Control method and device for flexible load to participate in power grid interaction, electronic equipment and storage medium

Abstract

The embodiment of the invention discloses a control method and device for a flexible load to participate in power grid interaction, electronic equipment and a storage medium. The control method comprises the steps that reservation data of a water heater cluster of a preset area is obtained; acquiring time sequence data of a target water heater from the preset area according to the reservation data; training a preset parameter adjustment model based on the time sequence data to obtain a heating parameter adjustment model; and according to the heating parameter adjusting model and the demand data of the intelligent power grid, a water heater cluster in a preset area is controlled. According to the scheme, the multiple water heaters can be dynamically adjusted in real time according to the demand data of the whole power grid, the power utilization peak can be effectively avoided, and normal operation of the whole power grid is guaranteed.

Description

Technical Field

[0001] The present application relates to the technical field of electric power grids, and in particular to a control method, device, electronic device and storage medium for flexible loads to participate in grid interaction. Background Art

[0002] The frequency of the power system reflects the active power balance relationship between the power generation side and the power consumption side of the power system. Active power imbalance will lead to frequency deviation. Large frequency deviation is not conducive to the safe and effective operation of power equipment and the safe, reliable and economical operation of the entire power system. Therefore, it is necessary to stabilize and control the frequency to control the frequency deviation within the allowable range. The main traditional frequency stabilization and control measures are to achieve power balance between the power generation side and the power consumption side by increasing / decreasing the output of the unit, so as to control the frequency deviation within the allowable range. Although this frequency regulation method can generally meet the frequency regulation requirements of the power system, it has the disadvantages of slow unit response speed, long ramp time, frequent operation and easy wear, high frequency regulation cost and heavy pollution. In addition, with the increase in the proportion of wind, solar and other new energy generators in the power system, their intermittent impact on the power grid makes the frequency regulation of the power system face greater challenges.

[0003] At present, the peak and valley load distribution of power users in 24 time periods is predicted based on the weights of the annual, monthly and weekly power loads of power users in 24 time periods; the 24 peak, valley and valley load prediction time periods of power users are used as the power consumption parameters of the virtual energy storage control technology to control the power load in peak shifting. Although this method can reduce the peak power consumption of electric water heaters of individual users, it cannot obtain the load information of the entire power grid, cannot make overall adjustments, and cannot protect the entire power grid. At the same time, users cannot make independent choices according to actual needs, lacking flexibility and being detrimental to user experience. Summary of the invention

[0004] The embodiments of the present application provide a control system, method, electronic device and storage medium for flexible loads to participate in grid interaction, which can dynamically adjust multiple water heaters in real time according to the demand data of the entire grid, effectively avoid peak power consumption and ensure the normal operation of the entire grid.

[0005] In a first aspect, an embodiment of the present application provides a control method for a flexible load to participate in grid interaction, including:

[0006] Get the reservation data of the water heater cluster in the preset area;

[0007] According to the reservation data, acquiring time series data of the target water heater from the preset area;

[0008] Training a preset parameter adjustment model based on the time series data to obtain a heating parameter adjustment model;

[0009] Based on the heating parameter adjustment model and demand data of the smart grid, a cluster of water heaters in a preset area is controlled.

[0010] In a second aspect, an embodiment of the present application provides a control device for a flexible load to participate in grid interaction, including:

[0011] A first acquisition module is used to acquire reservation data of a water heater cluster in a preset area;

[0012] A second acquisition module, configured to acquire time series data of a target water heater from the preset area according to the reservation data;

[0013] A training model is used to train a preset parameter adjustment model based on the time series data to obtain a heating parameter adjustment model;

[0014] The control model is used to control the water heater cluster in a preset area according to the heating parameter adjustment model and the demand data of the smart grid.

[0015] Correspondingly, the present application also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the steps of any of the above methods when executing the program.

[0016] The present application also provides a storage medium, wherein the storage medium stores a computer program, and the computer program implements any of the above methods when the computer program is executed by a processor.

[0017] The present application provides a control method, device, electronic device and storage medium for flexible loads to participate in grid interaction. After obtaining the reservation data of a water heater cluster in a preset area, the time series data of the target water heater is obtained from the preset area according to the reservation data. Then, the preset parameter adjustment model is trained based on the time series data to obtain a heating parameter adjustment model. Finally, the water heater cluster in the preset area is controlled according to the heating parameter adjustment model and the demand data of the smart grid. In the control scheme for flexible loads to participate in grid interaction provided in the present application, the preset parameter adjustment model can be trained based on the time series data to obtain a heating parameter adjustment model. Finally, the water heater cluster in the preset area can be controlled according to the heating parameter adjustment model and the demand data of the smart grid. That is, multiple water heaters can be adjusted dynamically in real time according to the demand data of the entire power grid, which can effectively avoid peak power consumption and ensure the normal operation of the entire power grid. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For those skilled in the art, other drawings can be obtained based on these drawings without creative work.

[0019] Figure 1 It is a flow chart of a control method for a flexible load to participate in grid interaction provided in an embodiment of the present application;

[0020] Figure 2 is another flow chart of the control method for flexible loads to participate in grid interaction provided by an embodiment of the present application;

[0021] Figure 3 It is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0022] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present application. Instead, they are merely examples of systems and methods consistent with the examples detailed in the attached claims, or with some aspects of the present application.

[0023] It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0024] In the subsequent description, the suffixes such as "module", "component" or "unit" used to represent elements are only used to facilitate the description of the present application, and have no specific meanings. Therefore, "module", "component" or "unit" can be used in a mixed manner.

[0025] The embodiments of the present application provide a control method, device, electronic device and storage medium for flexible loads to participate in grid interaction.

[0026] It should be noted that the order of description of the following embodiments is not intended to limit the priority order of the embodiments.

[0027] The present application provides a control method for flexible loads participating in grid interaction, including: obtaining reservation data of a water heater cluster in a preset area; obtaining time series data of a target water heater from the preset area based on the reservation data; training a preset parameter adjustment model based on the time series data to obtain a heating parameter adjustment model; and controlling the water heater cluster in the preset area based on the heating parameter adjustment model and demand data of the smart grid.

[0028] See also Figure 1 , Figure 1 : is a flow chart of a control method for a flexible load to participate in grid interaction provided by the present application. The control method for a flexible load to participate in grid interaction specifically includes the following process:

[0029] 101. Obtain reservation data for a water heater cluster in a preset area.

[0030] First, a user account system can be established to allow users to register and create a personal profile, including information related to the water heater, to ensure that the water heater device can be integrated with the user account system and the reservation platform, allowing remote control and monitoring, and users can select the time period they wish to use and the desired water temperature through the reservation interface. The user account system can record the user's reservation information, including the reservation time, duration, target water temperature, etc.

[0031] 102. According to the reservation data, time series data of the target water heater is obtained from the preset area.

