A method and device for dynamically sharing energy storage allocation based on power demand response

Abstract

The present invention discloses a method and device for dynamic shared energy storage allocation based on power demand response, which belongs to the technical field of power system. The method includes: constructing a collective demand response prediction model of a virtual power plant; based on the collective demand response prediction model, constructing an objective function with the goal of maximizing the collective demand response, and constructing constraint conditions of the objective function based on energy storage equipment; using a column generation algorithm to generate a main problem and a sub-problem for the objective function for iterative solution to obtain a scheduling strategy for the energy storage equipment; obtaining the user's historical load data, using a load clustering method based on deep learning and adaptive clustering to cluster and analyze the historical load data, and identifying the user's load pattern; according to the user's load pattern and the scheduling strategy of the energy storage equipment, the energy storage equipment is allocated to the corresponding user. The present invention can help aggregators improve the utilization efficiency of energy storage systems and enhance the scheduling flexibility and responsiveness of virtual power plants.

Description

Technical Field

[0001] The present invention relates to the technical field of power systems, and in particular to a method and device for dynamically sharing energy storage allocation based on power demand response. Background Art

[0002] With the large-scale access to renewable energy sources such as solar and wind power, the volatility and uncertainty of the power system have placed higher demands on the resource coordination and scheduling capabilities of virtual power plant aggregators. The output of solar and wind power generation is significantly affected by factors such as weather and time, and is difficult to accurately predict, which increases the complexity of virtual power plants in balancing grid loads and optimizing energy distribution. At the same time, during peak load periods, the rapid rise in electricity demand may lead to inefficient resource allocation. Traditional scheduling methods are insufficient to meet this dynamic demand, further exacerbating the scheduling pressure on aggregators and potentially affecting service quality and profitability.

[0003] Although energy storage equipment plays an important role in regulating electricity demand, its limited capacity and loss of charging and discharging efficiency restrict its flexible application in multiple users and scenarios. For virtual power plant aggregators, how to effectively coordinate user demand and energy storage resource allocation and improve scheduling flexibility and economy has become the key to solving the above challenges. Summary of the invention

[0004] The purpose of the present invention is to overcome the deficiencies in the prior art and to provide a dynamic shared energy storage allocation method and device based on power demand response, which can help aggregators improve the utilization efficiency of energy storage systems, enhance the scheduling flexibility and responsiveness of virtual power plants, effectively alleviate peak load pressure, and ultimately achieve high efficiency, economy and stability in the operation of the power system.

[0005] To achieve the above object, the present invention is implemented by adopting the following technical solutions:

[0006] In a first aspect, the present invention provides a dynamic shared energy storage allocation method based on power demand response, comprising:

[0007] Constructing individual demand response prediction models for users, and integrating the individual demand response prediction models through an aggregation mechanism to generate a collective demand response prediction model for a virtual power plant;

[0008] Based on the collective demand response prediction model, an objective function is constructed with the goal of maximizing the collective demand response, and constraints of the objective function are constructed based on the energy storage device;

[0009] The objective function is subjected to iterative solution by using a column generation algorithm to generate a main problem and sub-problems to obtain a scheduling strategy for the energy storage device;

[0010] Acquire historical load data of users, perform cluster analysis on the historical load data using a load clustering method based on deep learning and adaptive clustering, and identify the load pattern of users;

[0011] The energy storage device is allocated to the corresponding user according to the load pattern of the user and the dispatching strategy of the energy storage device.

[0012] Optionally, the individual demand response prediction model is:

[0013] ;

[0014] In the formula, for The moment The individual demand response of each user, for The moment The electricity price sensitivity coefficient of each user, for The change in electricity price at each moment;

[0015] The collective demand response prediction model is:

[0016] ;

[0017] In the formula, for Collective demand response at all times, is the number of users.

[0018] Optionally, the objective function is:

[0019] ;

[0020] The constraints of the objective function are:

[0021] ;

[0022] ;

[0023] ;

[0024] ;

[0025] ;

[0026] ;

[0027] In the formula, For the The minimum and maximum capacity of each energy storage device, for The moment The remaining capacity and charge and discharge times of each energy storage device; The charging efficiency and discharging efficiency of the energy storage device; is the maximum number of charge and discharge times of the energy storage device; for The moment The discharge power of an energy storage device, No. The maximum discharge power of an energy storage device, is the time step.

