumption prediction value and the objective function, the power distribution index is optimized, which can meet the power consumption needs of each user terminal, reduce power losses, and prevent resource waste.

Description

Technical Field

[0001] The present invention relates to the technical field of power management, and in particular to an optimization method and system for electric energy distribution based on demand response. Background Art

[0002] In traditional power systems, energy management relies primarily on manual methods, which presents numerous challenges. For example, irrational energy distribution within the power system can lead to energy waste, negatively impacting user experience, and even causing system instability. Therefore, optimizing energy distribution within the power system to improve energy utilization, reduce energy consumption, and enhance user experience is an important and urgent task.

[0003] Currently, researchers are using deep neural networks or intelligent algorithms to assist with power management. Deep neural networks, such as those using LSTM (Long Short-Term Memory) networks or SVR (Support Vector Regression) models to assist with power demand optimization, can better capture user time series. However, these methods require high data quality. Poor data quality significantly reduces accuracy and requires a large amount of training data. Some studies have used machine learning algorithms to address power distribution issues in power systems, but these methods ignore transformer and converter losses, fail to dynamically adjust objective functions based on actual conditions, and fail to accurately predict user demand. Summary of the Invention

[0004] The object of the present invention is to provide an optimization method and system for power distribution based on demand response, so as to improve the above technical problems.

[0005] In order to achieve the above-mentioned object of the invention, the embodiment of the present invention provides the following technical solutions:

[0006] A method for optimizing electric energy distribution based on user demand, comprising:

[0007] S1. Obtaining the initial power distribution instruction of the power system;

[0008] S2. constructing a corresponding objective function based on the constraints of the power system;

[0009] S3. Build a power forecasting model; obtain user power data and input it into the power forecasting model, and output the power consumption of each user terminal in the next time period, i.e., user demand;

[0010] S4. Based on the power consumption of each user terminal in the next time period and the objective function, the initial power allocation instruction is updated to obtain an updated power allocation instruction;

[0011] S5. Optimize the power distribution instruction after the first update based on the objective function to obtain an optimized power distribution instruction.

[0012] S6. Operate the power system based on the optimized power distribution instruction to complete the optimization of power distribution.

[0013] Further, the power system includes power generation equipment, the main power grid, the distribution network, and the power consumption end; the initial power distribution instruction includes a power generation scheduling instruction, an output control instruction, and a power distribution instruction.

[0014] Further, the constraint conditions include power balance constraints, power flow security constraints, reactive power distribution constraints, and flow constraints of transmission lines.

[0015] Further, the formula of the objective function is:

[0016] ;

[0017] ;

[0018] ;

[0019] ;

[0020] ;

[0021] Wherein, represents the objective function, , respectively represent the minimum value function and the maximum value function, , , respectively represent the transformer loss, the converter loss, and the line loss, represents the penalty function, represents the summation function, , respectively represent the number of units of the power generation equipment and the number of converter operations of the distribution network, represents the converter loss of the th unit of the power generation equipment, represents the loss corresponding to the rd converter operation of the distribution network, represents the total number of nodes of the power grid corresponding to the power system, represents the th node and the th node of the power grid corresponding to the power system, , respectively represent the th node of the power grid corresponding to the power system and the The voltage value of a node, represents a cosine function, represents the th node and the th node of the power grid corresponding to the power system, and respectively represent the voltage limit coefficient and the power limit coefficient, and respectively represent the maximum voltage and the minimum voltage allowed for the th node of the power grid corresponding to the power system, represents the output power of the th unit of the power generation equipment, and respectively represent the maximum output power and the minimum output power allowed for the th unit of the power generation equipment, and respectively represent the transformer loss of the power generation equipment and the transformer loss of the distribution network, and respectively represent the no-load loss of the transformer of the power generation equipment and the no-load loss of the transformer of the distribution network, and respectively represent the load loss of the transformer of the power generation equipment and the load loss of the transformer of the distribution network.