[0032] Time series data refers to a collection of data points arranged in chronological order, and is usually used to analyze and predict data change patterns within consecutive time intervals. In the reservation and control scenario of a water heater cluster, time series data can include various time-related measurements, such as water temperature, energy consumption, ambient temperature, etc.

[0033] For example, based on the user's reservation information, data related to a specific water heater is filtered out from the database. This may include the user's reservation time, the model and location of the water heater, etc. Specifically, data can be collected regularly from the water heater's sensors or smart meters, such as recording water temperature and energy consumption every minute or every hour. Among them, each data point corresponds to a precise timestamp, indicating the time when the data was collected. The time series data type can be temperature, energy consumption, pressure or flow, etc. It should be noted that the frequency of data collection can be determined according to the analysis requirements. For example, for real-time control, high-frequency data (such as once a minute) is required.

[0034] 103. Train the preset parameter adjustment model based on the time series data to obtain a heating parameter adjustment model.

[0035] The parameter adjustment model refers to a model used to optimize model parameters to improve model performance in machine learning and statistical modeling. In the scenario of water heater cluster control, the parameter adjustment model focuses on optimizing the parameters of the heating process, such as heating time, temperature setting, energy consumption, etc., to achieve more efficient energy use and meet user needs.

[0036] For example, specifically, useful time series features are extracted from the original time series data, and the time series features can be generated using a sliding window technology. Then, a preset parameter adjustment model is trained based on the time series features to obtain a heating parameter adjustment model. That is, optionally, in some embodiments of the present application, the step of "training a preset parameter adjustment model based on the time series data to obtain a heating parameter adjustment model" can specifically include:

[0037] Obtain temperature information, heating power, and user demand information corresponding to the water heater cluster from time series data;

[0038] Constructing a first temperature change rate formula corresponding to the water heater cluster according to the temperature information, heating power and user demand information;

[0039] The preset parameter adjustment model is trained based on the first temperature change rate formula and the time series data to obtain a heating parameter adjustment model.

[0040] Constructing the first temperature change rate formula corresponding to the water heater cluster is a physical and mathematical modeling process that aims to describe the rate at which the water temperature in the water heater changes over time. Specifically,

[0041] The heat loss coefficient k(T) is expressed as a polynomial function that depends only on temperature, which captures the effect of temperature on heat loss without the need for a reference temperature:

[0042] k(T)=k 0 ·(1+α 1 ·T(t) β +α 2 ·T(t) γ )

[0043] Where: k 0 is the initial heat loss coefficient, α 1 and α 2 is the coefficient that controls the effect of temperature on heat loss, and β and γ are nonlinear exponents of temperature.

[0044] At heating power P heat 、Ambient temperature T ambient Under the premise that the heat dissipation area A is constant, the first temperature change rate formula is:

[0045]

[0046] Where: k(T) is the heat loss coefficient, as a nonlinear function of temperature, A is the heat dissipation area of ​​the water tank (constant), m water is the mass of water, C water is the specific heat capacity of water, P heat is the heating power (constant), T ambientis the ambient temperature (constant).

[0047] The parameters in the above formula can be fitted by existing experimental data, such as k 0 , α 1 , α 2 , β, γ, etc. By fitting these parameters, the temperature change rate (i.e., the second temperature change rate) can be predicted at an unknown temperature.

[0048] For example, the following parameters were obtained through experimental fitting: k 0 =0.002, α 1 =0.003, α 2 =0.001, β=1.5, γ=2.5, A=1.5m 2 、m water =150kg, C water =4184J / (kg·℃),P heat =4000W, T ambient =25℃

[0049] At 55°C, the first temperature change rate can be calculated:

[0050]

[0051] Calculated

[0052] It can be seen that there are five parameters in the first temperature change rate formula. Given some temperature change curves, we need to determine these five parameters (k 0 , α 1 , α 2 , β, γ).

[0053] The ultimate goal is to determine these five parameters (k 0 , α 1 , α 2 , β, γ) make the formula have a good fitting effect on the known curve.

[0054] In order to determine these five parameters (k 0 , α 1 , α 2 , β, γ), a global optimization algorithm such as a genetic algorithm or simulated annealing may be used. Optionally, in some embodiments of the present application, a genetic algorithm may be used to fit these parameters.

[0055] First, collect a data set of temperature changes over time. This data should include the temperature changes at different initial temperatures and record the temperature at each moment. The data set is Data = {T(t), t, T ambient ,Pheat ,A}

[0056] Among them, temperature data: T(t), that is, the temperature at time t. Time data: the time t corresponding to each temperature data. Ambient temperature: T ambient , assumed to be a constant. Heating power: P heat , assumed to be a constant. The cooling area of ​​the water tank: A, assumed to be a constant. Then, define the objective function, which is used to measure the error between the model prediction value and the actual value. The goal is to minimize these errors. The second temperature change rate formula fitted is:

[0057]

[0058] Finally, the preset parameter adjustment model is trained based on the second temperature change rate formula and the time series data to obtain the heating parameter adjustment model. That is, optionally, in some embodiments of the present application, the step of “training the preset parameter adjustment model based on the first temperature change rate formula and the time series data to obtain the heating parameter adjustment model” may specifically include:

[0059] Get the preset optimization algorithm;

[0060] Constructing a second temperature change rate formula according to the optimization algorithm and the first temperature change rate formula;

[0061] The preset parameter adjustment model is trained based on the second temperature change rate formula and time series data to obtain a heating parameter adjustment model.

[0062] First, it can be defined as the mean square error between the first temperature change rate and the second temperature change rate

[0063]

[0064] in, is the first temperature change rate, is the second temperature change rate.

[0065] Genetic algorithm is a global optimization algorithm based on natural selection and genetic mechanism. The following are the steps to fit parameters using genetic algorithm:

[0066] a. Initialize the population

[0067] Coding scheme: define each individual as parameter k 0 , α 1 , α 2 , β, γ, can be represented by a vector, expressed as [k 0 ,α 1 ,α 2 ,β,γ].

[0068] Initial population: A set of initial solutions is randomly generated, each solution represents a combination of parameter values. The population size can be set to 100 to 200 individuals.

[0069] b. Fitness function

[0070] Calculate the fitness of each individual, that is, the corresponding objective function value. The goal is to minimize the error, and set the fitness to the inverse of the objective function value.

[0071]

[0072] Here ò is a small positive number (e.g. 10 -6 ), which is used to prevent division by zero errors when the MSE is very small.

[0073] When the objective function value is very small, the fitness value is large, which means that the model prediction is very close to the actual data and the fitness is high. The genetic algorithm will be more inclined to retain and use such individuals.

[0074] When the objective function value is large, the fitness value is small, indicating that the model prediction is poor and the fitness is low, and the genetic algorithm tends to eliminate such individuals.

[0075] c.Select

[0076] Using the tournament selection method, individuals with higher fitness are selected from the current population to enter the next generation.

[0077] Tournament selection is a commonly used selection strategy that randomly selects multiple individuals from the population for comparison and selects the individual with the highest fitness as the parent. The main idea of ​​tournament selection is to select the winner from the local competition instead of relying on the global probability distribution. This selection strategy has a high selection pressure in the genetic algorithm, which can help maintain the diversity of the population while effectively avoiding falling into the local optimum.

