Virtual power plant collaborative scheduling optimization method, system, equipment and medium

Abstract

The invention discloses a virtual power plant collaborative scheduling optimization method, system and device and a medium, and relates to the technical field of electric energy collaborative control, and the technical scheme is characterized in that real-time demand response data of an aggregation resource load of a virtual power plant is acquired; calculating a predicted value of the electric energy dependent variable data of the real-time demand response data according to a regression relationship between the electric energy independent variable data and the dependent variable data; wherein the regression relationship between the electric energy independent variable data and the dependent variable data is obtained by performing regression analysis on historical demand response data based on a partial least square method; clustering the predicted values to determine the response type of the real-time demand response data; and obtaining an electric energy distribution data set of the virtual power plant, constructing a cooperative scheduling model of the virtual power plant according to the electric energy distribution data set, and optimizing the scheduling of the electric energy distribution of the response type in combination with the cooperative scheduling model.

Description

Technical Field

[0001] The present invention relates to the technical field of electric energy collaborative control, and more specifically, to a virtual power plant collaborative dispatch optimization method, system, equipment and medium. Background Art

[0002] The resource interaction of virtual power plants is a complex and dynamic process, which involves the aggregation, optimization and coordination of multiple distributed energy resources. Through efficient information interaction and collaborative control mechanisms, virtual power plants can achieve efficient resource utilization and maximize overall benefits.

[0003] Virtual power plants first need to effectively aggregate dispersed distributed energy resources. These distributed energy resources include distributed power generation (such as solar photovoltaic, wind power, etc.), energy storage systems, controllable loads (such as building air conditioning, charging stations, etc.), electric vehicles, etc. Through advanced information and communication technologies and software systems, virtual power plants can monitor and control the operating status of these resources in real time and incorporate them into a unified energy management system.

[0004] Based on the aggregation of distributed energy resources, virtual power plants further optimize and coordinate these resources. This includes rationally arranging the generation plan of distributed power sources, the charging and discharging strategy of energy storage systems, and the power consumption behavior of controllable loads according to the needs of the power market and the dispatch instructions of the power grid. Through optimization algorithms and models, virtual power plants can achieve efficient use of resources and minimize costs.

[0005] In the application process of virtual power plants, the distribution of distributed energy is often affected by many factors, such as unreasonable resource allocation and errors generated in the process of power distribution data optimization, which leads to waste of distributed energy resources and cannot effectively improve the execution efficiency of virtual power plant scheduling strategies. Summary of the invention

[0006] The purpose of the present invention is to provide a virtual power plant collaborative scheduling optimization method, system, device and medium to solve the problem that the prior art fails to reasonably allocate distributed energy resources.

[0007] In a first aspect of the present application, a virtual power plant collaborative scheduling optimization method is provided, the method comprising:

[0008] Obtain real-time demand response data on aggregated resource loads of virtual power plants;

[0009] According to the regression relationship between the electric energy independent variable data and the dependent variable data, the predicted value of the electric energy dependent variable data of the real-time demand response data is calculated; wherein the regression relationship between the electric energy independent variable data and the dependent variable data is obtained by performing regression analysis on the historical demand response data based on the partial least squares method;

[0010] Clustering the predicted values ​​to determine the response type of the real-time demand response data; wherein the response type includes similar aggregated resource load demand response and independent aggregated resource load demand response;

[0011] The electric energy distribution data set of the virtual power plant is obtained, and a collaborative scheduling model of the virtual power plant is constructed according to the electric energy distribution data set. The scheduling of the response type of electric energy distribution is optimized in combination with the collaborative scheduling model.

[0012] In one implementation, a regression analysis is performed on the historical demand response data based on the partial least squares method to obtain the regression relationship between the electric energy independent variable data and the dependent variable data of the historical demand response data, including:

[0013] Extracting an electric energy independent variable data group and an electric energy dependent variable data group of historical demand response data;

[0014] The electric energy independent variable data group and the electric energy dependent variable data group are processed respectively to obtain a standard electric energy independent variable data group and a standard electric energy dependent variable data group;

[0015] Extracting the largest first data group from the standard electric energy independent variable data group, and extracting the largest second data group from the standard electric energy dependent variable data group;

[0016] calculating a covariance matrix between the first data set and the second data set;

[0017] Determine, based on singular value decomposition of the covariance matrix, a weight vector group of electric energy independent variables of the first data group and a weight vector group of electric energy dependent variables of the second data group;

[0018] Calculate a first vector according to the first data group and the electric energy independent variable weight vector group;

[0019] Calculate a first regression coefficient of the influence of the independent variable according to the first vector and the first data group, and calculate a second regression coefficient of the influence of the dependent variable according to the first vector and the second data group;

[0020] Calculating a first residual matrix of the electric energy independent variable data set according to the first regression coefficient, the first vector and the first data set;

[0021] Calculating a second residual matrix of the electric energy dependent variable data set based on the second regression coefficient, the first vector, and the second data set;

[0022] The first residual matrix and the second residual matrix are used as the new electric energy independent variable data group and the electric energy dependent variable data group, and the process of calculating the new first residual matrix and the second residual matrix from the electric energy independent variable data group and the electric energy dependent variable data group is repeatedly performed until the number of residual matrices reaches a preset threshold, and the first regression coefficient matrix of the independent variable influence degree, the second regression coefficient matrix of the dependent variable influence degree and the electric energy independent variable weight vector matrix are obtained;

[0023] A regression relationship between the electric energy independent variable data and the dependent variable data is established based on the first regression coefficient matrix, the second regression coefficient matrix and the electric energy independent variable weight vector matrix.

