Power dispatching strategy regulation and control method and system based on user load prediction

Abstract

The embodiment of the invention provides a regulation and control method and system for a power dispatching strategy based on user load prediction, and belongs to the technical field of energy dispatching of a power system. The regulation and control method comprises the following steps: acquiring historical load data of a user, and dividing the historical load data according to a preset time period to obtain user load distribution data in different time periods; acquiring historical weather data of the location, and dividing weather distribution data according to time periods to obtain weather distribution data in different time periods; determining user load increase data according to the user load distribution data in two adjacent time periods to obtain a plurality of user load increase data; taking the weather distribution data and the user load increase data as input, taking the user load distribution data as output, and constructing a data set for training a user load prediction model; and training and verifying a preset improved autoregressive comprehensive moving average model by adopting the data set to obtain a user load prediction model.

Description

Technical Field

[0001] The present invention relates to the technical field of energy dispatching of power systems, and in particular to a control method and system for a power dispatching strategy based on user load prediction. Background Art

[0002] In traditional power system energy dispatch, dispatch strategies are often based on real-time data, which includes immediate response to factors such as monitoring of the current power grid status, output capacity of power generation units, and capacity limitations of transmission networks. This dispatch mode usually focuses on solving short-term supply and demand balance problems and ensuring the stable operation of the power system. However, since this method takes less consideration of the expected power supply situation and user load change trends in the future, it may not be able to fully utilize existing resources or optimize power generation plans, thus affecting the economy and reliability of the entire power system.

[0003] The limitation of real-time dispatch is that it relies heavily on immediate information and makes insufficient use of future forecast information. For example, if the production plan of a large industrial user is known in advance, or if a large amount of solar and wind energy is predicted to be connected to the grid within a certain period of time, the dispatch center can plan in advance, adjust the output of traditional generators, or arrange the charging and discharging plan of energy storage equipment to better adapt to the expected load changes.

[0004] In addition, real-time dispatch often fails to fully consider the impact of various emergencies on the power system, such as reduced power generation capacity or damaged transmission lines caused by extreme weather conditions. These events may lead to power shortages or even blackouts, and the lack of foresight in dispatch strategies makes it difficult to take effective countermeasures in a timely manner.

[0005] Therefore, although real-time dispatch plays a key role in maintaining the immediate balance of the power system, its insufficient consideration of expected changes may lead to low dispatch efficiency, failure to maximize the efficiency of resource utilization, and may also affect the stability and economy of power supply. Summary of the invention

[0006] The purpose of the embodiments of the present invention is to provide a control method and system for a power dispatching strategy based on user load prediction, which can improve the efficiency of power dispatching.

[0007] In order to achieve the above object, an embodiment of the present invention provides a control method of a power dispatching strategy based on user load forecasting, comprising:

[0008] Acquire historical load data of users, and divide the historical load data according to preset time periods to obtain user load distribution data in different time periods;

[0009] Acquire historical weather data of the location, and divide the weather distribution data according to the time period to obtain weather distribution data in different time periods;

[0010] Determine user load growth data according to user load distribution data in two adjacent time periods to obtain a plurality of user load growth data;

[0011] Taking the weather distribution data and the user load growth data as input and the user load distribution data as output, a data set for training the user load prediction model is constructed;

[0012] The data set is used to train and verify a preset improved autoregressive integrated moving average model to obtain a user load forecasting model;

[0013] Obtaining user load distribution data within a predetermined time period according to the user load prediction model;

[0014] A power dispatching strategy is determined according to the user load distribution data.

[0015] Optionally, the time period is 48 hours.

[0016] Optionally, historical load data of users is obtained, and the historical load data is divided according to preset time periods to obtain user load distribution data in different time periods, including:

[0017] Obtaining multiple point values ​​of historical load data within the time period;

[0018] Divide the time period into a plurality of unit times of the same length, and divide the corresponding point values ​​at the same time, so that the unit time corresponds to a plurality of the point values, and each unit time corresponds to the same number of the point values;

[0019] Traversing each of the unit time, calculating the average and median of the corresponding multiple point values;

[0020] The average of the mean and the median is taken as the point value corresponding to the unit time.