[0078] Here are the detailed steps for tournament selection:

[0079] Set the tournament size: Choose the size k of the tournament (usually 2 to 5). The larger the tournament size, the greater the selection pressure, which means that each tournament has a higher probability of selecting individuals with higher fitness.

[0080] Random selection of participants: k individuals are randomly selected from the population (without replacement). These individuals will compete in a "tournament".

[0081] Compare fitness: Calculate the fitness values ​​of these k individuals and select the individual with the highest fitness as the winner. This individual will be selected as the parent individual until a sufficient number of parent individuals are selected for the next crossover operation.

[0082] By adjusting the tournament size k, the selection pressure can be flexibly controlled. If k is large, the algorithm tends to select individuals with higher fitness, thereby accelerating convergence; if k is small, the selection is more random, increasing population diversity.

[0083] Simple implementation: The implementation of tournament selection is very intuitive and does not require complex probability calculations or sorting.

[0084] Maintaining diversity: Tournament selection is better at maintaining population diversity even when the fitness between individuals varies greatly, because each tournament involves only a few individuals rather than the entire population.

[0085] d. Cross

[0086] Here, a single-point crossover operation is used to exchange some genes (parameter values) of two parent individuals to generate new individuals (offspring). The crossover probability is set to 80% to determine whether to perform a crossover. Individuals that do not perform a crossover are directly copied to the next generation.

[0087] Single-point crossover is a basic and commonly used genetic algorithm crossover method. It generates two new offspring individuals by selecting a crossover point (i.e., cutting point) between two parent individuals and then exchanging the gene fragments after the crossover point. This method is simple and easy to implement, and can effectively combine the genetic information of the two parents to generate new individuals with diversity.

[0088] The following are the detailed steps for single point crossover:

[0089] Two parent individuals are randomly selected from the population, usually determined by a selection operation such as tournament selection. A crossover point (position) p is randomly selected. The crossover point is a position in the gene sequence of the parent, usually in the middle of the gene sequence, but can be located anywhere in the sequence.

[0090] The part of the parent individual before the crossover point p is retained as part of the offspring, and the part after the crossover point p is exchanged into another parent, thereby generating two new offspring individuals.

[0091] Examples of offspring individuals:

[0092] For example, suppose the gene sequences of the two parents are

[0093] Parent 1: [k0 1 ,α1 1 ,α2 1 ,β 1 ,γ 1 ]

[0094] Parent 2: [k0 2 ,α1 2,α2 2 ,β 2 ,γ 2 ]

[0095] Assuming that the selected crossover point p=3 (i.e., crossover is performed after the third gene), the two offspring individuals generated are:

[0096] Child 1: [k0 1 ,α1 1 ,α2 1 ,β 2 ,γ 2 ]

[0097] Child 2: [k0 2 ,α1 2 ,α2 2 ,β 1 ,γ 1 ]

[0098] 5. Add offspring to the new population:

[0099] The two generated offspring individuals are added to the new population. By repeating this process, a new generation of population can be generated.

[0100] e. Variation

[0101] The individuals in the offspring are mutated, and certain genes (parameter values) are randomly changed with a certain probability, thereby increasing the diversity of the population. The mutation rate is usually set between 0.01 and 0.1.

[0102] Mutation method: The mutation operation is to randomly change some genes (parameter values) of individuals to increase the diversity of the population and prevent premature convergence.

[0103] Mutation probability: usually set between 0.01 and 0.1.

[0104] Gaussian mutation is selected as the mutation method. This mutation method perturbs the gene value by adding a small random Gaussian noise to the gene value.

[0105] Implementation steps of Gaussian mutation: Set the mutation probability p mut : Set the probability of mutation to 3%. This probability controls the frequency of mutation, that is, the probability of each gene being mutated in each generation.

[0106] Check whether each gene (parameter) in the individual has mutated. If the random number is less than the mutation probability p mut , the gene will mutate.

[0107] For the gene selected for mutation, a random number following Gaussian distribution N(0,σ) is added as noise, where σ is the standard deviation, which determines the amplitude of the mutation. The size of the standard deviation is usually adjusted according to the specific problem.

[0108] Formula: new_gene=old_gene+N(0,σ)

[0109] Make sure the gene value is within a reasonable range:

[0110] The mutated gene value may need to be limited to a reasonable range (for example, the parameter cannot be negative). After the mutation, the gene value needs to be checked and adjusted to keep it within the preset range.

[0111] f. Update population

[0112] Replace some individuals in the old population with new offspring to form a new generation. Repeat the above steps until the maximum number of generations is reached or the objective function converges.

[0113] A method combining the elite retention strategy with partial replacement is selected. This ensures that the best individuals are retained while introducing new offspring to maintain the diversity of the population.

[0114] Partial replacement: Only some of the offspring are used to replace individuals in the current population, usually the worst individuals.

[0115] Elite retention strategy: retain the best individuals in the current population to the next generation to ensure that the optimal solution is not lost.

[0116] 2. Implementation steps for updating population

[0117] a. Elite Retention

[0118] Determine the number of elite individuals:

[0119] Set the number of elite individuals N elite , which is set to 5% of the total population. These individuals are the ones with the highest fitness in the current population.

[0120] Selecting elite individuals:

[0121] According to the fitness value, sort the individuals in the current population and select the top N elite The individuals with the highest fitness are directly retained to the next generation.

[0122] b. Generate offspring

[0123] Perform selection, crossover, and mutation:

[0124] Use the selection, crossover and mutation operations described above to generate new offspring individuals. The number of offspring is N.offspring =N population -N elite

[0125] Check the diversity of the offspring:

[0126] Ensure that there is enough diversity in the generated offspring to prevent too many repeated individuals. If the diversity is insufficient, you can increase the mutation rate or adjust the position of the crossover point.

[0127] c. Update population

[0128] Merge elite individuals and offspring:

[0129] Merge elite individuals with offspring to form a new generation of population. The total number of the new population should be equal to the original population. Recalculate the fitness value of each individual in the new generation of population for use in the next round of selection operations. Check the termination conditions (such as reaching the maximum number of generations or fitness no longer significantly improving). If the conditions are not met, enter the next generation cycle.

[0130] g. Termination conditions

[0131] The maximum number of iterations is set to 1000 generations, and the threshold is set to 0.001. If the increase in the fitness value of the best individual is less than 0.001 in 10 consecutive generations, it is considered that the fitness has not been significantly improved.

[0132] Judgment criteria: When the fitness value no longer improves significantly or reaches the maximum number of iterations, the iteration is terminated.

[0133] After multiple generations of iterations, the genetic algorithm will converge to a parameter set close to the global optimal solution. 0 , α 1 , α 2 , β, and γ will be able to fit your experimental data well and be used to predict the changes in water tank temperature under different conditions.

