Power load prediction method and device, storage medium and electronic device

Abstract

The invention discloses a power load prediction method and device, a storage medium, an electronic device and a computer program product. The method comprises the following steps: acquiring workday load data and holiday load data of each enterprise in a target area; respectively determining the correlation between the workday load data and the holiday load data of each enterprise; for the first enterprise whose correlation is smaller than a preset threshold, predicting future holiday load data of the first enterprise according to historical holiday load data of the first enterprise; and for the second enterprise whose correlation is greater than or equal to a preset threshold, predicting future holiday load data of the second enterprise according to the historical load data of the second enterprise. By analyzing the correlation between the power loads of the workdays and the holidays of the enterprises, the enterprises with low correlation are predicted by using the historical load data of the holidays, and the enterprises with high correlation are predicted by directly using the historical load data, so that the prediction precision and reliability are effectively improved.

Description

Technical Field

[0001] The present application relates to the field of power load, and specifically, to a method and device for predicting power load, a storage medium, an electronic device, and a computer program product. Background Art

[0002] With the continued growth of electricity demand and the increasing complexity of power grid structures, power load forecasting has become a key basis for power dispatch and planning. Holiday load forecasting is particularly important because holiday load characteristics differ from weekdays, resulting in greater uncertainty and posing significant challenges to power system operation and dispatch.

[0003] In related technologies, historical average load is usually used for load forecasting.

[0004] However, the related technical approach does not produce accurate prediction results for some enterprises with large differences in load between holidays and weekdays. Summary of the Invention

[0005] Embodiments of the present application provide a method and device for predicting power load, a storage medium, an electronic device, and a computer program product.

[0006] According to one aspect of an embodiment of the present application, a method for predicting power load is provided, the method comprising: obtaining working day load data and holiday load data of each enterprise in a target area; determining the correlation between the working day load data and holiday load data of each enterprise based on the working day load data and holiday load data of each enterprise; for a first enterprise whose correlation is less than a preset threshold, predicting the future holiday load data of the first enterprise based on the historical holiday load data of the first enterprise; and, for a second enterprise whose correlation is greater than or equal to the preset threshold, predicting the future holiday load data of the second enterprise based on the historical load data of the second enterprise, wherein the historical load data of the second enterprise includes working day load data and holiday load data, and the future holiday load data of the first enterprise and the future holiday load data of the second enterprise are used to indicate the future holiday load prediction results of the enterprises in the target area.

[0007] In an exemplary embodiment, the correlation between the working day load data and the holiday load data of each enterprise is determined based on the working day load data and the holiday load data of each enterprise, including: obtaining a first time series of the working day load data and a second time series of the holiday load data of the target enterprise, wherein the first time series and the second time series both include multiple load data sampled at a preset frequency, and the target enterprise is any one of the enterprises; constructing a sample point set based on the first time series and the second time series, wherein the sample point set includes multiple sample points, and each sample point includes the load data of the first time series and the load data of the second time series corresponding to the time; determining the distribution of multiple sample points in the sample point set in a preset coordinate system, wherein the horizontal axis of the preset coordinate system is the working day load data, and the vertical axis of the preset coordinate system is the holiday load data; determining the correlation between the working day load data and the holiday load data of the target enterprise based on the distribution of the multiple sample points.

[0008] In an exemplary embodiment, the correlation between the working day load data and the holiday load data of a target enterprise is determined based on the distribution of multiple sample points, including: dividing a preset coordinate system into multiple grid units, wherein each grid unit corresponds to a value interval of the working day load data and a value interval of the holiday load data; counting the number of sample points that fall into each grid unit among the multiple sample points; determining the probability of the sample point falling into each grid unit according to the number of sample points in each grid unit; determining the mutual information value between the working day load data and the holiday load data of the target enterprise according to the probability of the sample point falling into each grid unit, wherein the mutual information value is used to indicate the correlation between the working day load data and the holiday load data of the target enterprise.

[0009] In an exemplary embodiment, for a first enterprise whose correlation is less than a preset threshold, future holiday load data of the first enterprise is predicted based on the historical holiday load data of the first enterprise, including: collecting characteristic data of historical holidays of the first enterprise, wherein the characteristic data includes environmental factors, enterprise operation factors, and holiday types; constructing a training sample set based on the characteristic data and historical holiday load data, wherein the training sample set includes historical holiday load data corresponding to different characteristic data; using the training sample set to train a preset neural network model so that the neural network model learns the relationship between the characteristic data and the historical holiday load data; inputting the characteristic data of future holidays of the first enterprise into the trained neural network model to obtain the future holiday load data of the first enterprise.

[0010] In an exemplary embodiment, for a second enterprise whose correlation is greater than or equal to a preset threshold, future holiday load data of the second enterprise is predicted based on the historical load data of the second enterprise, including: determining the changing trend of the load data of the second enterprise based on the historical load data of the second enterprise; determining the future holiday load data of the second enterprise based on the current load data of the second enterprise and the changing trend of the load data of the second enterprise.

[0011] In an exemplary embodiment, the method also includes: respectively determining the proportion of the historical holiday load data of the first enterprise and the historical holiday load data of the second enterprise in the historical holiday load data of all enterprises in the target area; respectively determining the weights corresponding to the first enterprise and the second enterprise based on the proportion of the historical holiday load data of the first enterprise and the second enterprise in the historical holiday load data of all enterprises in the target area; and weightedly summing the future holiday load data of the first enterprise and the second enterprise based on the weights corresponding to the first enterprise and the second enterprise to determine the future holiday load forecast results for the enterprises in the target area.

[0012] According to another aspect of an embodiment of the present application, a power load prediction device is also provided, which includes: a data acquisition module for acquiring working day load data and holiday load data of each enterprise in a target area; a correlation determination module for determining the correlation between the working day load data and holiday load data of each enterprise based on the working day load data and holiday load data of each enterprise; a first prediction module for predicting the future holiday load data of the first enterprise based on the historical holiday load data of the first enterprise for a first enterprise whose correlation is less than a preset threshold; a second prediction module for predicting the future holiday load data of the second enterprise based on the historical load data of the second enterprise for a second enterprise whose correlation is greater than or equal to a preset threshold, wherein the historical load data of the second enterprise includes working day load data and holiday load data, and the future holiday load data of the first enterprise and the future holiday load data of the second enterprise are used to indicate the future holiday load prediction results of the enterprises in the target area.

[0013] According to another aspect of the embodiments of the present application, a computer-readable storage medium is provided, in which a computer program is stored, wherein the computer program is configured to execute the above-mentioned power load prediction method when running.

[0014] According to another aspect of an embodiment of the present application, an electronic device is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the power load prediction method through the computer program.

[0015] According to another aspect of the embodiments of the present application, a computer program product is provided, including a computer program, which implements the steps of the method described in each embodiment of the present application when executed by a processor.