[0021] Optionally, the unit time is 1 hour.

[0022] Optionally, obtaining historical weather data of a location and dividing the weather distribution data according to the time period to obtain weather distribution data in different time periods includes:

[0023] Integer encoding for different types of weather;

[0024] Obtaining a plurality of encoded values ​​of historical weather data within the time period;

[0025] Divide the time period into a plurality of unit times of the same length, and divide the corresponding code values ​​at the same time, so that the unit time corresponds to a plurality of the code values, and each unit time corresponds to the same number of the code values;

[0026] Traversing each of the unit time, and determining the mode of the code value;

[0027] The mode is used as the encoding value corresponding to the unit time.

[0028] Optionally, the improved autoregressive integrated moving average model includes:

[0029] The load data input layer is used to obtain the user load growth data to obtain a user feature matrix;

[0030] The weather data input layer is used to obtain the weather distribution data to obtain a weather matrix;

[0031] A data fusion layer, used for performing a matrix weighted fusion operation on the user feature matrix and the weather matrix to obtain a fusion feature;

[0032] The ARIMA layer is used to predict the user load distribution data according to the fusion features.

[0033] Optionally, the data set is used to train and verify a preset improved autoregressive integrated moving average model to obtain a user load forecasting model, including:

[0034] The improved autoregressive integrated moving average model is determined according to formula (1):

[0035]

[0036] Wherein, Y is the output of the improved autoregressive integrated moving average model, c is a constant term, u is a mean term, is the autocorrelation coefficient, θ1, θ2, θ q is the moving average coefficient, ∈ t-1 ,∈ t-q ,∈ t is the white noise error term, and p and q are the orders.

[0037] Optionally, the data set is used to train and verify a preset improved autoregressive integrated moving average model to obtain a user load forecasting model, including:

[0038] Formula (2) is used as the loss function.

[0039]

[0040] Where Y is the value of the loss function, n is the number of observations, and Y i is the actual value, is the predicted value.

[0041] On the other hand, the present invention further provides a control system for a power dispatching strategy based on user load forecasting, the control system comprising a processor, and the processor is configured to execute any of the control methods described above.

[0042] On the other hand, the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores instructions, wherein the instructions are used to be read by a machine so that the machine executes any of the control methods described above.

[0043] Through the above technical scheme, the embodiment of the present invention provides a control method and system for power dispatching strategy based on user load forecasting. The control method and system adopt an improved autoregressive integrated moving average model by combining the weather variables that affect the user load and the variables updated over time by the user equipment, thereby realizing the prediction of the future trend of changes in power load demand, thereby improving the efficiency and rationality of the power dispatching strategy.

[0044] Other features and advantages of the embodiments of the present invention will be described in detail in the subsequent detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] The accompanying drawings are used to provide a further understanding of the embodiments of the present invention and constitute a part of the specification. Together with the following specific implementations, they are used to explain the embodiments of the present invention, but do not constitute a limitation on the embodiments of the present invention. In the accompanying drawings:

[0046] Figure 1 is a flow chart of a control method of a power dispatching strategy based on user load forecasting according to an embodiment of the present invention;

[0047] Figure 2 is a flow chart of a method for determining point values ​​according to one embodiment of the present invention;

[0048] Figure 3 is a flow chart of a method for determining encoding of weather data according to one embodiment of the present invention. DETAILED DESCRIPTION

[0049] The specific implementation of the embodiment of the present invention is described in detail below in conjunction with the accompanying drawings. It should be understood that the specific implementation described here is only used to illustrate and explain the embodiment of the present invention, and is not used to limit the embodiment of the present invention.