[0134] Use sliding window technology to generate time series features. Here, the window size is selected as 24 hours and the sliding step is 1 hour.

[0135] Such as trends, seasonality, cyclicality, etc.

[0136] Create lagged features, which are data from past points in time that may be useful in predicting future values.

[0137] Select the model type:

[0138] Choose an appropriate model based on the nature of the problem, such as statistical models (ARIMA, SARIMA), machine learning models (Random Forest, Support Vector Machine, Neural Network), or deep learning models (LSTM, GRU).

[0139] Divide the dataset into training set, validation set, and test set. Use the training set data to train the model. Adjust the model parameters (hyperparameter tuning) to improve the performance of the model. Perform cross-validation to evaluate the generalization ability of the model. Use the validation set to evaluate the performance of the model. Common evaluation indicators include mean square error (MSE), root mean square error (RMSE), mean absolute error (MAE), etc.

[0140] Optionally, in some embodiments of the present application, the step of “training a preset parameter adjustment model based on the second temperature change rate formula and the time series data to obtain a heating parameter adjustment model” may specifically include:

[0141] Extract time series features from time series data;

[0142] The preset parameter adjustment model is trained according to the time series characteristics and the second temperature change rate formula to obtain the heating parameter adjustment model.

[0143] Time series features are variables used to describe the characteristics of time series data, which can be used for modeling and prediction. Based on the first temperature change rate formula, more complex factors need to be considered, such as the specific heating mode of the water heater, user behavior patterns, etc., to construct the second temperature change rate formula. This formula can contain additional terms or parameters to more accurately describe the temperature change.

[0144] Optionally, in some embodiments of the present application, the step of “constructing a second temperature change rate formula according to the optimization algorithm and the first temperature change rate formula” may specifically include:

[0145] Get the preset mean square error function;

[0146] A second temperature change rate formula is constructed according to the mean square error function, the optimization algorithm and the first temperature change rate formula.

[0147] 104. Based on the heating parameter adjustment model and the demand data of the smart grid, the water heater cluster in the preset area is controlled.

[0148] Obtain real-time demand data for smart grids, including grid load, electricity price information, demand response signals, etc. Collect real-time data from water heater clusters, including current water temperature, heating status, user reservation time, etc.

[0149] Load the trained heating parameter adjustment model, which can predict the optimal heating strategy based on the input data. Update the model parameters according to the latest smart grid demand data and environmental changes (such as temperature changes, user behavior changes, etc.). According to the demand response signal of the smart grid, formulate corresponding control strategies, such as reducing heating during peak electricity price periods, or increasing heating during low electricity price periods. Ensure that the control strategy is implemented under the premise of meeting the user's reservation time and water temperature requirements. Generate specific control instructions for each water heater based on the heating parameter adjustment model and control strategy. Reduce energy waste and improve energy efficiency by optimizing the heating plan.

[0150] Optionally, in some embodiments of the present application, the step of “controlling a water heater cluster in a preset area according to the heating parameter adjustment model and demand data of the smart grid” may specifically include:

[0151] Obtaining load information and demand response information of the smart grid from demand data of the smart grid;

[0152] Determine the heating plan corresponding to the water heater cluster in the preset area based on the heating parameter adjustment model, load information and demand response information;

[0153] Control clusters of water heaters in pre-set areas based on heating schedules.

[0154] For example, real-time load data of the smart grid is collected, including total power consumption, load forecasts for different time periods, etc. Then, demand response signals from the grid are obtained, which may involve requests from grid operators to reduce or increase power consumption, especially during peak hours or in emergency situations. The acquired grid data is integrated with historical and real-time data of the water heater cluster (such as current water temperature, set temperature, user reservation time, etc.). Next, the integrated data is used as input to apply the heating parameter adjustment model. The model should be able to adjust the heating strategy of the water heater according to the grid load and demand response information. The optimization objectives of the model include minimizing energy consumption, balancing grid load, and meeting user needs. Based on the output of the model, a heating plan for the water heater cluster is generated. It includes the heating time, heating power, and target temperature of each water heater. Ensure that the heating plan meets the demand response requirements of the grid and the usage needs of users, while considering the operating limitations of the water heater. The generated heating plan is converted into control instructions and sent to each water heater through an appropriate communication protocol. Monitor the operating status of the water heaters to ensure that they are executed according to the plan. This may involve a feedback control system to adjust the heating strategy in real time.

[0155] Optionally, in some embodiments of the present application, the step of “determining a heating plan corresponding to a water heater cluster in a preset area according to a heating parameter adjustment model, load information, and demand response information” may specifically include:

[0156] Fusion of load information and demand response information;

[0157] The fused information is input into the heating parameter adjustment model to determine the heating plan corresponding to the water heater cluster in the preset area.

[0158] Obtain real-time and predicted load data from the smart grid, including total power consumption, peak and valley periods, etc. At the same time, obtain the demand response information of the power grid, which indicates to reduce or increase power consumption during specific periods. Next, fuse the load information and demand response information with the historical and real-time data of the water heater cluster (such as current water temperature, set temperature, user reservation time, etc.). This can be achieved through data integration technology to ensure that all relevant information is integrated into a unified data framework. Provide the preprocessed data as input to the heating parameter adjustment model. The model should be able to adjust the heating strategy of the water heater based on the load and demand response information of the power grid. Finally, adjust the model parameters based on the latest power grid data and user needs to ensure that the heating plan output by the model meets both user needs and responds to power grid needs. Generate a heating plan for the water heater cluster using the heating parameter adjustment model. This should include the heating time, heating power and target temperature of each water heater.

[0159] In order to further understand the control scheme of the flexible load participating in the grid interaction of this application,

[0160] Using deep learning methods to plan the heating plan of a household water heater cluster, the total power of the household water heater cluster is reduced

[0161] Constraints need to be set here, that is, for the i-th water tank, assuming that the adjustment time starts at 5 pm and the i-th user expects to take a shower at 8 pm, then the load adjustment constraint condition requires that the temperature change rate of the i-th water tank integrated from 5 pm to 8 pm must reach the set temperature value minus the initial temperature value.

[0162] That is, there is a formula:

[0163] 1. Data preparation

[0164] Data Collection:

[0165] The main goal of data collection is to obtain time series data, including water heater temperature, ambient temperature, heating power, and user demand. The data record format is as follows:

[0166] Timestamp i

[0167] Current water temperature T(t i )

[0168] Ambient temperature T ambient (t i )

[0169] User sets the target temperature T target

[0170] Current heating power P(t i )

[0171] Water heater status S(t i )(Heating / keeping / off, represented by 1 / 0 / -1 respectively)

[0172] 2. Data preprocessing:

[0173] Normalize the temperature, power and time data. Assume that the temperature data is in the range T min =20℃ T max =80℃, then the normalized formula is:

[0174]

[0175] Similarly, the power data is normalized:

[0176]

[0177] Here P max Take P max =3000W

[0178] Interpolation method: For missing data points, linear interpolation is used:

[0179]

[0180] 3. Feature Engineering:

[0181] Use sliding window technology to generate time series features. Here, the window size is selected as 24 hours and the sliding step is 1 hour.