[0022] Furthermore, the power consumption prediction model includes a data screening layer, a feature extraction layer, and a power consumption prediction layer connected in series; the feature extraction layer uses a GRU gated recurrent unit; the power consumption prediction layer uses an SVR support vector regression model.

[0023] Furthermore, the training process of the power consumption prediction model includes:

[0024] S3-1. Obtain the training user power consumption data;

[0025] The training user power consumption data includes historical meteorological data, historical user data, and historical time data; the historical meteorological data includes temperature, humidity, wind speed, and rainfall; the historical user data includes user power consumption time, user power consumption behavior, user working time, and user rest time; the historical time factors include month and holiday;

[0026] S3-2. Input the training user power consumption data into the data screening layer, calculate the corresponding Pearson correlation coefficient, sort the Pearson correlation coefficients in descending order, and output the training user power consumption data corresponding to the top 15% of the Pearson correlation coefficients;

[0027] S3-3. Input the training power consumption data output by S3-2 into the GRU (Gated Recurrent Unit) for feature extraction, and output the corresponding training time features.

[0028] S3-4. Input the training time features into the SVR (Support Vector Regression) model, and use the simulated annealing algorithm to adjust the parameters of the SVR model to complete the training of the power consumption prediction model.

[0029] Further, updating the initial power distribution instruction based on the power consumption in the next time period and the objective function to obtain the power distribution instruction after the first update includes:

[0030] Determine the remaining power of the energy storage device of the power generation equipment, and summarize the total power consumption corresponding to the power consumption in the next time period of each user terminal.

[0031] Based on the remaining power, the total power consumption and the objective function, use the particle swarm optimization algorithm to adjust the output and reactive power distribution of each unit of the power generation equipment, that is, update the power generation dispatch instruction to obtain the power distribution instruction after the first update.

[0032] Further, optimizing the power distribution instruction after the first update based on the objective function to obtain the optimized power distribution instruction includes:

[0033] S5-1. Based on the objective function and the updated power generation dispatch instruction, adjust the tap position of the regulating transformer of the power system, control the switch state of the transmission line, adjust the voltage and frequency of the power grid, the operation mode of the distribution transformer, and control the switch state of the distribution line, that is, update the output control instruction and the distribution instruction to obtain the corresponding updated output control instruction and updated distribution instruction.

[0034] S5-2. Based on the updated power generation dispatch instruction, updated output control instruction, and updated distribution instruction, use simulation software to simulate the operation process of the power system to obtain the corresponding power quality index and power generation amount.

[0035] S5-3. Judge whether the power quality index meets the regulations of the power industry standard and whether the difference between the power generation amount and the total power consumption belongs to the preset reference range; if so, use the updated power generation dispatch instruction, updated output control instruction, and updated distribution instruction as the optimized power distribution instruction; otherwise, return to S3.

[0036] An optimization system for power distribution based on user requirements is provided, which includes a data acquisition module, an objective function design module, a power consumption prediction module, and an instruction optimization module; the power consumption prediction module includes a data screening sub-module, a feature extraction sub-module, and a power consumption prediction sub-module; where:

[0037] The data acquisition module is used to obtain the initial power distribution instruction and user power consumption data of the power system;

[0038] The objective function design module is used to construct the objective function based on the constraint conditions of the power system;

[0039] The data screening sub-module is used to calculate the Pearson correlation coefficient of the user power consumption data, sort the Pearson correlation coefficients in descending order, and output the user power consumption data corresponding to the top 15% of the Pearson correlation coefficients;

[0040] The feature extraction sub-module is used to input the user power consumption data output by the data screening sub-module into the GRU gated recurrent unit and output the time feature;

[0041] The power consumption prediction sub-module is used to input the time feature into the SVR support vector regression model to obtain the user demand;

[0042] The instruction optimization module is used to optimize the initial power distribution instruction based on the objective function and the power consumption of each user end in the next time period.