[0050] like Figure 1FIG. 1 is a flow chart of a control method for a power dispatch strategy based on user load forecasting according to an embodiment of the present invention. Figure 1 In the method, the control method may include the following steps:

[0051] In step S10, historical load data of users is obtained, and the historical load data is divided according to a preset time period to obtain user load distribution data in different time periods;

[0052] In step S11, historical weather data of the location is obtained, and weather distribution data is divided according to time periods to obtain weather distribution data in different time periods;

[0053] In step S12, user load growth data is determined according to user load distribution data in two adjacent time periods to obtain a plurality of user load growth data;

[0054] In step S13, the weather distribution data and the user load growth data are used as inputs, and the user load distribution data is used as output to construct a data set for training a user load prediction model;

[0055] In step S14, the data set is used to train and verify the preset improved autoregressive integrated moving average model to obtain a user load prediction model;

[0056] In step S15, according to the user load prediction model, the user load distribution data within the predetermined time period is obtained;

[0057] In step S16, a power dispatching strategy is determined according to the user load distribution data.

[0058] In this Figure 1 In the method shown, step S10 is used to obtain the user's historical load data, and divide the historical load data according to a preset time period to obtain the user's load distribution data in different time periods. The historical load data may be data related to the power load of local low-voltage users and high-voltage users directly captured by the power system end and time. Since the historical load data includes a power load curve associated with time, in order to facilitate subsequent processing, in this embodiment, the power load curve may be segmented to obtain multiple time-related point values. Furthermore, the historical load data of the segmented point values ​​may be expressed using the following formula (1),

[0059]

[0060] Among them, X ij is the historical load data, x ijis the load value of the jth load point on the i-th group of lines, and 1≤i≤m, 1≤j≤n, m is the number of line groups, and n is the number of load points on each group of lines.

[0061] As for the time period, for historical load data of different total time lengths, the time period can be multiple values ​​known to those skilled in the art. Considering that in order to ensure the accuracy of the prediction, the historical load data generally spans several years or decades, in this embodiment, the time period can be 48 hours.

[0062] In addition, considering that the number of segmented point values ​​is large, if the subsequent network training is directly performed, it will obviously cause the problem of high complexity. Therefore, in an example of the present invention, the step S10 can also include the following Figure 2 In the method shown in Figure 2 In the step S10, the step S10 may further include the following method:

[0063] In step S20, multiple point values ​​of historical load data within a time period are obtained;

[0064] In step S21, the time period is divided into a plurality of unit times of the same length, and the corresponding point values ​​are divided at the same time, so that the unit time corresponds to a plurality of point values, and each unit time corresponds to the same number of point values. The unit time may be, for example, 1 hour;

[0065] In step S22, traverse each unit time and calculate the average and median of the corresponding multiple point values;

[0066] In step S23, the average of the mean and the median is taken as the point value corresponding to the unit time.

[0067] Step S11 can be used to obtain the historical weather data of the location, and divide the weather distribution data according to the time period to obtain the weather distribution data in different time periods. In this embodiment, it is considered that the weather distribution data is different from the historical load data. The values ​​of the historical load data are relatively discrete, and it is difficult for the same or identical point values ​​to appear. The weather distribution data is generally the code of different types of weather, and the possibility of the same code value appearing is relatively large. Therefore, step S11 can be further adopted as follows Figure 3 The method shown in is used to obtain the weather distribution data. Specifically, in Figure 3 In the step S11, the step S11 may further include the following method:

[0068] In step S30, different types of weather are integer-encoded;

[0069] In step S31, a plurality of coded values ​​of historical weather data within a time period are obtained;

[0070] In step S32, the time period is divided into a plurality of unit times of the same time length, and the corresponding code values ​​are divided at the same time, so that the unit time corresponds to a plurality of code values, and each unit time corresponds to the same number of code values;

[0071] In step S33, each unit time is traversed to determine the mode of the code value;

[0072] In step S34, the mode is used as the coding value corresponding to the unit time.