[0182] For time series data {T(t i )}, the data set generated by the sliding window is:

[0183] X i =[T(t i ),T(t i-1 ),…,T(t i-n+1 )]

[0184] Here, n is the window length, which represents the number of data points within 24 hours.

[0185] 2. Model selection

[0186] Model Architecture:

[0187] Select LSTM (Long Short-Term Memory Network), the model architecture is as follows:

[0188] Input layer: The input data is the feature sequence X generated by the sliding window i

[0189] Hidden layer: Single-layer LSTM network with 128 hidden units.

[0190] Output layer: The fully connected layer outputs the heating plan for the next hour, with one data point every 5 minutes. Assume that the output is y i , then:

[0191] y i =LSTM(X i )

[0192] LSTM unit formula: The core of LSTM is its three gating mechanisms, input gate, forget gate and output gate. The specific formula is:

[0193] Forget Gate:

[0194] f t =σ(W f ·[h t-1 ,x t ]+b f )

[0195] The forget gate determines the state C of the memory cell at the previous time step t-1 How much information can be retained to the memory cell state C at the current time step t This gate can choose which information to "forget".

[0196] Input: h t-1 is the hidden state of the previous time step, x t is the input for the current time step.

[0197] Weight matrix: W f is a weight matrix used to map the hidden state and current input to a new space.

[0198] Bias: b f is the bias vector.

[0199] Activation function: σ is the Sigmoid activation function, and the output value is between 0 and 1, which is used to control how much information is forgotten (0 means completely forgotten, 1 means completely retained).

[0200] Input Gate:

[0201] i t =σ(W i ·[h t-1 ,x t ]+b i )

[0202]

[0203] The input gate determines how new information at the current time step is written into the memory cell.

[0204] Input gate control: i t Controls what new information will be added to the memory cell.

[0205] Input: Same as the forget gate, h t-1 and x t as input.

[0206] Output: Through the Sigmoid activation function, the selection range of information is compressed to between 0 and 1.

[0207] Candidate memory cell status: It is a candidate memory value generated by the current input and the previous hidden state, representing the new information to be added to the memory cell.

[0208] Activation function: tanh is a hyperbolic tangent activation function with an output between -1 and 1, which is used to generate new memory information.

[0209] Output Gate:

[0210] o t =σ(W o ·[h t-1 ,x t ]+b o )

[0211] h t =o t *tanh(C t )

[0212] The output gate determines the hidden state h at the current time step t What to output is the final output of the LSTM unit.

[0213] Output gate control: o t Controls how much information is output from the memory cell.

[0214] Input: Also use h t-1 and x t as input.

[0215] Output: The output ratio is controlled through the Sigmoid activation function.

[0216] Hidden state: h t It is the transformation result of the memory cell state controlled by the output gate. The final hidden state is the memory cell state C t A gated output.

[0217] Activation function: tanh transforms the memory cell state again and then multiplies it by the output o of the output gate t , which determines the final output content.

[0218] Memory unit status update:

[0219]

[0220] Memory cell state C t It is the core part of LSTM, which records long time series information and decides when to update or delete it through a gating mechanism.

[0221] Update process:

[0222] The forget gate f controls the memory C of the previous time step t-1 How much should be retained.

[0223] Input Gate i t and candidate memory cell states Controls how new information should be added to the current memory cell state C t middle.

[0224] Final memory: After combining the two, we get the memory cell state C at the current time step t .

[0225] 3. Model training

[0226] Objective function:

[0227] Use the mean square error (MSE) loss function as the objective function:

[0228]

[0229] Among them, y i is the actual power, is the predicted power.

[0230] Training process:

[0231] Select the Adam optimization algorithm,

[0232] The Adam optimization algorithm (Adaptive Moment Estimation) is an optimization algorithm for training deep learning models that combines the advantages of momentum and RMSprop. Adam dynamically adjusts the learning rate of each parameter by calculating the first-order moment estimate (momentum) and the second-order moment estimate (squared average of the gradient) of the gradient. The formula of the Adam optimization algorithm and its meaning are explained in detail below.

[0233] Its update rule is:

[0234] 1. Gradient calculation

[0235] In each round of iteration, we first calculate the gradient g of the loss function L(θ) with respect to the model parameter θ. t :

[0236]

[0237] g t is the gradient of the loss function with respect to the parameter θ, indicating that at the current parameter value θ t The direction and rate of change of the loss function.

[0238] It means to find the gradient of the parameter θ, with the goal of finding the direction that minimizes the loss function L(θ).

[0239] 2. First-order moment estimation (momentum)

[0240] Calculate the first-order moment estimate of the gradient, i.e., momentum, using m t express:

[0241] m t =β 1 m t-1 +(1-β 1 ) t

[0242] m t It is the exponentially weighted average of the gradient (i.e. momentum), which is used to smooth the gradient during the optimization process to avoid oscillations caused by large gradient changes.

[0243] β 1 is the decay rate of momentum (usually close to 1, here it is 0.9), which determines the influence of past gradients on the current gradient estimate.

[0244] mt -1 is the first moment estimate of the previous moment.

[0245] (1-β 1 ) is a correction factor to ensure that the newly calculated gradient value g t There is enough influence in momentum estimation.

[0246] 3. Second-order moment estimation (squared average of gradient)

[0247] Calculate the second-order moment estimate of the gradient, which is the exponentially weighted average of the square of the gradient, using v t express:

[0248]

[0249] vt It is the exponentially weighted average of the square of the gradient, which is used to estimate the variance of the gradient and ensure that the learning rate dynamically adapts to changes in the gradient.

[0250] β 2 It is the decay rate of the second-order moment estimate (usually close to 1, often 0.999), which determines the influence of the past squared gradient on the current estimate.

[0251] It is the square of the current gradient, indicating the "amplitude" of the current gradient.

[0252] 4. Deviation correction

[0253] Since the momentum and second-order moment estimates will have deviations (close to 0) in the early stage, deviation correction is required:

[0254]

[0255] and is the bias-corrected average of momentum and squared gradient. t and v t The values ​​may be biased towards 0 in the early stages, but this bias decays exponentially over time, so they are corrected in each round to get unbiased estimates.

[0256] 5. Parameter Update

[0257] Update the parameters using the bias-corrected momentum and second-order moment estimates:

[0258]

[0259] Among them, g t is the gradient, η is the learning rate, β 1 ,β 2 is the momentum parameter.

[0260] θ t is the updated parameter value.

[0261] η is the learning rate, which controls the size of the step of parameter update. It is the core part of Adam, which represents the update amount of each parameter in the gradient direction. is the square root of the second moment estimate plus a small constant ò = 10 -8 , used to prevent the denominator from being zero.

[0262] 6. Parameter Update

[0263] Set the number of iterations to 500 and terminate the iteration after reaching the maximum number of iterations. At this time, the final parameters are determined.