[0016] The above-described power load forecasting method, by analyzing the correlation between weekday and holiday load data, can more accurately predict an enterprise's future holiday power load. For enterprises with low correlation, historical holiday load data and characteristic data are used for forecasting. For enterprises with high correlation, historical load data from both weekdays and holidays is combined for forecasting, effectively improving the accuracy and reliability of the forecast. By analyzing the correlation between an enterprise's weekday and holiday power load, refined management of future holiday load forecasts for enterprises within a target area is achieved. Dividing enterprises into two categories fully accounts for changes in their electricity usage patterns during holidays, thereby improving the accuracy of load forecasts. For the second type of enterprise, no additional holiday data collection and analysis is required; historical load data can be directly used for forecasting, reducing the workload of data collection and preprocessing. While ensuring forecast accuracy, the present invention can focus more resources and computing power on analyzing the first type of enterprises, which are significantly affected by holidays, thereby achieving optimal resource allocation. This method effectively addresses the problem of inaccurate load forecasts caused by significant differences in enterprise holiday and weekday power usage. Through refined analysis and forecasting strategies, forecasting efficiency is improved while ensuring accuracy, providing strong support for power system scheduling and power market operations. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0018] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0019] Figure 1 This is a hardware structure block diagram of the power load prediction method according to an embodiment of the present application;

[0020] Figure 2 is a flow chart of a method for predicting power load according to an embodiment of the present application;

[0021] Figure 3 This is a second flow chart of a method for predicting power load according to an embodiment of the present application;

[0022] Figure 4This is a third flow chart of a method for predicting power load according to an embodiment of the present application;

[0023] Figure 5 This is a fourth flow chart of a method for predicting power load according to an embodiment of the present application;

[0024] Figure 6 This is a fifth flow chart of a method for predicting power load according to an embodiment of the present application;

[0025] Figure 7 This is a sixth flowchart of a method for predicting power load according to an embodiment of the present application;

[0026] Figure 8 This is a structural block diagram of a power load prediction device according to an embodiment of the present application. DETAILED DESCRIPTION

[0027] In order to enable those skilled in the art to better understand the present invention, the following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of this application.

[0028] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequential order. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in a sequence other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0029] The method embodiments provided in the embodiments of the present application can be executed in a computer terminal or similar computing device. Taking running on a computer terminal as an example, Figure 1 This is a hardware structure block diagram of a computer terminal for the power load prediction method of the embodiment of the present application. Figure 1 As shown, the computer terminal may include one or more ( Figure 1Only one is shown in the figure) a processor 102 (the processor 102 may include but is not limited to a microprocessor (Microprocessor Unit, referred to as MPU) or a programmable logic device (Programmable logic device, referred to as PLD)) and a memory 104 for storing data. In an exemplary embodiment, the computer terminal may further include a transmission device 106 for communication functions and an input / output device 108. It will be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the above-mentioned computer terminal. For example, the computer terminal may also include Figure 1 More or fewer components than shown, or with Figure 1 Equivalent functions or comparisons shown Figure 1 Shown are different configurations with more functionality.

[0030] The memory 104 can be used to store computer programs, for example, software programs and modules of application software, such as the computer program corresponding to the power load prediction method in the embodiment of the present application. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, that is, implementing the above-mentioned method. The memory 104 may include a high-speed random access memory and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 104 may further include a memory remotely located relative to the processor 102, and these remote memories may be connected to the computer terminal via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0031] The transmission device 106 is used to receive or transmit data via a network. A specific example of the aforementioned network may include a wireless network provided by a computer terminal's communications provider. In one embodiment, the transmission device 106 includes a network interface controller (NIC), which can be connected to other network devices via a base station to enable communication with the Internet. In another embodiment, the transmission device 106 may be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0032] As described in the background art, power companies and related institutions in the related art tend to rely on historical data for trend analysis and statistical deduction when making load forecasts. However, the emergence of holidays is often accompanied by changes in the business operation model. For example, manufacturing companies may suspend production, shorten working hours or adjust shifts on holidays, while retail and tourism companies may increase business hours, hold promotional activities or provide additional services on holidays. These changes in operating models directly affect the electricity demand of enterprises, resulting in a large difference in electricity load on holidays and weekdays. This leads to inaccurate prediction results in the related art. The power load prediction method provided in this application can fully consider the nonlinear relationship between the load on working days and holidays of an enterprise, conduct targeted analysis on enterprises with large differences, and improve the accuracy of enterprise load forecasting.

[0033] In this embodiment, a method for predicting power load is provided. Figure 2 4 is a flow chart of an optional power load prediction method according to an embodiment of the present application, the flow includes the following steps S200-S230:

[0034] Step S200: Obtain the workday load data and holiday load data of each enterprise in the target area.

[0035] Specifically, the weekday load data and holiday load data of each enterprise record the electricity consumption of each enterprise in different periods, which facilitates the subsequent analysis of the differences in the electricity consumption habits of enterprises between weekdays and holidays.

[0036] For example, you can collaborate with a power company to extract historical electricity usage records from its database. This data can include weekday and holiday electricity usage from past years, along with specific date and time stamps, to facilitate time series analysis. After data collection, preprocessing is required, including data cleaning (removing outliers and missing values), data alignment (ensuring temporal consistency between weekday and holiday data), and data standardization (making data comparable across different companies or at different time points).

[0037] Step S210 , determining the correlation between the working day load data and the holiday load data of each enterprise based on the working day load data and the holiday load data of each enterprise.

[0038] Specifically, determining the correlation between load data on weekdays and holidays can help identify which companies' electricity usage patterns change during holidays and which remain relatively stable. The results of this correlation analysis will serve as the basis for subsequent forecasting strategies.

[0039] For example, the Maximum Information Coefficient (MIC) algorithm can be used for nonlinear correlation analysis. First, the weekday load data is paired with the holiday load data to form a data set. The MIC method is then used to calculate the correlation between the two sets of data. The resulting value ranges from 0 to 1, with 1 indicating perfect correlation and 0 indicating no correlation. A threshold value (e.g., 0.5) is set, and the results are divided into two groups: the first enterprise with a correlation less than the threshold, and the second enterprise with a correlation greater than or equal to the threshold.

[0040] Step S220 : For the first enterprise whose correlation is less than a preset threshold, predict the future holiday load data of the first enterprise based on the historical holiday load data of the first enterprise.

[0041] Specifically, for the first type of enterprises, ie, those whose electricity consumption patterns on holidays are significantly different from those on weekdays, it is necessary to pay special attention to their historical electricity consumption data on holidays for prediction in order to improve the accuracy of the prediction.

[0042] For example, a time series-based approach, such as a long short-term memory (LSTM) network, is used to model historical holiday load data. The forecasting model utilizes only historical holiday data, ignoring weekday data, to ensure it accurately reflects holiday electricity usage patterns. After model training, a rolling forecast is performed using the latest holiday data points to obtain load forecasts for future holidays.