[0073] Step S12 can be used to determine user load growth data based on user load distribution data in two adjacent time periods to obtain multiple user load growth data. Specifically, in this embodiment, step S12 can be obtained by subtracting user load distribution data in two adjacent time periods.

[0074] Step S14 is used to train and verify the preset improved autoregressive integrated moving average model using a data set to obtain a user load prediction model. The improved autoregressive integrated moving average model may be in various forms known to those skilled in the art. In one example of the present invention, considering that the input of the data set is weather distribution data and user load growth data, and there are differences in the data dimensions of the two, in order to achieve feature fusion, the improved autoregressive integrated moving average model may include a load data input layer, a weather data input layer, a data fusion layer, and an ARIMA layer. The load data input layer may be used to obtain user load growth data to obtain a user feature matrix; the weather data input layer may be used to obtain weather distribution data to obtain a weather matrix; the data fusion layer may be used to perform a matrix weighted fusion operation on the user feature matrix and the weather matrix to obtain fusion features; the ARIMA layer may be used to predict user load distribution data based on fusion features. In addition, the specific method for training the improved autoregressive integrated moving average model may be in various forms known to those skilled in the art. In one example of the present invention, the improved autoregressive integrated moving average model may be determined using the following formula (2):

[0075]

[0076] Among them, Y t To improve the output of the autoregressive integrated moving average model, c is a constant term, u is a mean term, is the autocorrelation coefficient, θ1, θ2, θ q is the moving average coefficient, ∈ t-1 ,∈ t-q ,∈ t is the white noise error term, and p and q are the orders.

[0077] At the same time, formula (3) is used as the loss function in the training process.

[0078]

[0079] Where Y is the value of the loss function, n is the number of observations, and Y i is the actual value, is the predicted value.

[0080] On the other hand, the present invention also provides a control system for a power dispatch strategy based on user load forecasting, the control system comprising a processor, the processor being configured to execute any of the control methods described above. Specifically, the control method may include the following steps:

[0081] In step S10, historical load data of users is obtained, and the historical load data is divided according to a preset time period to obtain user load distribution data in different time periods;

[0082] In step S11, historical weather data of the location is obtained, and weather distribution data is divided according to time periods to obtain weather distribution data in different time periods;

[0083] In step S12, user load growth data is determined according to user load distribution data in two adjacent time periods to obtain a plurality of user load growth data;

[0084] In step S13, the weather distribution data and the user load growth data are used as inputs, and the user load distribution data is used as output to construct a data set for training a user load prediction model;

[0085] In step S14, the data set is used to train and verify the preset improved autoregressive integrated moving average model to obtain a user load prediction model;

[0086] In step S15, according to the user load prediction model, the user load distribution data within the predetermined time period is obtained;

[0087] In step S16, a power dispatching strategy is determined according to the user load distribution data.

[0088] In this Figure 1In the method shown, step S10 is used to obtain the user's historical load data, and divide the historical load data according to a preset time period to obtain the user's load distribution data in different time periods. The historical load data may be data related to the power load of local low-voltage users and high-voltage users directly captured by the power system end and time. Since the historical load data includes a power load curve associated with time, in order to facilitate subsequent processing, in this embodiment, the power load curve may be segmented to obtain multiple time-related point values. Furthermore, the historical load data of the segmented point values ​​may be expressed using the following formula (1),

[0089]

[0090] Among them, X ij is the historical load data, x ij is the load value of the jth load point on the i-th group of lines, and 1≤i≤m, 1≤j≤n, m is the number of line groups, and n is the number of load points on each group of lines.

[0091] As for the time period, for historical load data of different total time lengths, the time period can be multiple values ​​known to those skilled in the art. Considering that in order to ensure the accuracy of the prediction, the historical load data generally spans several years or decades, in this embodiment, the time period can be 48 hours.