[0028] Optionally, the generating a main problem and a sub-problem by using a column generation algorithm for the objective function for iterative solution includes:

[0029] Build the main question:

[0030] ;

[0031] In the formula, for The moment The contribution coefficient and decision variables of each energy storage device, including charging power and discharging power;

[0032] Constructing sub-problems:

[0033] ;

[0034] In the formula, is the dual variable of the objective function, is the dual variable of the constraint condition, is the dispatch strategy, which includes the decision variables of each energy storage device;

[0035] Iterate and solve the main problem and the sub-problem until the sub-problem cannot generate a new scheduling strategy Or when a preset number of iterations is reached, the optimal solution of the main problem is output.

[0036] Optionally, the adopting a load clustering method based on deep learning and adaptive clustering to perform cluster analysis on the historical load data includes:

[0037] Preprocessing the historical load data;

[0038] The pre-processed historical load data is subjected to feature extraction and dimension reduction processing by using a stacked autoencoder;

[0039] For the obtained low-dimensional features, the density peak clustering algorithm is used to identify the cluster centers and generate clustering results.

[0040] Optionally, the preprocessing of the historical load data includes:

[0041] Based on the box plot and mean-variance method, detecting abnormal data points in the historical load data, and removing the abnormal data points;

[0042] Using linear interpolation and time series models, predict missing data points in the historical load data, and fill in the predicted values ​​into the historical load data;

[0043] The historical load data is normalized.

[0044] In a second aspect, the present invention provides a dynamic shared energy storage allocation device based on power demand response, comprising:

[0045] A prediction model building module is configured to build an individual demand response prediction model for a user, and to integrate the individual demand response prediction models through an aggregation mechanism to generate a collective demand response prediction model for a virtual power plant;

[0046] An objective function construction module is configured to construct an objective function based on the collective demand response prediction model with the goal of maximizing the collective demand response, and to construct constraints of the objective function based on the energy storage device;

[0047] A scheduling strategy solving module is configured to use a column generation algorithm to generate a main problem and a sub-problem for the objective function and perform iterative solutions to obtain a scheduling strategy for the energy storage device;

[0048] A load pattern recognition module is configured to obtain historical load data of a user, perform cluster analysis on the historical load data using a load clustering method based on deep learning and adaptive clustering, and recognize the load pattern of the user;

[0049] The energy storage device allocation module is configured to allocate energy storage devices to corresponding users according to the user's load pattern and the scheduling strategy of the energy storage device.

[0050] In a third aspect, the present invention provides an electronic device, including a processor and a storage medium;

[0051] The storage medium is used to store instructions;

[0052] The processor is used to operate according to the instructions to execute the steps according to the above method.

[0053] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, which implements the steps of the above method when executed by a processor.

[0054] In a fifth aspect, the present invention provides a computer program product, comprising a computer program / instruction, which implements the steps of the above method when executed by a processor.

[0055] Compared with the prior art, the present invention has the following beneficial effects:

[0056] The present invention provides a dynamic shared energy storage allocation method and device based on power demand response, which 1) can flexibly and dynamically adjust the allocation of energy storage resources between different regions and users through the combination of real-time load forecasting, energy storage status monitoring and demand response mechanism. 2) Through the shared energy storage mode, the energy storage needs of multiple users are aggregated together, thereby greatly reducing construction and operating costs. 3) It provides a flexible, economical and sustainable operation plan for virtual power plant aggregators, which not only improves the stability and economic benefits of the power grid, but also promotes the development of green energy and promotes the realization of smart grids. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 It is a flow chart of a dynamic shared energy storage allocation method based on power demand response provided by an embodiment of the present invention;

[0058] Figure 2 It is a flowchart of solving the objective function based on the column generation algorithm provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0059] The present invention will be further described below in conjunction with the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and cannot be used to limit the protection scope of the present invention.

[0060] Embodiment 1:

[0061] like Figure 1 As shown, an embodiment of the present invention provides a dynamic shared energy storage allocation method based on power demand response, comprising the following steps:

[0062] Step S1: construct an individual demand response prediction model for users, and integrate the individual demand response prediction model through an aggregation mechanism to generate a collective demand response prediction model for the virtual power plant.