[0264] When using the Adam optimization algorithm to train the model, through the above steps, the Adam algorithm can update the parameters of the model in each iteration, so that the loss function (such as mean square error MSE) gradually decreases. During the training process, the Adam algorithm dynamically adjusts the learning rate of each parameter according to the gradient information, and flexibly updates the parameters according to the gradient changes.

[0265] The final Adam optimization algorithm output: optimized model parameters θ * (including all weights and biases in the LSTM network), so the parameters θ of the optimized LSTM model are obtained * (including all weights and biases in the LSTM network) so that the model can best fit the training data. The optimized parameters θ * It is the direct output of the Adam algorithm. After obtaining the optimized parameters θ of the LSTM model * , thus providing a basis for generating prediction output.

[0266] The parameters θ generated by the Adam algorithm t The values ​​are assigned to the LSTM model to generate a trained and optimized LSTM model. The Adam optimization algorithm indirectly helps to obtain the final prediction output. It optimizes the internal parameters of the model so that the LSTM model can generate the required heating plan prediction results based on the new input data.

[0267] Use the trained and optimized model to use these optimized parameters to make predictions or generate results. The specific process is as follows:

[0268] After the training is completed, an optimized LSTM model is obtained. At this point, new input data (such as future ambient temperature, historical temperature and other features) can be input into the trained model.

[0269] The input data is in the form of X new , which are the new features required for model prediction.

[0270] The model uses the optimized parameters θ * (output by the Adam optimization algorithm) performs forward propagation calculation. Specifically, the model uses LSTM units to gradually process the input time series data and obtains the output through forward propagation calculation.

[0271] The LSTM layer passes the time series features to the hidden layer, which is then mapped to the output space through the fully connected layer.

[0272] In the model calculation, set additional constraints, that is, for the i-th water tank, assuming that the regulation time starts at 5 pm, and the i-th user expects to take a shower at 8 pm, then the load regulation constraint condition is that the temperature change rate of the i-th water tank is integrated from 5 pm to 8 pm to obtain the temperature change value that reaches the set temperature value minus the initial temperature value. That is, there is a formula:

[0273]

[0274] Temperature change rate It can reflect both heating and insulation. Heating is positive, and insulation is negative. From 5 to 8 o'clock, the i-th water tank may be heating, insulation, heating, insulation, etc.

[0275] If the formula is not satisfied Then regenerate the plan until it is satisfied

[0276] Output prediction results:

[0277] Finally, the model calculates and outputs a prediction based on the input data This result is the heating schedule you need. For example, the model output is This could be a forecast of heating power every 5 minutes for the next hour:

[0278] These predictions indicate what the heating power of the water heater should be in the next 12 time steps (i.e. one hour).

[0279] In this deep learning model, the input and output are defined as:

[0280] Time series features: Past temperature data (current and previous water temperatures): used to capture the temperature change trend of the water heater.

[0281] Form: [T(t i-n+1 ),T(t i-n+2 ),…,T(t i )]

[0282] Past ambient temperature data (current and previous ambient temperature): Ambient temperature affects the heat loss of the water heater, so it is helpful for the model. Form: [T ambient (t i-n+1 ),T ambient (t i-n+2 ),…,T ambient (t i )]

[0283] Past heating power data (current and previous heating power): used to understand historical power consumption to help predict future power demand. Form: [P(t i-n+1 ),P(t i-n+2 ),…,P(t i )]

[0284] Past water heater status data (heating / insulation / off): Understanding the working status of the water heater is helpful for temperature change prediction. Format: [S(t i-n+1 ),S(t i-n+2 ),…,S(t i )]

[0285] Time features: such as hours, days of the week, etc. This information helps the model understand the user's usage habits and time patterns. Form:

[0286] User-set target temperature:

[0287] Target water temperature: The water temperature that the user expects to reach within a specified time period. The model needs to ensure that the water heater reaches this temperature before that time point. Form: T target

[0288] Sliding window data:

[0289] The above data is usually input into the model in the form of a sliding window. For example, the window size is 24 hours and the sliding step is 1 hour. The data in the window will be input into the LSTM model as a whole for processing. Form: [X i-n+1 ,X i-n+2 ,…,X i ], where X i Represents at time t i The data vector at .

[0290] Output

[0291] Heating plan for the future time period:

[0292] Heating power prediction: The main output of the model is the heating power plan for the water heater in the future. This includes predicting the heating power at each moment to ensure that the target temperature is reached at the time specified by the user while optimizing the total power consumption. i+1 ),P(t i+2 ),…,P(t i+m )], where m is the number of time steps into the future for prediction (e.g. one data point every 5 minutes for 1 hour).

[0293] Water heater status prediction:

[0294] State control instructions: Based on power prediction, the model can also output the working state instructions of the water heater, that is, whether it is heating, keeping warm or shutting down at each moment. Form: [S(t i+1 ),S(t i+2 ),…,S(t i+m )], the status value can be heating (1), keeping warm (0) or off (-1).

[0295] Assuming that the input data is data from the past 24 hours, divided into 48 time steps (one data point every 30 minutes), then: the input feature matrix X size is [48,5], where 5 represents the five input variables of temperature, ambient temperature, power, state, and time characteristics.

[0296] The model outputs a heating plan for the next hour, which is divided into 12 time steps (one data point every 5 minutes), that is, the output size is [12, 1] The heating power sequence.

[0297] For example, the peak load time of the power grid is from 19:00 to 21:00, especially at 20:00. For users who require water heaters to heat up between 19:00 and 21:00, their household water heaters are adjusted to achieve the purpose of grid demand response.

[0298] For users, a reward mechanism is set up to reward users according to the time they set. After signing an agreement to participate in grid demand response, users can get rewards every time their household water heater responds to grid demand regulation.

[0299] Single reward value Reward i for

[0300] Reward i =A×|required time-20 points|+B (5)

[0301] The total reward value of n adjustments _total for:

[0302]

[0303] The required time here means that the user requires the water heater to reach the set temperature at that time

[0304] |Requested time - 20 o'clock| refers to the difference between the user's requested time and 20 o'clock. The larger the difference is, the farther the user's requested time is from 20 o'clock, and the smaller the impact of the corresponding water heater's heating power on the power balance constraint of the power grid.

[0305] 1. Single Reward Value Formula

[0306] A: This coefficient indicates the degree of influence of time deviation. The more time deviation, the higher the reward value will be.

[0307] B: This is a fixed base reward. The user will receive this part of the reward regardless of the time of response.

[0308] 2. Total Reward Value Formula

[0309] Total Reward Value: This is the sum of all rewards a user receives throughout their participation.

[0310] Single reward value: The reward a user receives each time he responds to grid demand, calculated by the previous formula.

[0311] Response adjustment times: refers to the total number of times the user participates in grid demand response during the entire agreement period.