[0092] In addition, considering that the number of segmented point values ​​is large, if the subsequent network training is directly performed, it will obviously cause the problem of high complexity. Therefore, in an example of the present invention, the step S10 can also include the following Figure 2 In the method shown in Figure 2 In the step S10, the step S10 may further include the following method:

[0093] In step S20, multiple point values ​​of historical load data within a time period are obtained;

[0094] In step S21, the time period is divided into a plurality of unit times of the same length, and the corresponding point values ​​are divided at the same time, so that the unit time corresponds to a plurality of point values, and each unit time corresponds to the same number of point values. The unit time may be, for example, 1 hour;

[0095] In step S22, traverse each unit time and calculate the average and median of the corresponding multiple point values;

[0096] In step S23, the average of the mean and the median is taken as the point value corresponding to the unit time.

[0097] Step S11 can be used to obtain the historical weather data of the location, and divide the weather distribution data according to the time period to obtain the weather distribution data in different time periods. In this embodiment, it is considered that the weather distribution data is different from the historical load data. The values ​​of the historical load data are relatively discrete, and it is difficult for the same or identical point values ​​to appear. The weather distribution data is generally the code of different types of weather, and the possibility of the same code value appearing is relatively large. Therefore, step S11 can be further adopted as follows Figure 3 The method shown in is used to obtain the weather distribution data. Specifically, in Figure 3 In the step S11, the step S11 may further include the following method:

[0098] In step S30, different types of weather are integer-encoded;

[0099] In step S31, a plurality of coded values ​​of historical weather data within a time period are obtained;

[0100] In step S32, the time period is divided into a plurality of unit times of the same time length, and the corresponding code values ​​are divided at the same time, so that the unit time corresponds to a plurality of code values, and each unit time corresponds to the same number of code values;

[0101] In step S33, each unit time is traversed to determine the mode of the code value;

[0102] In step S34, the mode is used as the coding value corresponding to the unit time.

[0103] Step S12 can be used to determine user load growth data based on user load distribution data in two adjacent time periods to obtain multiple user load growth data. Specifically, in this embodiment, step S12 can be obtained by subtracting user load distribution data in two adjacent time periods.

[0104] Step S14 is used to train and verify the preset improved autoregressive integrated moving average model using a data set to obtain a user load prediction model. The improved autoregressive integrated moving average model may be in various forms known to those skilled in the art. In one example of the present invention, considering that the input of the data set is weather distribution data and user load growth data, and there are differences in the data dimensions of the two, in order to achieve feature fusion, the improved autoregressive integrated moving average model may include a load data input layer, a weather data input layer, a data fusion layer, and an ARIMA layer. The load data input layer may be used to obtain user load growth data to obtain a user feature matrix; the weather data input layer may be used to obtain weather distribution data to obtain a weather matrix; the data fusion layer may be used to perform a matrix weighted fusion operation on the user feature matrix and the weather matrix to obtain fusion features; the ARIMA layer may be used to predict user load distribution data based on fusion features. In addition, the specific method for training the improved autoregressive integrated moving average model may be in various forms known to those skilled in the art. In one example of the present invention, the improved autoregressive integrated moving average model may be determined using the following formula (2):

[0105]

[0106] Among them, Y t To improve the output of the autoregressive integrated moving average model, c is a constant term, u is a mean term, is the autocorrelation coefficient, θ1, θ2, θ q is the moving average coefficient, ∈ t-1 ,∈ t-q ,∈ t is the white noise error term, and p and q are the orders.

[0107] At the same time, formula (3) is used as the loss function in the training process.

[0108]

[0109] Where Y is the value of the loss function, n is the number of observations, and Y i is the actual value, is the predicted value.

[0110] On the other hand, the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores instructions, wherein the instructions are used to be read by a machine so that the machine executes any of the control methods described above.