[0043] The beneficial effects of the present invention are:

[0044] Based on line loss, converter loss, and transformer loss, and considering the over-limit behavior of nodes, the present invention formulates a corresponding objective function, which more comprehensively describes the power loss situation in the power system, assists in optimizing the power distribution instruction and achieving the expected effect, and accurately controls the power distribution to the corresponding user end; combines the GRU gated unit and the SVR support vector regression model to construct a power consumption prediction model, uses the GRU gated unit to extract deep time features, improves data quality, and uses the SVR support vector regression model for power consumption prediction, which can improve the accuracy of power consumption prediction; optimizes the power distribution index based on the power consumption prediction value and the objective function, can meet the power consumption needs of each user end, reduce power loss, and prevent the occurrence of resource waste. Description of the Drawings

[0045] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and should not be regarded as limiting the scope. For those of ordinary skill in the art, without creative efforts, other related drawings can also be obtained based on these drawings.

[0046] Figure 1 It is the data flow diagram in the embodiments of the present invention;

[0047] Figure 2 It is the optimized system structure diagram in the embodiments of the present invention. Detailed implementation manners

[0048] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Usually, the components of the embodiments of the present invention described and shown in the accompanying drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed present invention, but only represents the selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0049] Please refer to Figure 1 , an optimization method for power distribution based on user requirements provided in this embodiment includes:

[0050] S1. Obtain the initial power distribution instruction of the power system;

[0051] The power system includes power generation equipment, the main power grid, the distribution network, and the power consumption end; the initial power distribution instruction includes a power generation scheduling instruction, an output control instruction, and a power distribution instruction.

[0052] S2. Based on the constraint conditions of the power system, construct a corresponding objective function;

[0053] The constraint conditions include power balance constraints, power flow security constraints, reactive power distribution constraints, and flow constraints of transmission lines.

[0054] The formula corresponding to the power balance constraint is:

[0055] ;

[0056] Wherein, , , They represent the number of generating units, the total number of users, and the total number of transmission lines in the power system. 、 、 Respectively represent The active power output of each unit, Active power demand of each user, The active power loss of the transmission line, 、 、 Respectively represent The reactive power output of each unit, Reactive power demand of each user, Reactive power loss of the transmission line, Represents a summation function.

[0057] The formula corresponding to the reactive power allocation constraint is:

[0058]

[0059] in, 、 Respectively represent the first The minimum and maximum values of reactive power output of each unit, Indicates the first The absolute value of the reactive power output of each unit.

[0060] The formula corresponding to the flow constraint of the transmission line is:

[0061] ;

[0062] in, Indicates the first nodes and The rated capacity of the transmission line consisting of nodes.

[0063] The formula of the objective function is:

[0064] ;

[0065] ;

[0066] ;

[0067] ;

[0068] ;

[0069] in, represents the objective function, , represent the minimum value function and the maximum value function respectively, , , represent the transformer loss, the converter loss and the line loss respectively, represents the penalty function, represents the summation function, , represent the number of generating units of the power generation equipment and the number of converter operations of the distribution network respectively, represents the th converter loss of the th generating unit of the power generation equipment, represents the loss corresponding to the th converter operation of the distribution network, represents the total number of nodes of the power grid corresponding to the power system, represents the conductance between the th node and the , th node of the power grid corresponding to the power system, represents the voltage value of the th node and the th node of the power grid corresponding to the power system, represents the cosine function, represents the voltage phase difference between the th node and the , represent the voltage limit coefficient and the power limit coefficient respectively, , represent the maximum allowable voltage and the minimum allowable voltage of the th node of the power grid corresponding to the power system, represents the output power of the th generating unit of the power generation equipment, , represent the maximum allowable output power and the minimum allowable output power of the th generating unit of the power generation equipment respectively, , represent the transformer loss of the power generation equipment and the transformer loss of the distribution network respectively, , represent the no-load loss of the transformer of the power generation equipment and the no-load loss of the transformer of the distribution network respectively, , represent the load loss of the transformer of the power generation equipment and the load loss of the transformer of the distribution network respectively.