[0043] And, step S230, for the second enterprise whose correlation is greater than or equal to the preset threshold, predicting the future holiday load data of the second enterprise based on the historical load data of the second enterprise.

[0044] The historical load data of the second enterprise includes workday load data and holiday load data, and the future holiday load data of the first enterprise and the future holiday load data of the second enterprise are used to indicate the future holiday load forecast results of the enterprises in the target area.

[0045] Specifically, for the second enterprise, its electricity consumption patterns on weekdays and holidays are relatively consistent, so the forecasting process can be simplified and conventional methods can be used for forecasting without special consideration of holiday factors.

[0046] For example, forecasts for this type of enterprise can be made using simpler time series forecasting models, such as exponential smoothing, seasonal decomposition models, or even linear regression models. The training dataset for the forecasting model includes mixed load data from weekdays and holidays to reflect the enterprise's daily electricity usage patterns. Forecast results are based on all historical data, not just holiday data.

[0047] Specifically, by analyzing the correlation between weekday and holiday load data, enterprises' future holiday power load can be more accurately predicted. For enterprises with low correlation, historical holiday load data and characteristic data are used for prediction. For enterprises with high correlation, historical load data from both weekdays and holidays is combined for prediction, effectively improving the accuracy and reliability of predictions. By analyzing the correlation between enterprise power loads on weekdays and holidays, refined management of future holiday load forecasts for enterprises within a target area is achieved. Enterprises are divided into two categories. For the first category, a prediction model specifically tailored to holiday data is used, fully accounting for changes in enterprise power consumption patterns during holidays, thereby improving load forecast accuracy. For the second category, historical load data is directly used for prediction, eliminating the need for additional holiday data collection and analysis, reducing the workload of data collection and preprocessing. While ensuring prediction accuracy, this invention allows more resources and computing power to be concentrated on analyzing the first category of enterprises, which are significantly affected by holidays, achieving optimal resource allocation. This effectively addresses the problem of inaccurate load forecasts caused by significant differences in enterprise power consumption during holidays and weekdays. Through refined analysis and prediction strategies, prediction efficiency is improved while accuracy is guaranteed, providing strong support for power system scheduling and power market operations.

[0048] In this embodiment, by analyzing the correlation between weekday and holiday load data, an enterprise's future holiday power load can be more accurately predicted. For enterprises with low correlation, historical holiday load data and characteristic data are used for prediction. For enterprises with high correlation, a combination of historical load data from both weekdays and holidays is used for prediction, effectively improving the accuracy and reliability of the prediction. By analyzing the correlation between an enterprise's weekday and holiday power load, refined management of future holiday load forecasts for enterprises within a target area is achieved. Enterprises are divided into two categories, fully accounting for changes in their electricity usage patterns during holidays, thereby improving the accuracy of load forecasts. For the second type of enterprise, no additional holiday data collection and analysis is required; historical load data can be directly used for prediction, reducing the workload of data collection and preprocessing. While ensuring prediction accuracy, this invention can focus more resources and computing power on analyzing the first type of enterprises, which are significantly affected by holidays, thereby achieving optimal resource allocation. This effectively addresses the problem of inaccurate load forecasts caused by significant differences in enterprise holiday and weekday power usage. Through refined analysis and prediction strategies, prediction efficiency is improved while accuracy is guaranteed, providing strong support for power system scheduling and power market operations.

[0049] In one embodiment, Figure 3As shown, step S210 determines the correlation between the working day load data and holiday load data of each enterprise based on the working day load data and holiday load data of each enterprise. It includes: steps S300-S330:

[0050] Step S300: Obtain a first time series of workday load data and a second time series of holiday load data of a target enterprise.

[0051] The first time series and the second time series both include a plurality of load data sampled at a preset frequency, and the target enterprise is any one of the enterprises.

[0052] Specifically, electricity load data for the target enterprise is collected for two specific time periods: weekdays and holidays. Both the first time series of weekday load data and the second time series of holiday load data must be sampled at a preset frequency (for example, every half hour) to ensure that the collected data accurately reflects the enterprise's electricity consumption patterns on different days.

[0053] For example, the target enterprise's electricity load data for weekdays and holidays over a specific period of time in the past (e.g., the past year) is extracted from the power company's historical database. During the extraction process, the frequency consistency of the data is ensured, for example, all data is converted into load records every half hour. For missing load data, interpolation methods can be used to fill in the gaps to ensure the continuity of the time series. For example, the first time series can be expressed as: represents the load sequence of the i-th working day, where l i (t) represents the load value of the t-th sampling interval on the i-th day, and T is the number of load sequences in a day. The second time series can be expressed as: represents the load sequence of the jth holiday, where l j (t) represents the load value at the t-th sampling interval on the j-th day, and T is the number of load sequences in 1 day.

[0054] Step S310: construct a sample point set according to the first time series and the second time series.

[0055] The sample point set includes a plurality of sample points, and each sample point includes load data of a first time series and load data of a second time series corresponding to a time.

[0056] Specifically, the load data obtained from the first step is converted into a set of sample points. Each sample point contains a weekday load data point and a holiday load data point at the corresponding time point. This allows the relationship between the two sets of data to be visualized, facilitating subsequent analysis.

[0057] For example, the two sets of time series data are paired to ensure that the time corresponding to each sample point is consistent. For example, if the load data is collected every half hour, then each point in the sample point set should contain load data for both weekdays and holidays within that time period.

[0058] Step S320: determining the distribution of the plurality of sample points in the sample point set in a preset coordinate system.

[0059] Among them, the horizontal axis of the preset coordinate system is the workday load data, and the vertical axis of the preset coordinate system is the holiday load data.

[0060] Specifically, the data points in the sample set are displayed in a two-dimensional coordinate system, with the horizontal axis representing weekday load data and the vertical axis representing holiday load data. By observing the distribution of the sample points in the coordinate system, we can gain a preliminary understanding of the correlations and differences between the load data.

[0061] For example, a scatter plot is drawn to visually display the distribution of the sample point set, which helps to discover potential relationships between load data, such as linear, nonlinear or clustering characteristics.

[0062] Step S330 : determining the correlation between the target enterprise's workday load data and holiday load data based on the distribution of the plurality of sample points.

[0063] Specifically, mathematical methods are used to quantitatively assess the correlation between weekday and holiday load data, providing a basis for subsequent load forecasting. A higher correlation indicates a stronger connection between weekday and holiday load data; conversely, a lower correlation indicates relative independence between the two.

[0064] For example, the Maximum Information Coefficient (MIC) is used to calculate the nonlinear correlation between two time series. In Python, this calculation can be performed using the MIC algorithm in the minepy library. Calculating the MIC value requires gridding the sample point set, finding the maximum mutual information value, and normalizing it. Ultimately, a value between 0 and 1 is obtained to quantify the degree of correlation between the two time series. Based on the MIC value, the strength of the correlation between weekday and holiday load data can be determined, providing guidance for the formulation of the next step of the forecasting strategy.