[0024] In one implementation, the historical demand response data includes user power consumption data, adjustable load resource data, grid demand data, and demand response execution data;

[0025] The electric energy independent variable data group includes user power consumption influencing factor data of user power consumption data, power grid dispatch influencing factor data of adjustable load resource data, power grid total demand influencing factor data of power grid demand data, and user power consumption behavior influencing factor data of demand response execution data;

[0026] The electric energy dependent variable data group includes the actual power consumption data of the user power consumption data, the actual adjusted power of the adjustable load of the adjustable load resource data, the actual total demand of the power grid of the power grid demand data and the actual transfer power of the user of the demand response execution data.

[0027] In one implementation, clustering the predicted values ​​to determine the response type of the real-time demand response data includes:

[0028] Calculate the similarity of the predicted values ​​of the electric energy dependent variable data, and construct a similarity matrix based on the similarity;

[0029] The similarity matrix is ​​clustered, and the response type of the real-time demand response data is determined based on the clustering result.

[0030] In one implementation, an energy distribution data set of a virtual power plant is obtained, and a collaborative scheduling model of the virtual power plant is constructed according to the energy distribution data set, including:

[0031] Obtaining power distribution constraint data of the virtual power plant, wherein the power distribution constraint data includes power generation side data, energy storage side data, load side data and power grid operation data;

[0032] According to the power distribution constraint condition data, determine the power distribution data set of the virtual power plant;

[0033] Using the gradient descent method, the energy distribution function of the virtual power plant is preset, the energy distribution data set is initialized, and the energy distribution initialization data set is obtained. The gradient of the energy distribution loss function with respect to the energy distribution function is calculated based on the energy distribution initialization data set. The number of iterations is preset, and the difference between the energy distribution initialization data of the virtual power plant and the gradient is calculated, and the product with the learning rate is calculated until the preset number of iterations is reached to obtain the optimal learning rate and the optimal parameter vector;

[0034] According to the optimal learning rate and optimal parameter vector, a collaborative scheduling model of the virtual power plant is constructed.

[0035] In a second aspect of the present application, a virtual power plant collaborative scheduling optimization system is provided, the system comprising:

[0036] A data acquisition module, used to acquire real-time demand response data of aggregated resource loads of the virtual power plant;

[0037] A prediction module, used to calculate the predicted value of the electric energy dependent variable data of the real-time demand response data according to the regression relationship between the electric energy independent variable data and the dependent variable data; wherein the regression relationship between the electric energy independent variable data and the dependent variable data is obtained by performing regression analysis on the historical demand response data based on the partial least squares method;

[0038] A response type determination module is used to cluster the predicted values ​​to determine the response type of the real-time demand response data; wherein the response type includes similar aggregated resource load demand response and independent aggregated resource load demand response;

[0039] The scheduling optimization module is used to obtain the power distribution data set of the virtual power plant, build a collaborative scheduling model of the virtual power plant based on the power distribution data set, and optimize the scheduling of response type power distribution in combination with the collaborative scheduling model.

[0040] In one implementation, the prediction module is further configured to:

[0041] Extracting an electric energy independent variable data group and an electric energy dependent variable data group of historical demand response data;

[0042] The electric energy independent variable data group and the electric energy dependent variable data group are processed respectively to obtain a standard electric energy independent variable data group and a standard electric energy dependent variable data group;

[0043] Extracting the largest first data group from the standard electric energy independent variable data group, and extracting the largest second data group from the standard electric energy dependent variable data group;

[0044] calculating a covariance matrix between the first data set and the second data set;

[0045] Determine, based on singular value decomposition of the covariance matrix, a weight vector group of electric energy independent variables of the first data group and a weight vector group of electric energy dependent variables of the second data group;

[0046] Calculate a first vector according to the first data group and the electric energy independent variable weight vector group;

[0047] Calculate a first regression coefficient of the influence of the independent variable according to the first vector and the first data group, and calculate a second regression coefficient of the influence of the dependent variable according to the first vector and the second data group;

[0048] Calculating a first residual matrix of the electric energy independent variable data set according to the first regression coefficient, the first vector and the first data set;

[0049] Calculating a second residual matrix of the electric energy dependent variable data set based on the second regression coefficient, the first vector, and the second data set;

[0050] The first residual matrix and the second residual matrix are used as the new electric energy independent variable data group and the electric energy dependent variable data group, and the process of calculating the new first residual matrix and the second residual matrix from the electric energy independent variable data group and the electric energy dependent variable data group is repeatedly performed until the number of residual matrices reaches a preset threshold, and the first regression coefficient matrix of the independent variable influence degree, the second regression coefficient matrix of the dependent variable influence degree and the electric energy independent variable weight vector matrix are obtained;

[0051] A regression relationship between the electric energy independent variable data and the dependent variable data is established based on the first regression coefficient matrix, the second regression coefficient matrix and the electric energy independent variable weight vector matrix.

[0052] In one implementation, the historical demand response data includes user power consumption data, adjustable load resource data, grid demand data, and demand response execution data;

[0053] The electric energy independent variable data group includes user power consumption influencing factor data of user power consumption data, power grid dispatch influencing factor data of adjustable load resource data, power grid total demand influencing factor data of power grid demand data, and user power consumption behavior influencing factor data of demand response execution data;

[0054] The electric energy dependent variable data group includes the actual power consumption data of the user power consumption data, the actual adjusted power of the adjustable load of the adjustable load resource data, the actual total demand of the power grid of the power grid demand data and the actual transfer power of the user of the demand response execution data.

[0055] The third aspect of the present application provides an electronic device, which includes a processor, a memory, and a computer program stored in the memory and executable by the processor, wherein when the computer program is executed by the processor, the steps of a virtual power plant collaborative scheduling optimization method provided in the first aspect of the present application are implemented.

[0056] The fourth aspect of the present application provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of a virtual power plant collaborative scheduling optimization method provided in the first aspect of the present application are implemented.