[0063] Specifically, in this embodiment, the key characteristics of user loads such as price sensitivity, load interruptibility, and response preference are analyzed to quantify the user's responsiveness to changes in electricity prices and load adjustment flexibility. These characteristics serve as the basis for demand response modeling and are used for subsequent dynamic response prediction.

[0064] (1) Price sensitivity: Define the price sensitivity of the user, assuming that the price change rate is proportional to the load response. Define the price sensitivity coefficient To represent the responsiveness of user 𝑖 to price changes:

[0065] ;

[0066] In the formula, For the The individual demand response of each user, For the The electricity price sensitivity coefficient of each user, The price sensitivity coefficient of different users is obtained through regression analysis of historical data.

[0067] (2) Load interruptibility analysis: Load interruptibility describes the upper limit of the load that the user can adjust. The maximum interruptible load ratio of a user is , that is, the part of user load that can be adjusted with the change of electricity price is:

[0068] ;

[0069] In the formula, For the The part of a user's load that can be adjusted with changes in electricity prices. For the The total load of users, .

[0070] (3) Different users have different acceptance of load regulation, so the user response flexibility coefficient is defined , and the permissible load regulation duration .

[0071] User demand is modeled based on the time series method, the user demand response curve is established, and the demand response changes at future moments are predicted as the input for dynamic shared energy storage allocation.

[0072] (1) Establish a demand response prediction model based on time series, and use the autoregressive integrated moving average model (ARIMA) to model the user load demand curve. Suppose the demand response of user 𝑖 is ,but:

[0073] ;

[0074] In the formula, are the model parameters of the autoregressive integrated moving average model, is the autoregressive integrated moving average function;

[0075] (2) User response coefficient in different time periods , , may change, and the coefficient value at future moments is predicted using the time series prediction model. The electricity price sensitivity coefficient of each user is , No. The maximum interruptible load ratio of a user is , No. The response flexibility coefficient of a user is ,but:

[0076] ;

[0077] ;

[0078] ;

[0079] In the formula, is the time series prediction function, are the model parameters of the time series forecasting model.

[0080] Specifically in this embodiment, by analyzing the user's sensitivity to electricity prices, an individual demand response prediction model is constructed:

[0081] ;

[0082] In the formula, for The moment The individual demand response of each user, for The moment The electricity price sensitivity coefficient of each user, for The change in electricity price at each moment;

[0083] The collective demand response prediction model is:

[0084] ;

[0085] In the formula, for Collective demand response at all times, is the number of users.

[0086] The system’s demand response volatility is analyzed through a predictive model, identifying peak and trough periods of demand so that peak periods can be prioritized when allocating energy storage. This analysis helps ensure the proper allocation of energy storage resources to reduce peak loads on the grid.

[0087] Step S2: Based on the collective demand response prediction model, an objective function is constructed with the goal of maximizing the collective demand response, and constraints of the objective function are constructed based on the energy storage device.

[0088] Specifically in this embodiment, the load response of the system is maximized so that the energy storage system can participate in the load response process as much as possible. Based on the load response data, the charging and discharging strategy of the energy storage device is designed to adjust the grid load and ultimately maximize the system benefits. The objective function is:

[0089] ;

[0090] The constraints of the objective function are:

[0091] ;

[0092] ;

[0093] ;

[0094] ;

[0095] ;

[0096] ;

[0097] In the formula, For the The minimum and maximum capacity of each energy storage device, for The moment The remaining capacity and charge and discharge times of each energy storage device; The charging efficiency and discharging efficiency of the energy storage device; is the maximum number of charge and discharge times of the energy storage device; for The moment The discharge power of an energy storage device, No. The maximum discharge power of an energy storage device, is the time step.

[0098] Step S3: Generate the main problem and sub-problems of the objective function using the column generation algorithm for iterative solution to obtain the scheduling strategy of the energy storage device, such as Figure 2 shown.

[0099] Build the main question:

[0100] ;

[0101] In the formula, for The moment The contribution coefficient and decision variables of each energy storage device, including charging power and discharging power;

[0102] In the initialization phase, a heuristic method is used to generate a set of feasible initial solutions as the starting solutions to the main problem.