[0065] For example, the calculation method of correlation, taking MIC as an example, is as follows: for a binary sample point set D(x,y)∈R 2, x is the first time series, which can be expressed as x1, x2, ..., xm, y is the second time series, which can be expressed as y1, y2, ..., yn, and the grid is divided into n columns and m rows. For the divided grid G, calculate the probability that the sample point in the sample point set falls into each grid cell in G (that is, the number of sample points falling into the grid cell is divided by the total number of sample points), and obtain the probability distribution D|G of the sample point set D on the grid G. Find its maximum mutual information maxI(D|G) and save it as I * [D(x,y)], as shown below:

[0066] I * [D(x,y)]=maxI(D|G)

[0067] The above formula means that for a given dataset D(x,y), we calculate the mutual information I(D|G) under all possible grid partitions G and take the maximum value as I*[D(x,y)]. I(D|G) represents the mutual information calculated under the grid partition G.

[0068] Then the maximum mutual information I * [D(x,y)] is normalized and the maximum mutual information coefficient is obtained as follows:

[0069]

[0070] xy <B(k)

[0071] F(D) MIC =max xy<B(k) {M(D) x,y}

[0072] Where k is the number of sample points, B(k) is a function of the sample size, which represents the constraint on the total number of squares xy divided by the grid G, and must be less than B(n). Generally, B(k)=k 0.6 MIC is a normalized maximum mutual information with a value range of [0,1]. The larger the MIC value between two variables, the stronger the correlation, and the smaller the MIC value, the weaker the correlation. The threshold of the MIC between the first time series and the second time series can be set to 0.5. When the obtained MIC is greater than 0.5, it is considered that the electricity load characteristics of the enterprise are not greatly affected by holidays, and it can be predicted according to the recent electricity load data on weekdays. When the obtained MIC is less than or equal to 0.5, it is considered that the electricity load characteristics of the enterprise are greatly affected by holidays, and it is necessary to conduct targeted analysis on the electricity consumption of the enterprise during holidays.

[0073] In this embodiment, to quantify the correlation between weekday and holiday load data, the present invention employs a time series analysis method. A time series is a series of observations arranged in chronological order. For example, an enterprise's power load data can be collected at a preset frequency, such as hourly, daily, or weekly. By constructing a set of sample points containing both weekday and holiday load data and displaying these sample points in a two-dimensional coordinate system, the relationship between the two types of data can be intuitively visualized. If the sample points exhibit a clear linear or nonlinear trend, it indicates a strong correlation between the weekday and holiday load data; conversely, if the sample points are more randomly distributed, the correlation is weak. This analysis method not only reveals the temporal patterns of power demand but also helps us understand how business operating models affect power consumption, thereby providing a basis for developing more reasonable power forecasting and scheduling strategies. Furthermore, this method can be applied to businesses of all sizes, from large industrial facilities to small commercial units, to effectively assess their power demand characteristics. By carefully analyzing business load data on weekdays and holidays, businesses significantly affected by holidays can be effectively identified, allowing differentiated forecasting strategies to be developed. This not only improves forecasting efficiency but also significantly enhances forecast accuracy, providing a solid foundation for efficient power system operation and power market optimization.

[0074] In one embodiment, Figure 4 As shown, step S330 determines the correlation between the target enterprise's workday load data and holiday load data based on the distribution of multiple sample points. It includes: steps S400-S430:

[0075] Step S400: Divide the preset coordinate system into a plurality of grid units.

[0076] Each grid unit corresponds to a value interval of the load data on working days and a value interval of the load data on holidays.

[0077] Specifically, when analyzing the correlation between weekday and holiday load data, dividing the coordinate system into multiple grid cells allows for a more detailed study of the mutual influence and distribution characteristics between the two dimensions. By setting the value range, each grid cell represents a specific combination of weekday and holiday load data.

[0078] For example, first, determine the value intervals for the horizontal axis (weekday load data) and the vertical axis (holiday load data). These intervals can be equally spaced or flexibly adjusted based on the density of the data distribution. For example, if the ranges of the weekday and holiday load data are [0, 1000] and [0.1500], respectively, the horizontal and vertical axes can be divided into 10 intervals, that is, each interval is [0, 100] and [0, 150]. Then, use the above function to divide the coordinate system into 100 grid cells.

[0079] Step S410 , counting the number of sample points that fall into each grid unit among the plurality of sample points.

[0080] Specifically, by counting the number of sample points in each grid cell, the distribution frequency of different load data combinations in the sample point set can be quantified, thereby providing basic data for the calculation of mutual information.

[0081] For example, based on the divided grid, all sample points are traversed and classified into corresponding grid cells based on the weekday and holiday load data values of each sample point. The statistical results will show the number of sample points in each grid cell, reflecting the combined frequency of weekday and holiday load data.

[0082] Step S420 : determining the probability of a sample point falling into each grid unit according to the number of sample points in each grid unit.

[0083] Specifically, the calculation of probability helps to quantify the distribution of sample points in each grid cell and provide the necessary probability distribution data for the subsequent calculation of mutual information.

[0084] Step S430 : determining the mutual information value between the target enterprise's workday load data and holiday load data based on the probability that the sample point falls into each grid cell.

[0085] Specifically, the mutual information value is an information theory metric that measures the degree of mutual dependence between two random variables. Here, it is used to evaluate the correlation between weekday load data and holiday load data. A larger mutual information value indicates a stronger dependence between the variables.

[0086] For example, based on the known probability distribution of grid cells, the mutual information value formula in information theory is used to calculate the mutual information value between the weekday load data and the holiday load data. The mutual information value formula is:

[0087]

[0088] Here, I(X;Y) represents the mutual information between two discrete random variables X and Y. X consists of multiple sample data with values x1, x2, …, xm, and Y consists of multiple sample data with values y1, y2, …, yn. p(xi,yj) is the joint probability distribution, which is the probability that X takes the value xi and Y takes the value yj. A higher mutual information value indicates a stronger correlation between the first and second time series.

[0089] Among them, the mutual information value is used to indicate the correlation between the working day load data and the holiday load data of the target enterprise. Mutual information is a measure of the degree of dependence between two random variables. It reflects the degree to which the information uncertainty of the other variable is reduced when one of the variables is known. In the present invention, by dividing the coordinate system into multiple grid units and calculating the probability of the occurrence of sample points in each grid unit, the joint probability distribution of the working day load data and the holiday load data can be constructed. Then, the mutual information value is calculated using this distribution. The larger the mutual information value, the stronger the correlation between the two data. Compared with the simple correlation coefficient calculation, this method can capture the nonlinear relationship between the data in more detail, and is particularly suitable for processing enterprises whose working day and holiday load data are unevenly distributed or have complex patterns.