[0057] Compared with the prior art, the present invention has the following beneficial effects:

[0058] In a virtual power plant collaborative scheduling optimization method provided in an embodiment of the present application, the predicted value of the electric energy dependent variable data of the real-time demand response data is predicted by the regression relationship obtained by regression analysis of the historical demand response data based on the partial least squares method, and the similarity is evaluated, so as to classify the demand response more accurately. Secondly, a collaborative scheduling model of a virtual power plant is constructed, and a joint optimization decision can be made for the parameter configuration of a higher dimension, and the current virtual power plant electric energy distribution data can be optimized to obtain a higher performance gain upper limit; finally, different treatments are performed for electric energy optimization for different situations (with high similarity and without high similarity), and the collaborative scheduling model of the virtual power plant is applied to the shared scheduling optimization strategy for the aggregated resource load demand response, and the collaborative scheduling model of the virtual power plant is applied to the independent scheduling optimization strategy for the aggregated resource load demand response, which can optimize all similar virtual power plant data at the same time, and the strategy obtained by reinforcement learning can quickly and dynamically adjust the parameter configuration when the electric energy state changes, so as to achieve the purpose of improving the execution efficiency of the virtual power plant scheduling strategy. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] The drawings described herein are used to provide a further understanding of the embodiments of the present invention, constitute a part of this application, and do not constitute a limitation of the embodiments of the present invention. In the drawings:

[0060] Figure 1 A flowchart of a virtual power plant collaborative scheduling optimization method provided in an embodiment of the present application;

[0061] Figure 2 A principle block diagram of a virtual power plant collaborative scheduling optimization system provided in an embodiment of the present application. DETAILED DESCRIPTION

[0062] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with embodiments and drawings. The exemplary embodiments of the present invention and their description are only used to explain the present invention and are not intended to limit the present invention.

[0063] It should be noted that the terms "include" or "may include" used in various embodiments of the present application indicate the presence of the function, operation or element applied for, and do not limit the addition of one or more functions, operations or elements. In addition, as used in various embodiments of the present application, the terms "include", "have" and their cognates are only intended to indicate specific features, numbers, steps, operations, elements, components or a combination of the foregoing items, and should not be understood as first excluding the presence of one or more other features, numbers, steps, operations, elements, components or a combination of the foregoing items or the possibility of adding one or more features, numbers, steps, operations, elements, components or a combination of the foregoing items.

[0064] It should be understood that terms such as "first" and "second" are only used for descriptive purposes and should not be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0065] Please refer to Figure 1 , Figure 1 A flowchart of a virtual power plant collaborative scheduling optimization method provided in an embodiment of the present application is shown in FIG. Figure 1 As shown, the method includes:

[0066] S101, obtaining real-time demand response data of the aggregated resource load of the virtual power plant.

[0067] In this embodiment, the real-time demand response data of the aggregated resource loads includes user power consumption data, adjustable load resource data, grid demand data, and demand response execution data.

[0068] S102, calculating the predicted value of the electric energy dependent variable data of the real-time demand response data according to the regression relationship between the electric energy independent variable data and the dependent variable data; wherein the regression relationship between the electric energy independent variable data and the dependent variable data is obtained by performing regression analysis on the historical demand response data based on the partial least squares method.

[0069] In this embodiment, determining the regression relationship between the electric energy independent variable data and the dependent variable data includes the following steps:

[0070] Extract the electric energy independent variable data group and the electric energy dependent variable data group of the historical demand response data; process the electric energy independent variable data group and the electric energy dependent variable data group respectively to obtain the standard electric energy independent variable data group and the standard electric energy dependent variable data group; extract the largest first data group in the standard electric energy independent variable data group, and extract the largest second data group in the standard electric energy dependent variable data group; calculate the covariance matrix between the first data group and the second data group; determine the electric energy independent variable weight vector group and the electric energy dependent variable weight vector group of the first data group based on the singular value decomposition of the covariance matrix; calculate the first vector according to the first data group and the electric energy independent variable weight vector group; calculate the first regression coefficient of the independent variable influence degree according to the first vector and the first data group, and calculate the second regression coefficient of the dependent variable influence degree according to the first vector and the second data group. number; calculate the first residual matrix of the electric energy independent variable data group according to the first regression coefficient, the first vector and the first data group; calculate the second residual matrix of the electric energy dependent variable data group according to the second regression coefficient, the first vector and the second data group; use the first residual matrix and the second residual matrix as new electric energy independent variable data group and electric energy dependent variable data group, and repeat the process of calculating the new first residual matrix and the second residual matrix from the electric energy independent variable data group and the electric energy dependent variable data group until the number of residual matrices reaches a preset threshold, and obtain the first regression coefficient matrix of the independent variable influence degree, the second regression coefficient matrix of the dependent variable influence degree and the electric energy independent variable weight vector matrix; establish the regression relationship between the electric energy independent variable data and the dependent variable data according to the first regression coefficient matrix, the second regression coefficient matrix and the electric energy independent variable weight vector matrix.

[0071] In this embodiment, the historical demand response data includes user power consumption data, adjustable load resource data, grid demand data and demand response execution data;

[0072] The electric energy independent variable data group includes user power consumption influencing factor data of user power consumption data, power grid dispatch influencing factor data of adjustable load resource data, power grid total demand influencing factor data of power grid demand data, and user power consumption behavior influencing factor data of demand response execution data;

[0073] The electric energy dependent variable data group includes the actual power consumption data of the user power consumption data, the actual adjusted power of the adjustable load of the adjustable load resource data, the actual total demand of the power grid of the power grid demand data and the actual transfer power of the user of the demand response execution data.