[0070] The objective function is formulated considering various losses, assisting in optimizing the power distribution instruction to optimize in the direction of minimizing losses, meeting the user demands at each user end with the minimum output of the power generation equipment, reducing the operation burden of the power generation equipment, extending the service life of the generator set, and achieving the power generation amount at the lowest cost.

[0071] S3. Construct an electricity consumption prediction model; obtain user electricity data and input it into the electricity consumption prediction model to output the electricity consumption of each user end in the next time period, that is, the user demand; the length of the next time period is indefinite and can be determined according to actual needs, and can be one week, one month, one quarter or one year.

[0072] The electricity consumption prediction model includes a data screening layer, a feature extraction layer, and an electricity consumption prediction layer connected in series; the feature extraction layer uses a GRU gated recurrent unit; the electricity consumption prediction layer uses an SVR support vector regression model.

[0073] The training process of the electricity consumption prediction model includes:

[0074] S3-1. Obtain training user electricity data;

[0075] The training user electricity data includes historical meteorological data, historical user data, and historical time data; the historical meteorological data includes temperature, humidity, wind speed, and rainfall; the historical user data includes user electricity consumption time, user electricity consumption behavior, user working time, and user rest time; the historical time factors include months and holidays;

[0076] S3-2. Input the training user electricity data into the data screening layer, calculate the corresponding Pearson correlation coefficient, sort the Pearson correlation coefficients in descending order, and output the training user electricity data corresponding to the top 15% of the Pearson correlation coefficients;

[0077] S3-3. Input the training electricity consumption data output by S3-2 into the GRU gated recurrent unit for feature extraction, and output the corresponding training time features;

[0078] S3-4. Input the training time features into the SVR support vector regression model, and use the simulated annealing algorithm to adjust the parameters of the SVR support vector regression model to complete the training of the electricity consumption prediction model. The parameters of the SVR support vector regression model include the kernel function, penalty parameter, kernel function coefficient, and value.

[0079] The simulated annealing algorithm is a heuristic global optimization method, and its principle is based on simulating the process of substances in nature transforming from a lower level to a higher level at different temperatures. The main idea of the simulated annealing algorithm is: by setting an initial solution and an objective function, calculating the fitness value of the current solution, and then through a certain strategy, gradually increasing the fitness value of the solution until the optimal solution is reached or the termination condition is satisfied. The simulated annealing algorithm can handle complex large-scale optimization problems, and can automatically adjust the step size of the algorithm during the search process to avoid falling into local optimal solutions. In addition, the simulated annealing algorithm has strong global search ability and can find the global optimal solution.

[0080] The specific process of using the simulated annealing algorithm to adjust the parameters of the SVR support vector regression model is as follows:

[0081] S3-4-1. Set the initial temperature, final temperature, and number of iterations, randomly select the parameters of the SVR support vector regression model as the initial solution and calculate the corresponding optimization objective function; among them, the optimization objective function can adopt the MSE mean square error function, MAE mean absolute error function, RMSE root mean square error function, MAPE mean absolute percentage error function; the current number of iterations is 0;

[0082] S3-4-2. According to the initial temperature, randomly perturb the initial solution to generate a new solution as the current solution;

[0083] S3-4-3. Calculate the optimization objective function of the current solution;

[0084] S3-4-4. Judge whether the optimization objective function of the initial solution is greater than that of the current solution; if so, accept the current solution and enter S3-4-7; otherwise, enter S3-4-5;

[0085] S3-4-5. According to the formula:

[0086] ;

[0087] Obtain a parameter value ; among them, represents the natural constant, represents the initial temperature, , respectively represent the optimization objective function values of the initial solution and the current solution;

[0088] S3-4-6. Judge whether the parameter value is less than ; if so, accept the current solution; otherwise, reject the current solution; enter S3-4-7; among them, represents a constant randomly generated in the interval from 0 to 1;

[0089] S3-4-7. Determine whether a sufficient search is performed at the initial temperature. If so, increment the current iteration count by 1 and proceed to S3-4-8; otherwise, return to S3-4-2.