[0090] In this embodiment, the division of grid units makes the calculation of mutual information more refined, which can capture the mutual dependence of load data in different intervals and provide a more refined analysis angle. By calculating the probability distribution of sample points within the grid unit, not only can the distribution of load data be intuitively reflected, but also a solid foundation is provided for the accurate calculation of mutual information values. Using mutual information values as a measure of the correlation between workday and holiday load data, compared with the traditional correlation coefficient, mutual information values can more accurately reflect nonlinear relationships, thereby improving the adaptability and prediction accuracy of the load forecasting model. Based on the analysis results of mutual information values, enterprises that are more affected by holidays and enterprises that are less affected can be distinguished, so as to formulate different forecasting strategies for each type of enterprise, reduce unnecessary data analysis workload, and ensure forecast accuracy. Through grid analysis and mutual information value calculation, the correlation between workday and holiday load data is effectively identified, providing key technical support for accurately predicting holiday loads, greatly optimizing the operating efficiency of the power system and the formulation process of market operation strategies.

[0091] In one embodiment, Figure 5 As shown, step S220, for a first enterprise whose correlation is less than a preset threshold, predicting the first enterprise's future holiday load data based on the first enterprise's historical holiday load data. It includes: steps S500-S530:

[0092] Step S500: collecting characteristic data of historical holidays of the first enterprise.

[0093] Among them, characteristic data include environmental factors, business operation factors, and holiday types.

[0094] Specifically, collecting characteristic data from historical holidays forms the basis for predicting future holiday electricity load. This characteristic data includes not only environmental factors (such as weather conditions and temperature), business operational factors (such as production plans and employee numbers), but also the type of holiday (such as Spring Festival and National Day), all of which collectively influence business electricity demand.

[0095] Exemplarily, data collection can be accomplished through a variety of channels, including but not limited to: Environmental factors: Obtain historical holiday weather data, such as temperature, humidity, wind speed, etc., from weather stations or public meteorological data sources. Enterprise operational factors: Work with enterprises to collect operational data such as their production plans, employee vacation arrangements, special events or promotions. Holiday type: Based on the Chinese statutory holiday calendar, determine the type of each holiday, the duration of the holiday, and its connection with working days. After data collection, preprocessing is required, such as data cleaning, missing value filling and outlier processing, as well as possible data standardization, to ensure the accuracy and stability of model training.

[0096] Step S510: construct a training sample set based on the feature data and historical holiday load data.

[0097] The training sample set includes historical holiday load data corresponding to different feature data.

[0098] Specifically, the training sample set is constructed by pairing the collected feature data with historical holiday load data to form input-output pairs for training the neural network. Each sample point contains a set of feature data and corresponding holiday load data, which the model uses to learn the correlation between features and load.

[0099] For example, use the Python pandas library or a similar tool to merge the feature data and load data into a data frame, ensuring that each feature data point is temporally aligned with its corresponding holiday load data point. The data frame is then split into training and validation sets to assess the model's generalization capabilities. When constructing the training sample set, a sliding window approach can be used to generate a continuous sequence of samples, facilitating the training of the time series forecasting model.

[0100] Step S520: training a preset neural network model using a training sample set, so that the neural network model learns the relationship between the feature data and the historical holiday load data.

[0101] Specifically, the neural network model is trained by using a training sample set to learn the complex relationship between feature data and historical holiday load data. This process enables the model to predict future holiday electricity load.

[0102] For example, deep learning frameworks such as TensorFlow or PyTorch are used to build and train neural network models. Models of choice include recurrent neural networks (RNNs), long short-term memory networks (LSTMs), and gated recurrent units (GRUs), which are good at processing time series data. During training, a loss function (such as mean squared error (MSE)) is used to quantify the prediction error, and the model parameters are adjusted through a backpropagation algorithm to minimize the prediction error.

[0103] Step S530 , inputting the feature data of the first enterprise's future holidays into the trained neural network model to obtain the first enterprise's future holiday load data.

[0104] Specifically, after the neural network model is trained and verified to be effective, it can be tested using characteristic data of future holidays to predict the electricity load of future holidays.

[0105] For example, characteristic data for future holidays (including weather forecasts, production plans, etc.) is collected and processed using the same data preprocessing methods used during training. This processed characteristic data is then input into the trained neural network model, which outputs predicted holiday load data. To improve forecast accuracy, a rolling forecast or dynamic adjustment of model inputs can be used. As future holidays approach, the forecast is continuously updated with the latest characteristic data to reflect the impact of short-term changes.

[0106] In this embodiment, for the first category of enterprises with low correlation between weekday and holiday load data, the present invention uses a deep learning neural network model to perform forecasting. Neural network models can automatically learn complex nonlinear relationships from large amounts of historical data, which is particularly important when dealing with highly uncertain problems such as power load forecasting. Feature data selection, such as environmental factors (weather, temperature), business operational factors (production plans, equipment status), and holiday types (such as Spring Festival and National Day), is designed to enable the model to comprehensively consider the various external conditions that influence power demand. By constructing a training sample set containing this feature data and corresponding historical holiday load data, the neural network model gradually grasps the inherent connections between these data during the training process. The advantage of this approach is that even when the correlation between weekday and holiday load data is low, it can predict future power demand by learning from the specific patterns of historical holidays. This is particularly useful for enterprises that are significantly affected by seasonal activities and special events. The predictive capabilities of neural networks can more accurately estimate holiday power demand, thereby optimizing power supply plans and ensuring stable operation of the power system.

[0107] In one embodiment, Figure 6 As shown, step S230, for a second enterprise whose correlation is greater than or equal to a preset threshold, predict the future holiday load data of the second enterprise based on the historical load data of the second enterprise. It includes: steps S600-S610:

[0108] Step S600: determining a change trend of the load data of the second enterprise based on the historical load data of the second enterprise.

[0109] Specifically, identifying changing trends is a key step in forecasting future load data. It helps us understand how electricity consumption changes over time and provides a basis for forecasting future holiday load data. In this step, we extract long-term and short-term trend features from historical load data.

[0110] For example, a time series analysis method, such as a moving average, exponential smoothing, or an autoregressive integrated moving average (ARIMA) model, is used to determine a changing trend in the load data of the second enterprise. For example, calculating a moving average of the load data for weekdays and holidays over the past 30 days can reveal a recent fluctuation trend in the load data.

[0111] Step S610 , determining the future holiday load data of the second enterprise based on the current load data of the second enterprise and the change trend of the load data of the second enterprise.

[0112] Specifically, based on the current load data and its changing trends, we can estimate the load data for future holidays. This step combines the current data status with historical trends to predict future conditions.

[0113] For example, analyze the current load data: First, analyze the current load data of the second enterprise to see if it conforms to the recent trend of change or whether there are abnormal fluctuations. In Python, this can be done through the data analysis function of pandas. Based on the determined trend of change, select a suitable time series forecasting model, such as seasonal ARIMA (SARIMA) or long short-term memory network (LSTM). SARIMA is suitable for data with obvious seasonality, while LSTM can handle more complex nonlinear relationships. Build and train these models in statsmodels or keras libraries. Use the constructed forecasting model, combined with the current load data and known trend of change, to predict the load data for future holidays.