[0074] Specifically, an electric energy independent variable data group X of historical demand response data is extracted, and the electric energy independent variable data group X includes the following: the electric energy independent variable data of user electricity consumption data is the data of factors affecting user electricity consumption (such as time (season, month, date, hour, etc.), temperature, humidity, economic activities, etc.), the electric energy independent variable data of adjustable load resource data is the data of factors affecting power grid dispatching (including factors such as power grid dispatching instructions, load forecast results, and electricity price changes), the electric energy independent variable data of power grid demand data is the data of factors affecting total power grid demand (such as macroeconomic conditions, population growth, urbanization process, industrial structure adjustment, etc.), and the electric energy independent variable data of demand response execution data is the factors affecting user electricity consumption behavior (including factors such as electricity price changes, power grid reliability threats, and government incentive policies).

[0075] The electric energy dependent variable data group Y of historical demand response data is extracted, and the electric energy dependent variable data group Y includes the following: the electric energy dependent variable data of user power consumption data is the actual power consumption data of the user, the electric energy dependent variable data of adjustable load resource data is the actual adjusted power of the adjustable load, the electric energy dependent variable data of power grid demand data is the actual total demand of the power grid, and the electric energy dependent variable data of demand response execution data is the actual transferred power of the user).

[0076] The electric energy independent variable data set X and the electric energy dependent variable data set Y are organized into matrix form. According to formula 1 and formula 2, each X and Y data set is centered and standardized to obtain the standard electric energy independent variable data set X * and the standard electric energy dependent variable data set Y * :

[0077]

[0078] Among them, μ represents the mean, which is a row vector composed of column means in matrix operations, and σ represents the standard deviation, which is a diagonal matrix composed of column standard deviations in matrix operations.

[0079] Extract each standard electric energy independent variable data set X * Maximum standard electric energy independent variable data set Data on factors affecting the maximum user's electricity consumption, data on factors affecting the maximum grid dispatch, data on factors affecting the total grid demand, factors affecting the maximum user's electricity consumption behavior, and data groups Y for each standard electricity dependent variable * Maximum standard electric energy dependent variable data set The actual power consumption data of the largest user, the actual power adjusted by the largest adjustable load, the maximum actual total demand of the power grid, and the actual power transferred by the largest user.

[0080] Based on each group and The weight vector of the maximum covariance is initialized. First, the weight vector of each group is calculated according to Formula 3. and The covariance matrix between : Among them, C XY yes and The covariance matrix between them, n is the number of samples.

[0081] Then calculate C according to Formula 4 XY Singular value decomposition of: C XY =UΣV Formula 4, where U is a left singular matrix, Σ is a diagonal matrix, and V is a right singular matrix.

[0082] Select the first column of U (the left singular vector corresponding to the largest singular value) as the weight vector w1, and the first column of V (the right singular vector corresponding to the largest singular value) as the weight vector c1, and then normalize the weight vectors to ensure that they are unit vectors. The corresponding power independent variable weight vector group W1, user power consumption influencing factor data, first power grid dispatch influencing factor data, power grid total demand influencing factor data, maximum user power consumption behavior influencing factor, and The corresponding power dependent variable weight vector group C1 includes the actual power consumption data of users, the actual power adjusted by the adjustable load, the actual total demand of the power grid and the actual power transferred by users.

[0083] The first vector t1 is calculated according to formula 5, and the first vector is normalized.

[0084] Perform variable influence degree regression analysis on W1 and C1 (used to analyze the influence degree of the first electric energy independent variable weight vector group on the first electric energy dependent variable weight vector group). Calculate according to formula 6 and formula 7 respectively The first regression coefficient p1 of the first set of independent variables, and The second regression coefficient q1 of the degree of influence of the first set of dependent variables is normalized.

[0085]

[0086] According to Formula 8 and Formula 9, the electric energy variable residual matrices E1 and F1 of the electric energy independent variable data X and the electric energy dependent variable data Y are calculated respectively: Among them, E1 is the first residual matrix, and F1 is the second residual matrix.

[0087] Since the first vector t1 is determined and the data groups X and Y are updated, E1 and F1 are used as new X and Y, and the above process is repeated. The value of the residual matrix of the electric energy variable is preset, and the iterative optimization is performed until the value of the residual matrix of the electric energy variable is reached. The first regression coefficient and the second regression coefficient of each data group can be calculated, and then the regression model of X and Y is established according to formula 10, Y = XW (PTW) -1 Q formula 10, where W is the weight vector matrix, each column is a weight vector w a , P is the loading vector matrix of the independent variables, each column is a first regression coefficient p a , Q is the loading vector matrix of the dependent variable, each column is a second regression coefficient q a .

[0088] Cross-validation is used to evaluate the performance of the regression coefficients. Specifically, the data is randomly divided into A equal and mutually exclusive subsets (usually A is selected as 5 or 10), and for each component number k (from 1 to the maximum component number k), max ), for each subset, the data other than the subset is used as the training set, a PLS model is trained, k components are used, and the trained model is used to predict the subset to obtain the predicted value, and the root mean square (RMSE) prediction error is calculated according to the formula. For each component number k, the average prediction error of all A subsets is calculated according to formula 11 and formula 12.

[0089]

[0090] Among them, N i is the number of samples in the i-th fold, RMSE(k) is the average root mean square error, y ij is the true value of the jth sample in the i-th fold, is the corresponding predicted value, and the number of components that minimizes the average error is selected as the optimal number of components k best .

[0091] S103, clustering the predicted values ​​to determine the response type of the real-time demand response data; wherein the response type includes similar aggregated resource load demand response and independent aggregated resource load demand response.

[0092] In this embodiment, the predicted values ​​are clustered to determine the response type of the real-time demand response data, including: calculating the similarity of the predicted values ​​of the electric energy dependent variable data, and constructing a similarity matrix based on the similarity; clustering the similarity matrix, and determining the response type of the real-time demand response data based on the clustering results.