[0103] Generate subproblems and add new columns. Subproblems are used to generate new candidate columns based on the solution of the main problem to further improve the objective function value. By fixing the dual variables of the main problem, the subproblem can be modeled as:

[0104] ;

[0105] In the formula, is the dual variable of the objective function, is the dual variable of the constraint condition, is the dispatch strategy, which includes the decision variables of each energy storage device;

[0106] Use dynamic programming or heuristic methods to solve subproblems and generate new energy storage allocation schemes (i.e., new columns) that can improve the target value. The generation rules of the new columns are as follows:

[0107] (1) Ensure that the energy storage equipment meets the constraints;

[0108] (2) Maximize the marginal benefit of the objective function;

[0109] (3) The contribution value of the new column must meet the improvement condition, that is, the marginal contribution value of the new column .

[0110] Marginal Contribution , is the value of the objective function after including the new column in the main problem, The objective function value when there are no new columns for the main problem.

[0111] Insert the generated new columns into the column set of the main problem and solve the main problem again. The objective value of the main problem is gradually improved through iteration.

[0112] Iterate and solve the main problem and sub-problems until the sub-problems cannot generate new scheduling strategies. (The marginal contribution values ​​for all new columns are ) or reaches the preset number of iterations, output the optimal solution to the main problem.

[0113] Optimization and acceleration of column generation algorithm When solving the main problem, the constraint matrix of the main problem is usually sparse, and only a few constraints and variables will change significantly in each iteration. Through sparse matrix technology, only non-zero elements related to the current solution are stored and calculated, avoiding global solution and thus reducing computational complexity.

[0114] In the main problem, only the constraints and variables involved in the current generated column are activated, and the redundant parts irrelevant to the current solution are ignored. Through the adaptive constraint management strategy, invalid constraints are dynamically deleted, thereby further improving the solution efficiency.

[0115] (1) Deletion criteria: If the dual variable value of a constraint does not affect the solution of the main problem in multiple consecutive iterations, the constraint can be temporarily ignored.

[0116] (2) Activation criteria: When a newly generated column takes effect on an ignored constraint, the constraint is immediately reactivated.

[0117] In order to speed up the generation of high-quality candidate columns in the subproblem, a column generation method based on a heuristic algorithm is adopted:

[0118] (1) Neighborhood search: Prioritize generating feasible sequences that satisfy key constraints such as capacity and power within the neighborhood of the current solution;

[0119] (2) Simulated annealing method: Use a random perturbation mechanism to search for the optimal solution region of candidate solutions, and control the exploration depth by annealing temperature.

[0120] The goal is to quickly obtain candidate columns that have significant marginal contributions to the main problem. After each iteration, the generated candidate columns are saved in the column pool and the priority of the columns is adjusted dynamically:

[0121] (1) If a column is not selected as the solution to the main problem for multiple consecutive times, its priority is lowered;

[0122] (2) If a column has made a significant contribution to the benefit of the main problem in recent iterations, its priority is increased.

[0123] Output the optimal energy storage allocation strategy for the main problem, including the charging and discharging decisions of each energy storage device at each moment. Calculate the total benefit based on the final solution and the efficiency of energy storage equipment , verify whether the allocation results meet the optimization objectives of actual demand response.

[0124] ;

[0125] ;

[0126] In the formula, for The moment The discharge power and charging power of an energy storage device.

[0127] By comparing with the historical benchmark value, verify whether this optimization has significantly improved the benefits. If the verification result does not meet expectations, the model and algorithm parameters can be further adjusted in combination with the feedback mechanism, and the column generation process can be rerun until the optimization goal is met.

[0128] Step S4: Obtain the user's historical load data, use a load clustering method based on deep learning and adaptive clustering to perform cluster analysis on the historical load data, and identify the user's load pattern.

[0129] Specifically in this embodiment, cluster analysis includes:

[0130] (1) Preprocessing of historical load data;

[0131] Based on the box plot and mean-variance method, detect and remove abnormal data points in historical load data;

[0132] Use linear interpolation and time series models to predict missing data points in historical load data and fill in the predicted values ​​into the historical load data;

[0133] Normalize the historical load data.

[0134] (2) For the preprocessed historical load data, a stacked autoencoder is used to perform feature extraction and dimensionality reduction.

[0135] Key characteristics that have a significant impact on load behavior include:

[0136] Time characteristics: Fluctuation pattern of daily load curve (e.g. peak time, trough time);

[0137] Price response characteristics: User sensitivity to electricity price fluctuations;

[0138] Seasonal characteristics: The changing trends of load demand in different seasons.