[0090] S3-4-8. Determine whether the current iteration count has reached the set iteration count. If so, use the current solution as the optimal solution; otherwise, perform temperature decay on the initial temperature to obtain a new temperature, and return to S3-4-2. In the next iteration, the new temperature is used as the initial temperature.

[0091] The data screening layer screens the user power consumption data and selects the highly relevant data, which can reduce the feature extraction time of the GRU (Gated Recurrent Unit), further reduce the size and running time of the power consumption prediction model, and improve the data quality. The GRU can flexibly capture the long-term dependence relationships of the data output by the data screening layer, output key feature data, and reduce the computational complexity. The support vector regression model demonstrates good performance in solving problems such as finite-scale samples and nonlinearity, is faster in prediction than deep learning methods, has a smaller empirical risk. Using the support vector regression model for power consumption prediction can improve the accuracy of power consumption prediction, and has a certain generalization ability and noise resistance, further ensuring the quality of power consumption prediction.

[0092] S4. Update the initial power distribution instruction based on the power consumption of each user terminal in the next time period and the objective function, and obtain the power distribution instruction after one update.

[0093] The step of updating the initial power distribution instruction based on the power consumption in the next time period and the objective function to obtain the power distribution instruction after one update includes:

[0094] Determine the remaining power of the energy storage device of the power generation equipment, and summarize the total power consumption corresponding to the power consumption of each user terminal in the next time period.

[0095] Based on the remaining power, the total power consumption, and the objective function, use the particle swarm optimization algorithm to adjust the power output and reactive power distribution of each unit of the power generation equipment, that is, update the power generation dispatch instruction, and obtain the power distribution instruction after one update.

[0096] S5. Optimize the power distribution instruction after one update based on the objective function to obtain the optimized power distribution instruction.

[0097] The step of optimizing the power distribution instruction after one update based on the objective function to obtain the optimized power distribution instruction includes:

[0098] S5-1. Based on the objective function and the updated power generation dispatch instruction, adjust the tap position of the regulating transformer in the power system, control the switch states of transmission lines, adjust the voltage and frequency of the power grid, the operation mode of distribution transformers, and control the switch states of distribution lines, that is, update the output control instruction and the distribution instruction to obtain the corresponding updated output control instruction and updated distribution instruction;

[0099] S5-2. Based on the updated power generation dispatch instruction, updated output control instruction, and updated distribution instruction, use simulation software (DSIM, ETAP) to simulate the operation process of the power system, monitor the state of the power system during the operation process, and obtain the corresponding power quality indexes and power generation;

[0100] S5-3. Determine whether the power quality indexes meet the regulations of the power industry standards and whether the difference between the power generation and the total power consumption belongs to a preset reference range; if so, use the updated power generation dispatch instruction, updated output control instruction, and updated distribution instruction as the optimized power distribution instruction; otherwise, return to S3.

[0101] S6. Operate the power system based on the optimized power distribution instruction to complete the optimization of power distribution.

[0102] As Figure 2 shown, an optimization system for power distribution based on user needs includes a data acquisition module, an objective function design module, a power consumption prediction module, and an instruction optimization module; the power consumption prediction module includes a data screening sub-module, a feature extraction sub-module, and a power consumption prediction sub-module; where:

[0103] The data acquisition module is used to obtain the initial power distribution instruction of the power system and user power consumption data;

[0104] The objective function design module is used to construct the objective function based on the constraint conditions of the power system;

[0105] The data screening sub-module is used to calculate the Pearson correlation coefficient of the user power consumption data, sort the Pearson correlation coefficients in descending order, and output the user power consumption data corresponding to the top 15% of the Pearson correlation coefficients;

[0106] The feature extraction sub-module is used to input the user power consumption data output by the data screening sub-module into the GRU gated recurrent unit and output the time feature;

[0107] The power consumption prediction sub-module is used to input the time feature into the SVR support vector regression model to obtain the user demand;

[0108] An instruction optimization module, configured to optimize an initial power distribution instruction based on the objective function and the power consumption of each client in the next time period.