[0093] Specifically, after calculating the predicted value of the electric energy dependent variable data of the real-time demand response data through the regression relationship, the predicted value is clustered. The clustering can be a similarity calculation method such as cosine similarity, Euclidean distance or Mahalanobis distance. The clustering algorithm uses DBSCAN or HDBSCAN clustering algorithm to cluster the data similarity evaluation, and divides the data with similarity into one group (for example, if the data is 1.1, 1.2, 1.3, 2, 5, then 1.1, 1.2, and 1.3 will eventually be grouped together). According to the results of the similarity evaluation, similar aggregated resource load demand responses are obtained, and the lower similarity (such as the remaining 2 and 5) are divided into another group to obtain independent aggregated resource load demand responses.

[0094] S104, obtaining an electric energy distribution data set of the virtual power plant, constructing a collaborative scheduling model of the virtual power plant according to the electric energy distribution data set, and optimizing the scheduling of the response type of electric energy distribution in combination with the collaborative scheduling model.

[0095] In this embodiment, a collaborative scheduling model of a virtual power plant is constructed based on an electric energy distribution data set, including: obtaining electric energy distribution constraint data of the virtual power plant, wherein the electric energy distribution constraint data includes power generation side data, energy storage side data, load side data and power grid operation data; determining an electric energy distribution data set of the virtual power plant based on the electric energy distribution constraint data; using a gradient descent method, presetting an electric energy distribution function of the virtual power plant, initializing the electric energy distribution data set, obtaining an electric energy distribution initialization data set, calculating the gradient of an electric energy distribution loss function with respect to the electric energy distribution function based on the electric energy distribution initialization data set, presetting the number of iterations, calculating the difference between the electric energy distribution initialization data of the virtual power plant and the gradient, and calculating the product with the learning rate until the preset number of iterations is reached to obtain an optimal learning rate and an optimal parameter vector; constructing a collaborative scheduling model of the virtual power plant based on the optimal learning rate and the optimal parameter vector.

[0096] Specifically, a collaborative initial dispatch optimization model of a virtual power plant is constructed, the power parameters of the collaborative dispatch model of the virtual power plant are adjusted, the number of iterations is preset, and the iterative optimization is performed until the preset number of iterations is reached, and finally the collaborative dispatch model of the virtual power plant is obtained. The construction steps of the collaborative dispatch model of the virtual power plant are as follows:

[0097] Obtain the power distribution constraint data of the virtual power plant. The power distribution constraint data set of the virtual power plant includes power generation side data, energy storage side data, load side data, and power grid operation data. The power distribution data set of the virtual power plant is obtained, and a collaborative initial dispatch model of the virtual power plant is constructed.

[0098] Using the gradient descent algorithm, the target virtual power plant energy allocation function (including minimum power generation operation cost, minimum energy storage cost, minimum grid operation cost, grid stability index data, etc.) is set, the virtual power plant energy allocation data set is initialized, the virtual power plant energy allocation initialization data is obtained, and the gradient of the energy allocation loss function MSE with respect to the target virtual power plant energy allocation function is calculated. The number of iterations is preset, and the difference between the virtual power plant energy allocation initialization data and the gradient is calculated, and the product with the learning rate H is calculated. The gradient is iteratively updated according to the above calculation steps until the preset number of iterations is reached (by iteratively updating each parameter, the optimal solution is gradually approached. In each iteration, the parameter value is adjusted according to the set gradient direction of the target virtual power plant energy allocation function), and the collaborative scheduling model of the virtual power plant is obtained, wherein the update formula of the gradient descent algorithm is shown in the following formula 13:

[0099]

[0100] Among them, θ new is the updated optimal parameter vector, θ old is the parameter vector before updating, H is the learning rate (a hyperparameter used to control the step size of parameter updating), It is the gradient of the objective function J(θ) with respect to the parameter θ (i.e., the vector of partial derivatives corresponding to each parameter).

[0101] For a single parameter θ i The update formula is shown in Formula 14:

[0102] Among them, θnew is the optimal parameter vector after update, θold is the parameter vector before update, H is the learning rate, It is the partial derivative (i.e. gradient) of the loss function MES with respect to the parameter θ.

[0103] Specifically, after the iteration, the optimal learning rate H and the optimal parameter vector θ are obtained. new , the collected real-time virtual power plant energy distribution data set is input into the collaborative scheduling model of the virtual power plant, and the optimal parameters (optimal learning rate H, optimal parameter vector θ new ) calculates the optimal data on the power generation side, energy storage side, load side, and grid operation data (so that the virtual power plant can balance the demand and supply on each side and reduce costs under normal operation).

[0104] The scheduling optimization strategies for similar aggregated resource load demand responses and independent aggregated resource load demand responses are handled separately. Specifically, the collaborative scheduling model is applied to the shared scheduling optimization strategy for aggregated resource load demand response, and the response data of aggregated resource loads with similar demand responses are adjusted simultaneously. In addition, the independent adjustment scheduling optimization strategy for virtual power plants includes separate adjustments based on the response data of the aggregated resource load demand of the current virtual power plant, so that the adjustment results meet the demand.

[0105] Aggregate resource load demand response shared scheduling optimization strategy: collect and input similar response data sets of virtual power plants corresponding to real-time similar aggregate resource load demand responses into the collaborative scheduling model of virtual power plants, output optimal power generation side data, optimal energy storage side data, optimal load side data, and optimal power grid operation data, and apply the optimal power generation side data, optimal energy storage side data, optimal load side data, and optimal power grid operation data to all similar virtual power plants, and complete data optimization of all similar virtual power plants at the same time.