[0114] In this embodiment, for the second category of businesses, whose weekday and holiday load data have a high correlation, the present invention utilizes a forecasting method based on trend analysis. The core of this method is to identify patterns in how a business's electricity demand changes over time, including long-term trends, seasonal fluctuations, and cyclical variations. For example, a restaurant business may have higher electricity demand on weekends and holiday evenings. This pattern can be discovered by analyzing the time series of historical load data. Once the trend is determined, current load data can be combined to predict future electricity demand, particularly during holidays. This simple and direct method is suitable for businesses with relatively stable electricity demand patterns. It enables rapid forecasting and reduces the computational cost of the forecasting model. Furthermore, by incorporating current load data, the forecast results can promptly reflect the latest operating conditions, improving the timeliness and accuracy of the forecast. This method provides an efficient and accurate means of forecasting electricity load, particularly for businesses less affected by weekday and holiday operating patterns. This helps power companies rationally plan power resources and meet the electricity needs of various businesses.

[0115] In one embodiment, Figure 7 As shown, the method further includes: steps S700-S720:

[0116] Step S700 : determining the proportion of the historical holiday load data of the first enterprise and the historical holiday load data of the second enterprise in the historical holiday load data of all enterprises in the target area.

[0117] Specifically, this step aims to quantify the contribution of Enterprise 1 and Enterprise 2 to the holiday load data for the entire target area, thereby assessing their impact on the overall load forecast. Determining these proportions provides a basis for subsequent weight calculations, facilitating fair and accurate forecasting.

[0118] For example, the respective proportions can be calculated by dividing the sum of the historical holiday load data of the first and second enterprises by the sum of the historical holiday load data of all enterprises in the target area. First, the historical holiday load data of all enterprises in the target area is extracted from the database to calculate the total load. Then, the sum of the load data of the first and second enterprises is calculated separately, and finally the respective proportions are calculated. For example, if the sum of the historical holiday load data of all enterprises in the target area is 10,000 kWh, and the sum of the historical holiday load data of the first and second enterprises is 3,000 and 1,500 kWh respectively, then their respective proportions are 30% and 15%, respectively.

[0119] Step S710 , determining weights corresponding to the first enterprise and the second enterprise respectively according to the proportion of the historical holiday load data of the first enterprise and the historical holiday load data of the second enterprise in the historical holiday load data of all enterprises in the target area.

[0120] Specifically, the weight is determined by the ratio calculated above, which reflects the importance of the enterprise in holiday load forecasting. Enterprises with higher weights contribute more to the overall forecast results, which means that they need to be given more attention and resources during the forecast.

[0121] For example, the weight setting can be determined based on the proportion of the company's historical holiday load data. A simple implementation method is to use the proportion directly as the weight, or to make appropriate adjustments to the proportion (such as normalization, quadratic amplification, etc.) to highlight the role of important companies. For example, if the first company accounts for 30% and the second company accounts for 15%, their initial weights can be set to 30% and 15%. Taking into account that the first company is more affected by holidays, its weight can be appropriately increased to 40% to emphasize its uniqueness in holiday load forecasting. Or the sum of the weights of the first company and the second company does not have to be 1. The weight of the first company can also be designed to be 1.1 and the weight of the second company to be 0.9 according to actual conditions. The weight can be used to determine which company's data has a greater contribution and influence on the overall situation.

[0122] Step S720 , weightedly summing the future holiday load data of the first enterprise and the future holiday load data of the second enterprise according to the corresponding weights of the first enterprise and the second enterprise, to determine the future holiday load forecast results of the enterprises in the target area.

[0123] Specifically, the weighted summation combines the future holiday load forecasts for the first and second enterprises into an overall forecast for the target area. The weights reflect the contribution of each enterprise to the forecast. This integration ensures that the forecast considers enterprises significantly affected by holidays while also not neglecting those less affected.

[0124] In this embodiment, in order to comprehensively consider the importance of different enterprises in regional electricity demand, the concept of weight is introduced. The weight is determined based on the proportion of each enterprise in the historical holiday load data, which means that those enterprises with a high proportion of electricity consumption during holidays will be given a greater weight. For example, if a large shopping mall consumes much more electricity on holidays than on weekdays and accounts for a large proportion of the total regional electricity demand, then its weight will increase accordingly. By weighted summation, the forecast results can more accurately reflect the overall electricity demand of the region, rather than simply averaging the forecast values of all enterprises. This method ensures the rationality of the forecast results, especially during periods of large fluctuations in electricity demand such as holidays. It can more accurately predict the total electricity demand of the region, avoid excessive conservatism or excessive optimism in power scheduling, and help power companies make more scientific power supply decisions.

[0125] In summary, this application realizes the refined management of power load forecasting by comprehensively considering the power demand characteristics of each enterprise and its proportion in the total regional demand, and provides strong support for the optimized operation of the power system. By analyzing the correlation between workday and holiday load data, the power load of the enterprise during future holidays can be predicted more accurately. For enterprises with low correlation, historical holiday load data and characteristic data are used for prediction. For enterprises with high correlation, historical load data of workdays and holidays are combined for prediction, which effectively improves the accuracy and reliability of the prediction. By taking into account the contribution of different enterprises to the total regional load through weighted summation, the prediction results are closer to reality.

[0126] Through the description of the above implementation methods, those skilled in the art can clearly understand that the method according to the above embodiment can be implemented by means of software plus the necessary general hardware platform, and of course it can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for enabling a terminal device (which can be a mobile phone, computer, server, or network device, etc.) to execute the methods of each embodiment of the present application.

[0127] This embodiment also provides a power load prediction device, which is used to implement the above-mentioned embodiments and preferred embodiments. Details already described will not be repeated here. As used below, the term "module" may refer to a combination of software and / or hardware that implements a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, implementation using hardware, or a combination of software and hardware, is also possible and contemplated.

[0128] Figure 8 This is a structural block diagram of an optional power load prediction device according to an embodiment of the present application. Figure 8 Shown, including:

[0129] The data acquisition module 801 is used to acquire the workday load data and holiday load data of each enterprise in the target area.

[0130] The correlation determination module 802 is used to determine the correlation between the working day load data and the holiday load data of each enterprise based on the working day load data and the holiday load data of each enterprise.

[0131] The first prediction module 803 is configured to predict, for a first enterprise whose correlation is less than a preset threshold, future holiday load data of the first enterprise based on the first enterprise's historical holiday load data.

[0132] The second prediction module 804 is used to predict the future holiday load data of the second enterprise based on the historical load data of the second enterprise for the second enterprise whose correlation is greater than or equal to the preset threshold, wherein the historical load data of the second enterprise includes weekday load data and holiday load data, and the future holiday load data of the first enterprise and the future holiday load data of the second enterprise are used to indicate the future holiday load prediction results of the enterprises in the target area.