[0312] Practical significance

[0313] The significance of this mechanism is that by setting a reward formula, users are encouraged to respond to demand in a specific time period, thereby helping to balance the load on the power grid. At the same time, the more times a user participates, the higher the total reward, which encourages users to participate multiple times and increase their help to the power grid.

[0314] The above is the control process of the flexible load participating in the grid interaction provided in the embodiment of the present application.

[0315] In summary, after obtaining the reservation data of the water heater cluster in the preset area, the embodiment of the present application obtains the time series data of the target water heater from the preset area according to the reservation data, and then trains the preset parameter adjustment model based on the time series data to obtain the heating parameter adjustment model. Finally, according to the heating parameter adjustment model and the demand data of the smart grid, the water heater cluster in the preset area is controlled. In the control scheme of the flexible load participating in the grid interaction provided in the present application, the preset parameter adjustment model can be trained based on the time series data to obtain the heating parameter adjustment model. Finally, according to the heating parameter adjustment model and the demand data of the smart grid, the water heater cluster in the preset area is controlled, that is, multiple water heaters can be adjusted dynamically in real time according to the demand data of the entire power grid, which can effectively avoid the peak of electricity consumption and ensure the normal operation of the entire power grid.

[0316] In order to facilitate better implementation of the control method of flexible load participating in grid interaction in the embodiment of the present application, the embodiment of the present application also provides a control device for flexible load participating in grid interaction. The meaning of the terms is the same as in the control system for flexible load participating in grid interaction mentioned above, and the specific implementation details can refer to the description in the system embodiment.

[0317] See also Figure 2 , Figure 2A schematic diagram of the structure of a control device for flexible loads participating in grid interaction provided in an embodiment of the present application, wherein the control device for flexible loads participating in grid interaction may specifically include a first acquisition module 201, a second acquisition module 202, a training model 203, and a control model 204, which may be specifically as follows:

[0318] The first acquisition module 201 is used to acquire the reservation data of the water heater cluster in a preset area.

[0319] The second acquisition module 202 is used to acquire time series data of a target water heater from a preset area according to the reservation data.

[0320] The training model 203 is used to train the preset parameter adjustment model based on the time series data to obtain the heating parameter adjustment model.

[0321] The control model 204 is used to control the water heater cluster in a preset area according to the heating parameter adjustment model and the demand data of the smart grid.

[0322] Optionally, in some embodiments of the present application, the training model 203 may specifically include:

[0323] An acquisition unit, used to acquire temperature information, heating power and user demand information corresponding to the water heater cluster from the time series data;

[0324] A construction unit, used to construct a first temperature change rate formula corresponding to the water heater cluster according to temperature information, heating power and user demand information;

[0325] A training unit is used to train a preset parameter adjustment model based on a first temperature change rate formula and time series data to obtain a heating parameter adjustment model.

[0326] Optionally, in some embodiments of the present application, the training unit may specifically include:

[0327] An acquisition subunit is used to acquire a preset optimization algorithm;

[0328] A construction subunit, used to construct a second temperature change rate formula according to the optimization algorithm and the first temperature change rate formula;

[0329] The training subunit is used to train the preset parameter adjustment model based on the second temperature change rate formula and time series data to obtain a heating parameter adjustment model.

[0330] Optionally, in some embodiments of the present application, the training subunit is specifically used for:

[0331] Extract time series features from time series data;

[0332] The preset parameter adjustment model is trained according to the time series characteristics and the second temperature change rate formula to obtain the heating parameter adjustment model.

[0333] Optionally, in some embodiments of the present application, the construction subunit may be specifically used for:

[0334] Get the preset mean square error function;

[0335] A second temperature change rate formula is constructed according to the mean square error function, the optimization algorithm and the first temperature change rate formula.

[0336] Optionally, in some embodiments of the present application, the control module 204 may be specifically used to:

[0337] Obtaining load information and demand response information of the smart grid from demand data of the smart grid;

[0338] Determine the heating plan corresponding to the water heater cluster in the preset area based on the heating parameter adjustment model, load information and demand response information;

[0339] Control clusters of water heaters in pre-set areas based on heating schedules.

[0340] Optionally, in some embodiments of the present application, the control module 204 may be specifically used to:

[0341] Fusion of load information and demand response information;

[0342] The fused information is input into the heating parameter adjustment model to determine the heating plan corresponding to the water heater cluster in the preset area.

[0343] The embodiment of the present application provides a control device for flexible loads to participate in grid interaction. After the first acquisition module 201 obtains the reservation data of the water heater cluster in the preset area, the second acquisition module 202 obtains the time series data of the target water heater from the preset area according to the reservation data. Then, the training model 203 trains the preset parameter adjustment model based on the time series data to obtain the heating parameter adjustment model. Finally, the control model 204 controls the water heater cluster in the preset area according to the heating parameter adjustment model and the demand data of the smart grid. In the control scheme for flexible loads to participate in grid interaction provided in the present application, the preset parameter adjustment model can be trained based on the time series data to obtain the heating parameter adjustment model. Finally, the control module 204 controls the water heater cluster in the preset area according to the heating parameter adjustment model and the demand data of the smart grid, that is, multiple water heaters can be adjusted dynamically in real time according to the demand data of the entire power grid, which can effectively avoid the peak of electricity consumption and ensure the normal operation of the entire power grid.

[0344] In addition, the present application also provides an electronic device, such as Figure 3As shown, it shows a schematic diagram of the structure of the electronic device involved in the embodiment of the present application, specifically:

[0345] The electronic device may include components such as a processor 301 with one or more processing cores, a memory 302 with one or more computer-readable storage media, a power supply 303, and an input unit 304. Those skilled in the art will appreciate that Figure 3 The electronic device structure shown in the figure does not constitute a limitation on the electronic device, and may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently.

[0346] The processor 301 is the control center of the electronic device. It uses various interfaces and lines to connect various parts of the entire electronic device. By running or executing software programs and / or modules stored in the memory 302, and calling data stored in the memory 302, it executes various functions of the electronic device and processes data, thereby monitoring the electronic device as a whole. Optionally, the processor 301 may include one or more processing cores; preferably, the processor 301 may integrate an application processor and a modem processor, wherein the application processor mainly processes the operating system, user interface, and application programs, and the modem processor mainly processes wireless communications. It is understandable that the above-mentioned modem processor may not be integrated into the processor 301.

[0347] The memory 302 can be used to store software programs and modules. The processor 301 executes various functional applications and a control method for flexible loads to participate in grid interaction by running the software programs and modules stored in the memory 302. The memory 302 may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other volatile solid-state storage device. Accordingly, the memory 302 may also include a memory controller to provide the processor 301 with access to the memory 302.

[0348] The electronic device also includes a power supply 303 for supplying power to each component. Preferably, the power supply 303 can be logically connected to the processor 301 through a power management system, so that the power management system can manage charging, discharging, power consumption and other functions. The power supply 303 can also include one or more DC or AC power supplies, recharging systems, power failure detection circuits, power converters or inverters, power status indicators and other arbitrary components.