[0106] Aggregate resource load demand response independent scheduling optimization strategy: According to the demand, the response data of the load demand of the aggregated resources of the current virtual power plant is adjusted separately to obtain a target independent response data set. The target independent response data set includes target power generation side data, target energy storage side data, target load side data, and target grid operation data.

[0107] In summary, in a virtual power plant collaborative scheduling optimization method provided in this embodiment, the predicted value of the electric energy dependent variable data of the real-time demand response data is predicted by the regression relationship obtained by regression analysis of the historical demand response data based on the partial least squares method, and the similarity is evaluated, so as to classify the demand response more accurately. Secondly, the collaborative scheduling model of the virtual power plant is constructed, and the joint optimization decision of the parameter configuration of higher dimensions can be made, and the current virtual power plant electric energy distribution data can be optimized to obtain a higher performance gain upper limit; finally, different treatments are performed for electric energy optimization for different situations (with high similarity and without high similarity), and the collaborative scheduling model of the virtual power plant is applied to the shared scheduling optimization strategy of the aggregated resource load demand response, and the collaborative scheduling model of the virtual power plant is applied to the independent scheduling optimization strategy of the aggregated resource load demand response, which can optimize all similar virtual power plant data at the same time, and the strategy obtained by reinforcement learning can quickly and dynamically adjust the parameter configuration when the electric energy state changes, so as to achieve the purpose of improving the execution efficiency of the virtual power plant scheduling strategy.

[0108] Please refer to Figure 2 , Figure 2 A block diagram of a virtual power plant collaborative dispatch optimization system provided in an embodiment of the present application, such as Figure 2 As shown, the system includes:

[0109] A data acquisition module 210, for acquiring real-time demand response data of aggregated resource loads of a virtual power plant;

[0110] The prediction module 220 is used to calculate the predicted value of the electric energy dependent variable data of the real-time demand response data according to the regression relationship between the electric energy independent variable data and the dependent variable data; wherein the regression relationship between the electric energy independent variable data and the dependent variable data is obtained by performing regression analysis on the historical demand response data based on the partial least squares method;

[0111] The response type determination module 230 is used to cluster the predicted values ​​to determine the response type of the real-time demand response data; wherein the response type includes similar aggregated resource load demand response and independent aggregated resource load demand response;

[0112] The scheduling optimization module 240 is used to obtain the power distribution data set of the virtual power plant, build a collaborative scheduling model of the virtual power plant based on the power distribution data set, and optimize the scheduling of the power distribution of the response type in combination with the collaborative scheduling model.

[0113] Accordingly, in a virtual power plant collaborative dispatch optimization system provided in this embodiment, the predicted value of the electric energy dependent variable data of the real-time demand response data is predicted by the regression relationship obtained by regression analysis of the historical demand response data based on the partial least squares method, and the similarity is evaluated, so as to classify the demand response more accurately. Secondly, the collaborative dispatch model of the virtual power plant is constructed, and the joint optimization decision of the parameter configuration of higher dimensions can be made, and the current virtual power plant electric energy distribution data can be optimized to obtain a higher performance gain upper limit; finally, different treatments are performed for electric energy optimization for different situations (with high similarity and without high similarity), and the collaborative dispatch model of the virtual power plant is applied to the shared dispatch optimization strategy of the aggregated resource load demand response, and the collaborative dispatch model of the virtual power plant is applied to the independent dispatch optimization strategy of the aggregated resource load demand response, which can optimize all similar virtual power plant data at the same time, and the strategy obtained by reinforcement learning can quickly and dynamically adjust the parameter configuration when the electric energy state changes, so as to achieve the purpose of improving the execution efficiency of the virtual power plant dispatch strategy.

[0114] In some embodiments, the prediction module 220 is further configured to:

[0115] Extracting an electric energy independent variable data group and an electric energy dependent variable data group of historical demand response data;

[0116] The electric energy independent variable data group and the electric energy dependent variable data group are processed respectively to obtain a standard electric energy independent variable data group and a standard electric energy dependent variable data group;

[0117] Extracting the largest first data group from the standard electric energy independent variable data group, and extracting the largest second data group from the standard electric energy dependent variable data group;

[0118] calculating a covariance matrix between the first data set and the second data set;

[0119] Determine, based on singular value decomposition of the covariance matrix, a weight vector group of electric energy independent variables of the first data group and a weight vector group of electric energy dependent variables of the second data group;

[0120] Calculate a first vector according to the first data group and the electric energy independent variable weight vector group;

[0121] Calculate a first regression coefficient of the influence of the independent variable according to the first vector and the first data group, and calculate a second regression coefficient of the influence of the dependent variable according to the first vector and the second data group;

[0122] Calculating a first residual matrix of the electric energy independent variable data set according to the first regression coefficient, the first vector and the first data set;

[0123] Calculating a second residual matrix of the electric energy dependent variable data set based on the second regression coefficient, the first vector, and the second data set;

[0124] The first residual matrix and the second residual matrix are used as the new electric energy independent variable data group and the electric energy dependent variable data group, and the process of calculating the new first residual matrix and the second residual matrix from the electric energy independent variable data group and the electric energy dependent variable data group is repeatedly performed until the number of residual matrices reaches a preset threshold, and the first regression coefficient matrix of the independent variable influence degree, the second regression coefficient matrix of the dependent variable influence degree and the electric energy independent variable weight vector matrix are obtained;

[0125] A regression relationship between the electric energy independent variable data and the dependent variable data is established based on the first regression coefficient matrix, the second regression coefficient matrix and the electric energy independent variable weight vector matrix.