[0133] By analyzing the correlation between weekday and holiday load data, the above-described device can more accurately predict an enterprise's future holiday power load. For enterprises with low correlation, historical holiday load data and characteristic data are used for prediction. For enterprises with high correlation, a combination of historical load data from both weekdays and holidays is used for prediction, effectively improving the accuracy and reliability of the prediction. By analyzing the correlation between an enterprise's weekday and holiday power load, refined management of future holiday load forecasts for enterprises within a target area is achieved. Enterprises are divided into two categories. For the first category, a prediction model specifically tailored to holiday data is used, fully accounting for changes in enterprise power usage patterns during holidays, thereby improving load forecast accuracy. For the second category, no additional holiday data collection and analysis is required; historical load data is directly used for prediction, reducing the workload of data collection and preprocessing. While ensuring prediction accuracy, the present invention can focus more resources and computing power on analyzing the first category of enterprises significantly affected by holidays, achieving optimal resource allocation. This effectively addresses the problem of inaccurate load forecasts caused by significant differences in enterprise power usage during holidays and weekdays. Through refined analysis and prediction strategies, prediction efficiency is improved while accuracy is guaranteed, providing strong support for power system scheduling and power market operations.

[0134] In an exemplary embodiment, the correlation determination module 802 is further used to: obtain a first time series of the target enterprise's workday load data and a second time series of the holiday load data, wherein the first time series and the second time series each include a plurality of load data sampled at a preset frequency, and the target enterprise is any one of the enterprises. A sample point set is constructed based on the first time series and the second time series, wherein the sample point set includes a plurality of sample points, and each sample point includes the load data of the first time series and the load data of the second time series corresponding to the time. The distribution of the plurality of sample points in the sample point set in a preset coordinate system is determined, wherein the horizontal axis of the preset coordinate system is the workday load data, and the vertical axis of the preset coordinate system is the holiday load data. The correlation between the workday load data and the holiday load data of the target enterprise is determined based on the distribution of the plurality of sample points.

[0135] In an exemplary embodiment, the correlation determination module 802 is further configured to: divide a preset coordinate system into a plurality of grid cells, wherein each grid cell corresponds to a value interval of the workday load data and a value interval of the holiday load data; count the number of sample points that fall into each grid cell among the plurality of sample points; determine the probability of a sample point falling into each grid cell based on the number of sample points in each grid cell; and determine the mutual information value between the workday load data and the holiday load data of the target enterprise based on the probability of the sample point falling into each grid cell, wherein the mutual information value is used to indicate the correlation between the workday load data and the holiday load data of the target enterprise.

[0136] In an exemplary embodiment, the first prediction module 803 is further configured to collect characteristic data of historical holidays for the first enterprise, where the characteristic data includes environmental factors, enterprise operating factors, and holiday types. A training sample set is constructed based on the characteristic data and historical holiday load data, where the training sample set includes historical holiday load data corresponding to different characteristic data. A preset neural network model is trained using the training sample set to enable the neural network model to learn the relationship between the characteristic data and the historical holiday load data. The characteristic data of future holidays for the first enterprise is input into the trained neural network model to obtain the future holiday load data for the first enterprise.

[0137] In an exemplary embodiment, the second prediction module 804 is further configured to determine a change trend of the load data of the second enterprise based on the historical load data of the second enterprise, and to determine future holiday load data of the second enterprise based on the current load data of the second enterprise and the change trend of the load data of the second enterprise.

[0138] In an exemplary embodiment, the apparatus further comprises:

[0139] The proportion determination module is used to respectively determine the proportion of the historical holiday load data of the first enterprise and the historical holiday load data of the second enterprise in the historical holiday load data of all enterprises in the target area.

[0140] The weight determination module is used to determine the weights corresponding to the first enterprise and the second enterprise respectively according to the proportion of the historical holiday load data of the first enterprise and the historical holiday load data of the second enterprise in the historical holiday load data of all enterprises in the target area.

[0141] The load determination module is used to weightedly sum the future holiday load data of the first enterprise and the future holiday load data of the second enterprise according to their corresponding weights to determine the future holiday load forecast results of the enterprises in the target area.

[0142] An embodiment of the present application further provides a storage medium, which includes a stored program, wherein the program executes any of the above methods when it is run.

[0143] Optionally, in this embodiment, the storage medium may be configured to store program codes for executing the following steps:

[0144] S1, obtain the working day load data and holiday load data of each enterprise in the target area.

[0145] S2, determining the correlation between the working day load data and the holiday load data of each enterprise according to the working day load data and the holiday load data of each enterprise.

[0146] S3: For the first enterprise whose correlation is less than a preset threshold, predict the future holiday load data of the first enterprise based on the historical holiday load data of the first enterprise.

[0147] And, S4, for the second enterprise whose correlation is greater than or equal to a preset threshold, predict the future holiday load data of the second enterprise based on the historical load data of the second enterprise, wherein the historical load data of the second enterprise includes weekday load data and holiday load data, and the future holiday load data of the first enterprise and the future holiday load data of the second enterprise are used to indicate the future holiday load prediction results of the enterprises in the target area.

[0148] An embodiment of the present application further provides an electronic device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor is configured to run the computer program to execute the steps in any one of the above method embodiments.

[0149] Optionally, the electronic device may further include a transmission device and an input / output device, wherein the transmission device is connected to the processor, and the input / output device is connected to the processor.

[0150] Optionally, in this embodiment, the processor may be configured to execute the following steps through a computer program:

[0151] S1, obtain the working day load data and holiday load data of each enterprise in the target area.

[0152] S2, determining the correlation between the working day load data and the holiday load data of each enterprise according to the working day load data and the holiday load data of each enterprise.

[0153] S3: For the first enterprise whose correlation is less than a preset threshold, predict the future holiday load data of the first enterprise based on the historical holiday load data of the first enterprise.

[0154] And, S4, for the second enterprise whose correlation is greater than or equal to a preset threshold, predict the future holiday load data of the second enterprise based on the historical load data of the second enterprise, wherein the historical load data of the second enterprise includes weekday load data and holiday load data, and the future holiday load data of the first enterprise and the future holiday load data of the second enterprise are used to indicate the future holiday load prediction results of the enterprises in the target area.

[0155] Optionally, in this embodiment, the above-mentioned storage medium may include but is not limited to: a USB flash drive, a read-only memory (ROM), a random access memory (RAM), a mobile hard disk, a magnetic disk or an optical disk, and other media that can store program codes.

[0156] An embodiment of the present application further provides a computer program product, including a non-volatile computer-readable storage medium, wherein the non-volatile computer-readable storage medium stores the computer program product, and when the computer program is executed by a processor, the steps of the method in each embodiment of the present application are implemented.

[0157] Optionally, in this embodiment, the computer program may be configured to implement the following steps when executed by a processor:

[0158] S1, obtain the working day load data and holiday load data of each enterprise in the target area.