[0111] Through the above technical scheme, the embodiment of the present invention provides a control method and system for power dispatching strategy based on user load forecasting. The control method and system adopt an improved autoregressive integrated moving average model by combining the weather variables that affect the user load and the variables updated over time by the user equipment, thereby realizing the prediction of the future trend of changes in power load demand, thereby improving the efficiency and rationality of the power dispatching strategy.

[0112] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0113] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes 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 generate 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.

[0114] 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.

[0115] 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.

[0116] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0117] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0118] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0119] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0120] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included within the scope of the claims of the present application.

Claims

1. A control method for power dispatching strategy based on user load forecasting, characterized in that: The control method comprises: Acquire historical load data of users, and divide the historical load data according to preset time periods to obtain user load distribution data in different time periods; Acquire historical weather data of the location, and divide the historical weather data according to the time period to obtain weather distribution data in different time periods; Determine user load growth data according to user load distribution data in two adjacent time periods to obtain a plurality of user load growth data; Taking the weather distribution data and the user load growth data as input and the user load distribution data as output, a data set for training a user load prediction model is constructed; The data set is used to train and verify a preset improved autoregressive integrated moving average model to obtain a user load forecasting model; Obtaining user load distribution data within a predetermined time period according to the user load prediction model; A power dispatching strategy is determined according to the user load distribution data.

2. The control method according to claim 1, characterized in that: The time period is 48 hours.

3. The control method according to claim 1, characterized in that: Obtaining historical load data of users, and dividing the historical load data according to preset time periods to obtain user load distribution data in different time periods, including: Obtaining multiple point values ​​of historical load data within the time period; Divide the time period into a plurality of unit times of the same length, and divide the corresponding point values ​​at the same time, so that the unit time corresponds to a plurality of the point values, and each unit time corresponds to the same number of the point values; Traversing each of the unit time, calculating the average and median of the corresponding multiple point values; The average of the mean and the median is taken as the point value corresponding to the unit time.

4. The control method according to claim 3, characterized in that: The unit time is 1 hour.

5. The control method according to claim 1, characterized in that: Acquiring historical weather data of the location, and dividing the weather distribution data according to the time period to obtain weather distribution data in different time periods, including: Integer encoding for different types of weather; Obtaining a plurality of encoded values ​​of historical weather data within the time period; Divide the time period into a plurality of unit times of the same length, and divide the corresponding code values ​​at the same time, so that the unit time corresponds to a plurality of the code values, and each unit time corresponds to the same number of the code values; Traversing each of the unit time, and determining the mode of the code value; The mode is used as the encoding value corresponding to the unit time.

6. The control method according to claim 1, characterized in that: The improved autoregressive integrated moving average model includes: The load data input layer is used to obtain the user load growth data to obtain a user feature matrix; The weather data input layer is used to obtain the weather distribution data to obtain a weather matrix; A data fusion layer, used for performing a matrix weighted fusion operation on the user feature matrix and the weather matrix to obtain a fusion feature; The ARIMA layer is used to predict the user load distribution data according to the fusion features.

7. The control method according to claim 1, characterized in that: The data set is used to train and verify a preset improved autoregressive integrated moving average model to obtain a user load forecasting model, including: The improved autoregressive integrated moving average model is determined according to formula (1): Among them, Y t is the output of the improved autoregressive integrated moving average model, c is a constant term, u is a mean term, is the autocorrelation coefficient, θ1, θ2, θ q is the moving average coefficient, ∈ t-1 ,∈ t-q ,∈ t is the white noise error term, and p and q are the orders.

8. The control method according to claim 1, characterized in that: The data set is used to train and verify a preset improved autoregressive integrated moving average model to obtain a user load forecasting model, including: Formula (2) is used as the loss function. Where Y is the value of the loss function, n is the number of observations, and Y i is the actual value, is the predicted value.

9. A control system for power dispatch strategy based on user load forecasting, characterized in that: The control system comprises a processor, and the processor is configured to execute the control method according to any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores instructions, and the instructions are used to be read by a machine so that the machine executes the control method according to any one of claims 1 to 8.

Images