[0126] In some embodiments, the historical demand response data includes user power consumption data, adjustable load resource data, grid demand data, and demand response execution data;

[0127] The electric energy independent variable data group includes user power consumption influencing factor data of user power consumption data, power grid dispatch influencing factor data of adjustable load resource data, power grid total demand influencing factor data of power grid demand data, and user power consumption behavior influencing factor data of demand response execution data;

[0128] The electric energy dependent variable data group includes the actual power consumption data of the user power consumption data, the actual adjusted power of the adjustable load of the adjustable load resource data, the actual total demand of the power grid of the power grid demand data and the actual transfer power of the user of the demand response execution data.

[0129] The embodiment of the present application also provides an electronic device, which includes a processor, a memory, a communication interface and at least one communication bus for connecting the processor, the memory and the communication interface. The memory includes but is not limited to a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (PROM) or a portable CD-ROM, and the memory is used for related instructions and data.

[0130] The communication interface is used to receive and send data. The processor may be one or more CPUs. When the processor is a CPU, the CPU may be a single-core CPU or a multi-core CPU. The processor in the electronic device is used to read one or more programs stored in the memory and perform the following operations: obtaining the real-time demand response data of the aggregated resource load of the virtual power plant; calculating the predicted value of the electric energy dependent variable data of the real-time demand response data according to the regression relationship between the electric energy independent variable data and the dependent variable data; wherein the regression relationship between the electric energy independent variable data and the dependent variable data is obtained by regression analysis of the historical demand response data based on the partial least squares method; clustering the predicted values ​​to determine the response type of the real-time demand response data; wherein the response type includes similar aggregated resource load demand response and independent aggregated resource load demand response; obtaining the electric energy distribution data set of the virtual power plant, constructing the collaborative scheduling model of the virtual power plant according to the electric energy distribution data set, and optimizing the scheduling of the electric energy distribution of the response type in combination with the collaborative scheduling model.

[0131] It should be noted that the specific implementation of each operation can be Figure 1 The corresponding description of the method embodiment shown, the electronic device can be used to execute a virtual power plant collaborative scheduling optimization method of the above method embodiment of the present application, which will not be described in detail here.

[0132] In the embodiment of the present application, a computer-readable storage medium is also provided, and the computer-readable storage medium is a memory device in a computer device for storing programs and data. It is understandable that the computer-readable storage medium here can include both built-in storage media in a computer device and, of course, extended storage media supported by the computer device. The computer-readable storage medium provides a storage space, which stores the operating system of the terminal. In addition, one or more instructions suitable for being loaded and executed by a processor are also stored in the storage space, and these instructions can be one or more computer programs (including program codes). It should be noted that the computer-readable storage medium here can be a high-speed RAM memory or a non-volatile memory (non-volatile memory), such as at least one disk memory. The processor can load and execute one or more instructions stored in the computer-readable storage medium to implement the corresponding steps of a virtual power plant collaborative scheduling optimization method in the above embodiment. Those skilled in the art should understand that the embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can 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 codes.

[0133] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A virtual power plant collaborative scheduling optimization method, characterized in that: Methods include: Obtain real-time demand response data on aggregated resource loads of virtual power plants; According to the regression relationship between the electric energy independent variable data and the dependent variable data, the predicted value of the electric energy dependent variable data of the real-time demand response data is calculated; wherein the regression relationship between the electric energy independent variable data and the dependent variable data is obtained by performing regression analysis on the historical demand response data based on the partial least squares method; Clustering the predicted values ​​to determine the response type of the real-time demand response data; wherein the response type includes similar aggregated resource load demand response and independent aggregated resource load demand response; The electric energy distribution data set of the virtual power plant is obtained, and a collaborative scheduling model of the virtual power plant is constructed according to the electric energy distribution data set. The scheduling of the response type of electric energy distribution is optimized in combination with the collaborative scheduling model.

2. A virtual power plant collaborative scheduling optimization method according to claim 1, characterized in that: Based on the partial least squares method, the historical demand response data is regressed and the regression relationship between the independent variable data and the dependent variable data of the historical demand response data is obtained, including: Extracting an electric energy independent variable data group and an electric energy dependent variable data group of historical demand response data; The electric energy independent variable data group and the electric energy dependent variable data group are processed respectively to obtain a standard electric energy independent variable data group and a standard electric energy dependent variable data group; Extracting the largest first data group from the standard electric energy independent variable data group, and extracting the largest second data group from the standard electric energy dependent variable data group; calculating a covariance matrix between the first data set and the second data set; Determine, based on singular value decomposition of the covariance matrix, a weight vector group of electric energy independent variables of the first data group and a weight vector group of electric energy dependent variables of the second data group; Calculate a first vector according to the first data group and the electric energy independent variable weight vector group; Calculate a first regression coefficient of the influence of the independent variable according to the first vector and the first data group, and calculate a second regression coefficient of the influence of the dependent variable according to the first vector and the second data group; Calculating a first residual matrix of the electric energy independent variable data set according to the first regression coefficient, the first vector and the first data set; Calculating a second residual matrix of the electric energy dependent variable data set based on the second regression coefficient, the first vector, and the second data set; The first residual matrix and the second residual matrix are used as the new electric energy independent variable data group and the electric energy dependent variable data group, and the process of calculating the new first residual matrix and the second residual matrix from the electric energy independent variable data group and the electric energy dependent variable data group is repeatedly performed until the number of residual matrices reaches a preset threshold, and the first regression coefficient matrix of the independent variable influence degree, the second regression coefficient matrix of the dependent variable influence degree and the electric energy independent variable weight vector matrix are obtained; A regression relationship between the electric energy independent variable data and the dependent variable data is established based on the first regression coefficient matrix, the second regression coefficient matrix and the electric energy independent variable weight vector matrix.