[0109] Based on line loss, converter loss, and transformer loss, and considering the over-limit behavior of nodes, the present invention formulates a corresponding objective function, which more comprehensively describes the power loss situation in the power system, assists in optimizing and distributing power instructions to achieve the expected effect, and precisely controls the power distribution to the corresponding clients. By combining the GRU gated unit and the SVR support vector regression model to construct a power consumption prediction model, using the GRU gated unit to extract deep time features to improve data quality, and adopting the SVR support vector regression model for power consumption prediction, the accuracy of power consumption prediction can be improved. Optimizing the power distribution index based on the power consumption prediction value and the objective function can meet the power consumption needs of each client, reduce power loss, and prevent resource waste.

[0110] The above is only the specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or substitutions, which should all be covered by the protection scope of the present invention. Therefore, the protection scope of the present invention should be subject to the protection scope of the claims.

Claims

1. An optimization method for power distribution based on user requirements, characterized in that Including: S1. Obtain the initial electric energy distribution instruction of the power system; the power system includes power generation equipment, the main power grid, the distribution network, and the power consumption end; S2. Based on the constraint conditions of the power system, construct a corresponding objective function; The formula of the objective function is: ； ； ； ； ； Among them, represents the objective function, , respectively represent the minimum value function and the maximum value function, , , respectively represent the transformer loss, the converter loss, and the line loss, represents the penalty function, represents the summation function, , respectively represent the number of units of the power generation equipment and the number of converter operations of the distribution network, represents the converter loss of the th unit of the power generation equipment, represents the loss corresponding to the th converter operation of the distribution network, represents the total number of nodes of the power grid corresponding to the power system, represents the conductance between the th node and the th node of the power grid corresponding to the power system, , respectively represent the voltage values of the th node and the th node of the power grid corresponding to the power system, represents the cosine function, represents the voltage phase difference between the th node and the th node of the power grid corresponding to the power system, , respectively represent the voltage violation coefficient and the power violation coefficient, , respectively represent the maximum allowable voltage and the minimum allowable voltage of the th node of the power grid corresponding to the power system, represents the output power of the th unit of the power generation equipment, , respectively represent the maximum allowable output power and the minimum allowable output power of the th unit of the power generation equipment, , respectively represent the transformer loss of the power generation equipment and the transformer loss of the distribution network, , respectively represent the no-load loss of the transformer of the power generation equipment and the no-load loss of the transformer of the distribution network, , respectively represent the load loss of the transformer of the power generation equipment and the load loss of the transformer of the distribution network; S3. Construct an electric quantity prediction model; obtain user electric quantity data and input it into the electric quantity prediction model, and output the electricity consumption of each user end in the next time period, that is, the user demand; S4. Based on the electricity consumption of each user end in the next time period and the objective function, update the initial electric energy distribution instruction to obtain the electric energy distribution instruction after the first update; S5. Optimize the electric energy distribution instruction after the first update based on the objective function to obtain the optimized electric energy distribution instruction; S6. Operate the power system based on the optimized electric energy distribution instruction to complete the optimization of the electric energy distribution.

2. An optimization method for power distribution based on user needs as described in claim 1, characterized in that, The initial electric energy distribution instruction includes a power generation scheduling instruction, an output control instruction, and a power distribution instruction.

3. An optimization method for power distribution based on user requirements according to claim 2, characterized in that, The constraint conditions include power balance constraint, power flow security constraint, reactive power distribution constraint, and flow constraint of the transmission line.

4. An optimization method for power distribution based on user requirements according to claim 2, characterized in that The electric quantity prediction model includes a series-connected data screening layer, a feature extraction layer, and an electric quantity prediction layer; the feature extraction layer uses a GRU gated recurrent unit; the electric quantity prediction layer uses an SVR support vector regression model.