[0349] The electronic device may further include an input unit 304, which may be used to receive input digital or character information and generate keyboard, mouse, joystick, optical or trackball signal input related to user settings and function control.

[0350] Specifically in this embodiment, the processor 301 in the electronic device will load the executable files corresponding to the processes of one or more applications into the memory 302 according to the following instructions, and the processor 301 will run the applications stored in the memory 302 to implement various functions, as follows:

[0351] Obtain reservation data for a water heater cluster in a preset area; obtain time series data for a target water heater from the preset area based on the reservation data; train a preset parameter adjustment model based on the time series data to obtain a heating parameter adjustment model; and control the water heater cluster in the preset area based on the heating parameter adjustment model and demand data of the smart grid.

[0352] The specific implementation of the above operations can be found in the previous embodiments, which will not be described in detail here.

[0353] After obtaining the reservation data of the water heater cluster in the preset area, the embodiment of the present application obtains the time series data of the target water heater from the preset area according to the reservation data, then trains the preset parameter adjustment model based on the time series data to obtain the heating parameter adjustment model, and finally controls the water heater cluster in the preset area according to the heating parameter adjustment model and the demand data of the smart grid. In the control scheme of the flexible load participating in the grid interaction provided by the present application, the preset parameter adjustment model can be trained based on the time series data to obtain the heating parameter adjustment model, and finally controls the water heater cluster in the preset area according to the heating parameter adjustment model and the demand data of the smart grid, that is, multiple water heaters can be adjusted dynamically in real time according to the demand data of the entire power grid, which can effectively avoid the peak of electricity consumption and ensure the normal operation of the entire power grid.

[0354] A person of ordinary skill in the art will appreciate that all or part of the steps in the various methods of the above embodiments may be completed by instructions, or by controlling related hardware through instructions. The instructions may be stored in a computer-readable storage medium and loaded and executed by a processor.

[0355] To this end, an embodiment of the present application provides a storage medium in which a plurality of instructions are stored, and the instructions can be loaded by a processor to execute the steps in any of the control methods for flexible loads to participate in grid interaction provided in the embodiments of the present application. For example, the instructions can execute the following steps:

[0356] Obtain reservation data for a water heater cluster in a preset area; obtain time series data for a target water heater from the preset area based on the reservation data; train a preset parameter adjustment model based on the time series data to obtain a heating parameter adjustment model; and control the water heater cluster in the preset area based on the heating parameter adjustment model and demand data of the smart grid.

[0357] The specific implementation of the above operations can be found in the previous embodiments, which will not be described in detail here.

[0358] The storage medium may include: a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0359] Since the instructions stored in the storage medium can execute the steps in any one of the control methods for flexible loads participating in grid interaction provided in the embodiments of the present application, the beneficial effects that can be achieved by any one of the control methods for flexible loads participating in grid interaction provided in the embodiments of the present application can be achieved. For details, please refer to the previous embodiments and will not be repeated here.

[0360] The above is a detailed introduction to a control method, device, electronic device and storage medium for flexible loads participating in grid interaction provided in an embodiment of the present application. Specific examples are used in this article to illustrate the principles and implementation methods of the present application. The description of the above embodiments is only used to help understand the method of the present application and its core idea; at the same time, for technical personnel in this field, according to the idea of ​​the present application, there will be changes in the specific implementation method and application scope. In summary, the content of this specification should not be understood as a limitation on the present application.

Claims

1. A control method for flexible loads to participate in grid interaction, characterized in that: include: Get the reservation data of the water heater cluster in the preset area; According to the reservation data, acquiring time series data of the target water heater from the preset area; Training a preset parameter adjustment model based on the time series data to obtain a heating parameter adjustment model; Based on the heating parameter adjustment model and demand data of the smart grid, a cluster of water heaters in a preset area is controlled.

2. The control method for flexible loads participating in grid interaction according to claim 1, characterized in that: The step of training a preset parameter adjustment model based on the time series data to obtain a heating parameter adjustment model includes: Obtaining temperature information, heating power and user demand information corresponding to the water heater cluster from the time series data; Constructing a first temperature change rate formula corresponding to the water heater cluster according to the temperature information, heating power and user demand information; The preset parameter adjustment model is trained based on the first temperature change rate formula and the time series data to obtain a heating parameter adjustment model.

3. The control method for flexible loads participating in grid interaction according to claim 2 is characterized in that: The method of training a preset parameter adjustment model based on the first temperature change rate formula and the time series data to obtain a heating parameter adjustment model includes: Get the preset optimization algorithm; Constructing a second temperature change rate formula according to the optimization algorithm and the first temperature change rate formula; The preset parameter adjustment model is trained based on the second temperature change rate formula and the time series data to obtain a heating parameter adjustment model.

4. The control method for flexible loads participating in grid interaction according to claim 3 is characterized in that: The preset parameter adjustment model is trained based on the second temperature change rate formula and the time series data to obtain a heating parameter adjustment model, including: Extracting time series features of the time series data; The preset parameter adjustment model is trained according to the time series characteristics and the second temperature change rate formula to obtain a heating parameter adjustment model.

5. The control method for flexible loads participating in grid interaction according to claim 3 is characterized in that: The step of constructing a second temperature change rate formula according to the optimization algorithm and the first temperature change rate formula includes: Get the preset mean square error function; A second temperature change rate formula is constructed based on the mean square error function, the optimization algorithm and the first temperature change rate formula.

6. The control method for flexible loads participating in grid interaction according to any one of claims 1 to 5, characterized in that: The controlling of the water heater cluster in a preset area according to the heating parameter adjustment model and the demand data of the smart grid includes: Obtaining load information and demand response information of the smart grid from demand data of the smart grid; Determine a heating plan corresponding to the water heater cluster in the preset area according to the heating parameter adjustment model, load information and demand response information; The water heater cluster in the preset area is controlled based on the heating plan.

7. The control method for flexible loads participating in grid interaction according to claim 6 is characterized in that: The step of determining a heating plan corresponding to the water heater cluster in the preset area according to the heating parameter adjustment model, load information and demand response information includes: fusing the load information and demand response information; The fused information is input into the heating parameter adjustment model to determine the heating plan corresponding to the water heater cluster in the preset area.

8. A control device for flexible loads to participate in grid interaction, characterized in that: include: A first acquisition module is used to acquire reservation data of a water heater cluster in a preset area; A second acquisition module, configured to acquire time series data of a target water heater from the preset area according to the reservation data; A training model is used to train a preset parameter adjustment model based on the time series data to obtain a heating parameter adjustment model; The control model is used to control the water heater cluster in a preset area according to the heating parameter adjustment model and the demand data of the smart grid.

9. An electronic device, characterized in that: include: A memory, a processor and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps of the control method for flexible loads participating in grid interaction as described in any one of claims 1 to 7 are performed.

10. A storage medium, characterized in that: A computer program is stored which can be loaded by a processor and execute the control method for flexible loads participating in grid interaction as claimed in any one of claims 1 to 7.

Images