3. A virtual power plant collaborative scheduling optimization method according to claim 2, characterized in that: The historical demand response data includes user power consumption data, adjustable load resource data, power grid demand data and demand response execution data; The electric energy independent variable data group includes user power consumption influencing factor data of user power consumption data, power grid dispatch influencing factor data of adjustable load resource data, power grid total demand influencing factor data of power grid demand data, and user power consumption behavior influencing factor data of demand response execution data; The electric energy dependent variable data group includes the actual power consumption data of the user power consumption data, the actual adjusted power of the adjustable load of the adjustable load resource data, the actual total demand of the power grid of the power grid demand data and the actual transfer power of the user of the demand response execution data.

4. A virtual power plant collaborative scheduling optimization method according to claim 1, characterized in that: Clustering the predicted values ​​determines the response type of the real-time demand response data, including: Calculate the similarity of the predicted values ​​of the electric energy dependent variable data, and construct a similarity matrix based on the similarity; The similarity matrix is ​​clustered, and the response type of the real-time demand response data is determined based on the clustering result.

5. The method for collaborative scheduling and optimization of a virtual power plant according to claim 1, characterized in that: Obtain the power distribution data set of the virtual power plant, and build a collaborative scheduling model of the virtual power plant based on the power distribution data set, including: Obtaining power distribution constraint data of the virtual power plant, wherein the power distribution constraint data includes power generation side data, energy storage side data, load side data and power grid operation data; According to the power distribution constraint condition data, determine the power distribution data set of the virtual power plant; Using the gradient descent method, the energy distribution function of the virtual power plant is preset, the energy distribution data set is initialized, and the energy distribution initialization data set is obtained. The gradient of the energy distribution loss function with respect to the energy distribution function is calculated based on the energy distribution initialization data set. The number of iterations is preset, and the difference between the energy distribution initialization data of the virtual power plant and the gradient is calculated, and the product with the learning rate is calculated until the preset number of iterations is reached to obtain the optimal learning rate and the optimal parameter vector; According to the optimal learning rate and optimal parameter vector, a collaborative scheduling model of the virtual power plant is constructed.

6. A virtual power plant collaborative dispatch optimization system, characterized in that: The system includes: A data acquisition module, used to acquire real-time demand response data of aggregated resource loads of the virtual power plant; A prediction module, used to calculate the predicted value of the electric energy dependent variable data of the real-time demand response data according to the regression relationship between the electric energy independent variable data and the dependent variable data; wherein the regression relationship between the electric energy independent variable data and the dependent variable data is obtained by performing regression analysis on the historical demand response data based on the partial least squares method; A response type determination module is used to cluster the predicted values ​​to determine the response type of the real-time demand response data; wherein the response type includes similar aggregated resource load demand response and independent aggregated resource load demand response; The scheduling optimization module is used to obtain the power distribution data set of the virtual power plant, build a collaborative scheduling model of the virtual power plant based on the power distribution data set, and optimize the scheduling of response type power distribution in combination with the collaborative scheduling model.

7. A virtual power plant collaborative dispatch optimization system according to claim 6, characterized in that: The prediction module is also used for: Extracting an electric energy independent variable data group and an electric energy dependent variable data group of historical demand response data; The electric energy independent variable data group and the electric energy dependent variable data group are processed respectively to obtain a standard electric energy independent variable data group and a standard electric energy dependent variable data group; Extracting the largest first data group from the standard electric energy independent variable data group, and extracting the largest second data group from the standard electric energy dependent variable data group; calculating a covariance matrix between the first data set and the second data set; Determine, based on singular value decomposition of the covariance matrix, a weight vector group of electric energy independent variables of the first data group and a weight vector group of electric energy dependent variables of the second data group; Calculate a first vector according to the first data group and the electric energy independent variable weight vector group; Calculate a first regression coefficient of the influence of the independent variable according to the first vector and the first data group, and calculate a second regression coefficient of the influence of the dependent variable according to the first vector and the second data group; Calculating a first residual matrix of the electric energy independent variable data set according to the first regression coefficient, the first vector and the first data set; Calculating a second residual matrix of the electric energy dependent variable data set based on the second regression coefficient, the first vector, and the second data set; The first residual matrix and the second residual matrix are used as the new electric energy independent variable data group and the electric energy dependent variable data group, and the process of calculating the new first residual matrix and the second residual matrix from the electric energy independent variable data group and the electric energy dependent variable data group is repeatedly performed until the number of residual matrices reaches a preset threshold, and the first regression coefficient matrix of the independent variable influence degree, the second regression coefficient matrix of the dependent variable influence degree and the electric energy independent variable weight vector matrix are obtained; A regression relationship between the electric energy independent variable data and the dependent variable data is established based on the first regression coefficient matrix, the second regression coefficient matrix and the electric energy independent variable weight vector matrix.

8. A virtual power plant collaborative dispatch optimization system according to claim 7, characterized in that: The historical demand response data includes user power consumption data, adjustable load resource data, power grid demand data and demand response execution data; The electric energy independent variable data group includes user power consumption influencing factor data of user power consumption data, power grid dispatch influencing factor data of adjustable load resource data, power grid total demand influencing factor data of power grid demand data, and user power consumption behavior influencing factor data of demand response execution data; The electric energy dependent variable data group includes the actual power consumption data of the user power consumption data, the actual adjusted power of the adjustable load of the adjustable load resource data, the actual total demand of the power grid of the power grid demand data and the actual transfer power of the user of the demand response execution data.

9. An electronic device, characterized in that: The electronic device includes a processor, a memory, and a computer program stored in the memory and executable by the processor, wherein when the computer program is executed by the processor, the steps of a virtual power plant collaborative scheduling optimization method as described in any one of claims 1 to 5 are implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, wherein when the computer program is executed by a processor, the steps of a virtual power plant collaborative scheduling optimization method as described in any one of claims 1 to 5 are implemented.

Images