[0139] The encoder part of the stacked autoencoder maps high-dimensional feature data to a low-dimensional space. The decoder part optimizes the reconstruction error through back-propagation to ensure that the low-dimensional features can effectively restore the original data.

[0140] Its model expression is:

[0141] ;

[0142] ;

[0143] In the formula, Input feature data; is a low-dimensional hidden feature; and is the activation function; is the model parameter. is the restored original data. After training, the hidden features extracted It can be used for subsequent cluster analysis.

[0144] (3) For the obtained low-dimensional features, the density peak clustering algorithm is used to identify the cluster centers and generate clustering results.

[0145] Calculate the local density of data points :

[0146] ;

[0147] In the formula, For data points The local density of For data points The Euclidean distance between is the cutoff distance, determined by the percentile of all point pair distances;

[0148] Minimum distance Describes the minimum distance from point 𝑖 to a point with higher density, defined as:

[0149] ;

[0150] in, The larger it is, the further away point 𝑖 is from the high-density region.

[0151] Choose a and high As the cluster center, set the number of cluster centers 𝑘. Assign cluster labels, and assign non-center points to corresponding clusters according to the nearest high-density points; after clustering, each data point belongs to a cluster category.

[0152] Dynamic cluster center adjustment: recalculate the density peak point in real time and update the cluster center according to the change of load data. Merge the categories with lower density peak and reallocate the clusters to the isolated points. Combined with the deep learning prediction model, dynamically monitor the cluster quality, such as the silhouette coefficient or cluster score. Re-trigger the clustering process when the cluster quality decreases.

[0153] Identify typical load patterns, output clustering results, and identify typical patterns of various loads (such as peak and valley periods). Design exclusive energy storage allocation strategies for different load categories. Optimize the charging and discharging strategies of energy storage equipment based on clustering results, giving priority to meeting the needs of users in high-load categories.

[0154] Step S5: Allocate the energy storage device to the corresponding user according to the user's load pattern and the scheduling strategy of the energy storage device.

[0155] For example, the load patterns of users are classified into household, commercial and industrial. Household users have less fluctuations, and their corresponding energy storage is mainly discharged during peak hours; commercial users are relatively stable, and their corresponding energy storage balances the grid by discharging during peak loads; industrial users have large fluctuations in electricity demand, and flexibly adjust energy storage discharge according to production needs.

[0156] Embodiment 2:

[0157] An embodiment of the present invention provides a dynamic shared energy storage allocation device based on power demand response, comprising:

[0158] A prediction model building module is configured to build an individual demand response prediction model of the user, and integrate the individual demand response prediction models through an aggregation mechanism to generate a collective demand response prediction model of the virtual power plant;

[0159] An objective function building module is configured to build an objective function based on a collective demand response prediction model with the goal of maximizing the collective demand response, and to build constraints of the objective function based on an energy storage device;

[0160] The scheduling strategy solving module is configured to use a column generation algorithm to generate a main problem and a sub-problem for the objective function to perform iterative solutions to obtain a scheduling strategy for the energy storage device;

[0161] A load pattern recognition module is configured to obtain historical load data of a user, perform cluster analysis on the historical load data using a load clustering method based on deep learning and adaptive clustering, and recognize the load pattern of the user;

[0162] The energy storage device allocation module is configured to allocate energy storage devices to corresponding users according to the user's load pattern and the scheduling strategy of the energy storage device.

[0163] Embodiment three:

[0164] Based on the dynamic shared energy storage allocation method based on power demand response provided in the first embodiment, an embodiment of the present invention provides an electronic device, including a processor and a storage medium;

[0165] The storage medium is used to store instructions;

[0166] The processor is used to operate according to the instructions to execute the steps according to the above method.

[0167] Embodiment 4:

[0168] Based on the dynamic shared energy storage allocation method based on power demand response provided in Example 1, an embodiment of the present invention provides a computer-readable storage medium on which a computer program is stored, and the program implements the steps of the above method when executed by a processor.

[0169] Embodiment five:

[0170] Based on the dynamic shared energy storage allocation method based on power demand response provided in Example 1, an embodiment of the present invention provides a computer program product, including a computer program / instructions, which implement the steps of the above method when executed by a processor.