[0159] S2, determining the correlation between the working day load data and the holiday load data of each enterprise according to the working day load data and the holiday load data of each enterprise.

[0160] S3: For the first enterprise whose correlation is less than a preset threshold, predict the future holiday load data of the first enterprise based on the historical holiday load data of the first enterprise.

[0161] And, S4, for the second enterprise whose correlation is greater than or equal to a preset threshold, predict the future holiday load data of the second enterprise based on the historical load data of the second enterprise, wherein the historical load data of the second enterprise includes weekday load data and holiday load data, and the future holiday load data of the first enterprise and the future holiday load data of the second enterprise are used to indicate the future holiday load prediction results of the enterprises in the target area.

[0162] Optionally, specific examples in this embodiment may refer to the examples described in the above embodiments and optional implementation modes, and this embodiment will not be described in detail here.

[0163] Obviously, those skilled in the art should understand that the modules or steps of the present application described above can be implemented using a general-purpose computing device, they can be concentrated on a single computing device, or distributed on a network composed of multiple computing devices. Alternatively, they can be implemented using program code executable by the computing device, so that they can be stored in a storage device and executed by the computing device. In some cases, the steps shown or described can be performed in a different order than herein, or they can be made into separate integrated circuit modules, or multiple modules or steps can be made into a single integrated circuit module for implementation. Thus, the present application is not limited to any specific combination of hardware and software.

[0164] The above description is merely a preferred embodiment of the present application and is not intended to limit the present application. Various modifications and variations are possible for those skilled in the art. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present application shall be included within the scope of protection of the present application.

Claims

1. A method for predicting power load, characterized in that: The method comprises: Obtain weekday load data and holiday load data for each enterprise in the target area; Determine the correlation between the working day load data and the holiday load data of each enterprise according to the working day load data and the holiday load data of each enterprise; For the first enterprise whose correlation is less than a preset threshold, predicting the future holiday load data of the first enterprise based on the historical holiday load data of the first enterprise; And, for the second enterprise whose correlation is greater than or equal to the preset threshold, the future holiday load data of the second enterprise is predicted based on the historical load data of the second enterprise, wherein the historical load data of the second enterprise includes weekday load data and holiday load data, and the future holiday load data of the first enterprise and the future holiday load data of the second enterprise are used to indicate the future holiday load prediction results of the enterprises in the target area.

2. The method for predicting power load according to claim 1, characterized in that: Determining the correlation between the working day load data and the holiday load data of each enterprise based on the working day load data and the holiday load data of each enterprise includes: Obtaining a first time series of workday load data and a second time series of holiday load data of a target enterprise, wherein the first time series and the second time series each include a plurality of load data sampled at a preset frequency, and the target enterprise is any one of the enterprises; Constructing a sample point set according to the first time series and the second time series, wherein the sample point set includes a plurality of sample points, and each sample point includes load data of the first time series and load data of the second time series corresponding to a time; Determining the distribution of multiple sample points in the sample point set in a preset coordinate system, wherein the horizontal axis of the preset coordinate system is the workday load data, and the vertical axis of the preset coordinate system is the holiday load data; The correlation between the working day load data and the holiday load data of the target enterprise is determined according to the distribution of the multiple sample points.

3. The method for predicting power load according to claim 2, characterized in that: The determining of the correlation between the working day load data and the holiday load data of the target enterprise according to the distribution of the plurality of sample points includes: Dividing the preset coordinate system into a plurality of grid units, wherein each grid unit corresponds to a value interval of the load data on a working day and a value interval of the load data on a holiday; Counting the number of sample points that fall into each grid unit respectively among the multiple sample points; Determining, according to the number of sample points in each grid unit, the probability that the sample point falls into each grid unit; According to the probability that the sample point falls into each of the grid cells, the mutual information value between the working day load data and the holiday load data of the target enterprise is determined, wherein the mutual information value is used to indicate the correlation between the working day load data and the holiday load data of the target enterprise.

4. The method for predicting power load according to any one of claims 1 to 3, characterized in that: For the first enterprise whose correlation is less than a preset threshold, predicting the future holiday load data of the first enterprise based on the historical holiday load data of the first enterprise includes: Collecting characteristic data of historical holidays of the first enterprise, wherein the characteristic data includes environmental factors, enterprise operation factors, and holiday types; Constructing a training sample set based on the characteristic data and the historical holiday load data, wherein the training sample set includes historical holiday load data corresponding to different characteristic data; Using the training sample set to train a preset neural network model, so that the neural network model learns the relationship between the feature data and the historical holiday load data; The feature data of the future holidays of the first enterprise are input into the trained neural network model to obtain the future holiday load data of the first enterprise.

5. The method for predicting power load according to any one of claims 1 to 3, characterized in that: For the second enterprise whose correlation is greater than or equal to the preset threshold, predicting the future holiday load data of the second enterprise based on the historical load data of the second enterprise includes: determining a change trend of the load data of the second enterprise based on the historical load data of the second enterprise; The future holiday load data of the second enterprise is determined according to the current load data of the second enterprise and the change trend of the load data of the second enterprise.

6. The method for predicting power load according to any one of claims 1 to 3, characterized in that: The method further comprises: respectively determining the proportions of the historical holiday load data of the first enterprise and the historical holiday load data of the second enterprise in the historical holiday load data of all enterprises in the target area; Determine weights corresponding to the first enterprise and the second enterprise respectively according to the proportions of the historical holiday load data of the first enterprise and the historical holiday load data of the second enterprise in the historical holiday load data of all enterprises in the target area; The future holiday load data of the first enterprise and the future holiday load data of the second enterprise are weightedly summed according to the corresponding weights of the first enterprise and the second enterprise to determine the future holiday load forecast result of the enterprises in the target area.

7. A power load prediction device, characterized in that: The device comprises: The data acquisition module is used to obtain the working day load data and holiday load data of each enterprise in the target area; a correlation determination module, configured to determine the correlation between the working day load data and the holiday load data of each enterprise based on the working day load data and the holiday load data of each enterprise; a first prediction module, configured to predict, for a first enterprise whose correlation is less than a preset threshold, future holiday load data of the first enterprise based on the historical holiday load data of the first enterprise; The second prediction module is used to predict the future holiday load data of the second enterprise whose correlation is greater than or equal to the preset threshold based on the historical load data of the second enterprise, wherein the historical load data of the second enterprise includes weekday load data and holiday load data, and the future holiday load data of the first enterprise and the future holiday load data of the second enterprise are used to indicate the future holiday load prediction results of the enterprises in the target area.

8. A computer-readable storage medium, characterized in that The computer-readable storage medium includes a stored program, wherein the program executes the method according to any one of claims 1 to 6 when executed.

9. An electronic device comprising a memory and a processor, characterized in that: A computer program is stored in the memory, and the processor is configured to execute the method according to any one of claims 1 to 6 through the computer program.

10. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

Images