5. An optimization method for power distribution based on user needs as described in claim 4, characterized in that, The training process of the electric quantity prediction model includes: S3-1. Obtain the training user electric quantity data; The training user electric quantity data includes historical meteorological data, historical user data, and historical time data; the historical meteorological data includes temperature, humidity, wind speed, and rainfall; the historical user data includes user electricity consumption time, user electricity consumption behavior, user working time, and user rest time; the historical time factor includes month and holiday; S3-2. Input the training user electric quantity data into the data screening layer, calculate the corresponding Pearson correlation coefficient, sort the Pearson correlation coefficients in descending order, and output the training user electric quantity data corresponding to the top 15% of the Pearson correlation coefficients; S3-3. Input the training electricity consumption data output by S3-2 into the GRU gated recurrent unit for feature extraction, and output the corresponding training time features; S3-4. Input the training time features into the SVR support vector regression model, and use the simulated annealing algorithm to adjust the parameters of the SVR support vector regression model to complete the training of the electric quantity prediction model.

6. An optimization method for power distribution based on user requirements as described in claim 5, characterized in that, The updating of the initial electric energy distribution instruction based on the electricity consumption in the next time period and the objective function to obtain the electric energy distribution instruction after the first update includes: Determine the remaining electric energy of the energy storage device of the power generation equipment, and summarize the total electricity consumption corresponding to the electricity consumption of each user end in the next time period; Based on the remaining electric energy, the total electricity consumption, and the objective function, use the particle swarm optimization algorithm to adjust the output and reactive power distribution of each unit of the power generation equipment, that is, update the power generation scheduling instruction, to obtain the electric energy distribution instruction after the first update.

7. An optimization method for power distribution based on user requirements as described in claim 6, characterized in that, Optimizing the power distribution instruction after the first update based on the objective function to obtain an optimized power distribution instruction, including: S5-1. Based on the objective function and the updated power generation scheduling instruction, adjust the tap position of the regulating transformer in the power system, control the switch state of the transmission line, adjust the voltage and frequency of the power grid, the operation mode of the distribution transformer, and control the switch state of the distribution line, that is, update the output control instruction and the distribution instruction to obtain the corresponding updated output control instruction and updated distribution instruction; S5-2. Based on the updated power generation scheduling instruction, updated output control instruction, and updated distribution instruction, use simulation software to simulate the operation process of the power system to obtain the corresponding power quality index and power generation amount; S5-3. Judge whether the power quality index meets the regulations of the power industry standard and whether the difference between the power generation amount and the total power consumption belongs to a preset reference range; if so, use the updated power generation scheduling instruction, updated output control instruction, and updated distribution instruction as the optimized power distribution instruction; otherwise, return to S3.

8. An optimization system for power distribution based on user requirements, which is used to implement an optimization method for power distribution based on user requirements according to any one of claims 1 to 7, characterized in that, Including a data acquisition module, an objective function design module, a power consumption prediction module, and an instruction optimization module; the power consumption prediction module includes a data screening sub-module, a feature extraction sub-module, and a power consumption prediction sub-module; where: The data acquisition module is used to obtain the initial power distribution instruction of the power system and the user power consumption data; The objective function design module is used to construct an objective function based on the constraint conditions of the power system; The data screening sub-module is used to calculate the Pearson correlation coefficient of the user power consumption data, sort the Pearson correlation coefficients in descending order, and output the user power consumption data corresponding to the top 15% of the Pearson correlation coefficients; The feature extraction sub-module is used to input the user power consumption data output by the data screening sub-module into the GRU gated recurrent unit and output the time feature; The power consumption prediction sub-module is used to input the time feature into the SVR support vector regression model to obtain the user demand; The instruction optimization module is used to optimize the initial power distribution instruction based on the objective function and the power consumption of each user terminal in the next time period.

Images