[0171] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0172] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0173] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0174] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0175] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A dynamic shared energy storage allocation method based on power demand response, characterized in that: include: Constructing individual demand response prediction models for users, and integrating the individual demand response prediction models through an aggregation mechanism to generate a collective demand response prediction model for a virtual power plant; Based on the collective demand response prediction model, an objective function is constructed with the goal of maximizing the collective demand response, and constraints of the objective function are constructed based on the energy storage device; The objective function is subjected to iterative solution by using a column generation algorithm to generate a main problem and sub-problems to obtain a scheduling strategy for the energy storage device; Acquire historical load data of users, perform cluster analysis on the historical load data using a load clustering method based on deep learning and adaptive clustering, and identify the load pattern of users; Allocating energy storage devices to corresponding users according to the load patterns of the users and the scheduling strategies of the energy storage devices; Wherein, the individual demand response prediction model is: ； In the formula, for The moment The individual demand response of each user, for The moment The electricity price sensitivity coefficient of each user, for The change in electricity price at each moment; The collective demand response prediction model is: ； In the formula, for Collective demand response at all times, is the number of users; The objective function is: ； The constraints of the objective function are: ； ； ； ； ； ； In the formula, For the The minimum and maximum capacity of each energy storage device, for The moment The remaining capacity and charge and discharge times of each energy storage device; The charging efficiency and discharging efficiency of the energy storage device; is the maximum number of charge and discharge times of the energy storage device; for The moment The discharge power of an energy storage device, No. The maximum discharge power of an energy storage device, is the time step; The step of using a column generation algorithm to generate a main problem and sub-problems for the objective function and iteratively solving the problem comprises: Build the main question: ； In the formula, for The moment The contribution coefficient and decision variables of each energy storage device, including charging power and discharging power; is the number of energy storage devices, is the number of moments; Constructing sub-problems: ； In the formula, is the dual variable of the objective function, is the dual variable of the constraint, is the dispatch strategy, which includes the decision variables of each energy storage device; Iterate and solve the main problem and the sub-problem until the sub-problem cannot generate a new scheduling strategy Or when a preset number of iterations is reached, the optimal solution of the main problem is output.

2. The dynamic shared energy storage allocation method based on power demand response according to claim 1 is characterized in that: The clustering analysis of the historical load data using the load clustering method based on deep learning and adaptive clustering includes: Preprocessing the historical load data; The pre-processed historical load data is subjected to feature extraction and dimension reduction processing by using a stacked autoencoder; For the obtained low-dimensional features, the density peak clustering algorithm is used to identify the cluster centers and generate clustering results.

3. The dynamic shared energy storage allocation method based on power demand response according to claim 2 is characterized in that: The preprocessing of the historical load data comprises: Based on the box plot and mean-variance method, detecting abnormal data points in the historical load data, and removing the abnormal data points; Using linear interpolation and time series models, predict missing data points in the historical load data, and fill in the predicted values ​​into the historical load data; The historical load data is normalized.

4. A dynamic shared energy storage allocation device based on power demand response, characterized in that: The dynamic shared energy storage allocation device is used to perform the steps of the method according to any one of claims 1 to 3, and the dynamic shared energy storage allocation device includes: A prediction model building module is configured to build an individual demand response prediction model for a user, and to integrate the individual demand response prediction models through an aggregation mechanism to generate a collective demand response prediction model for a virtual power plant; An objective function construction module is configured to construct an objective function based on the collective demand response prediction model with the goal of maximizing the collective demand response, and to construct constraints of the objective function based on the energy storage device; A scheduling strategy solving module is configured to use a column generation algorithm to generate a main problem and a sub-problem for the objective function and perform iterative solutions to obtain a scheduling strategy for the energy storage device; A load pattern recognition module is configured to obtain historical load data of a user, perform cluster analysis on the historical load data using a load clustering method based on deep learning and adaptive clustering, and recognize the load pattern of the user; The energy storage device allocation module is configured to allocate energy storage devices to corresponding users according to the user's load pattern and the scheduling strategy of the energy storage device.

5. An electronic device, characterized in that: including processor and storage medium; The storage medium is used to store instructions; The processor is configured to operate according to the instructions to execute the steps of the method according to any one of claims 1-3.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the method according to any one of claims 1 to 3 are implemented.

7. A computer program product comprising a computer program / instructions, characterized in that When the computer program / instructions are executed by a processor, the steps of the method according to any one of claims 1 to 3 are implemented.

Images