Power grid load high-precision prediction method based on big data analysis and storage medium

Abstract

The application relates to the technical field of data processing, in particular to a power grid load high-precision prediction method based on big data analysis and a storage medium. The method comprises the following steps: collecting meteorological data and power grid load data; determining a period coordination coefficient based on period differences of spectrum graph peak values of the meteorological data and the power grid load data; screening period points based on the period coordination coefficient, and obtaining a periodic prediction value based on the meteorological data and the power grid load data of the period points; constructing a contribution function based on the power grid load data and the meteorological data, and determining a contribution degree based on a partial derivative of the meteorological data and a maximum difference; constructing an optimization function based on the contribution degree to reduce the dimension of the meteorological data; determining an aperiodic prediction value based on the reduced-dimension meteorological data and the power grid load data; and obtaining a prediction value of the power grid load data based on the periodic prediction value and the aperiodic prediction value. The application improves the prediction accuracy of the power grid load.

Description

Technical Field

[0001] This application relates to the field of data processing technology, specifically to a high-precision power grid load prediction method and storage medium based on big data analysis. Background Technology

[0002] Compared to traditional power grids, smart grids utilize advanced infrastructure and information technology to provide smarter, more efficient, sustainable, and reliable power services. Power load forecasting plays an increasingly indispensable role in smart grids. Advance modeling and forecasting of power load allows for the development of frequency regulation plans, improving grid frequency stability. Artificial neural networks are a common forecasting method in power load forecasting, adept at capturing complex nonlinear relationships. However, load access on the load side is constrained by daily weather, temperature, and other external changes, exhibiting a random characteristic. Current technologies do not consider the periodic and non-periodic variations of power load during forecasting, resulting in insufficient handling of weather sensitivity and weak adaptability to extreme weather, thus leading to low accuracy in power grid load forecasting. Summary of the Invention

[0003] To address the technical problem of poor forecast accuracy due to weak weather adaptability, this application provides a high-precision power grid load forecasting method and storage medium based on big data analysis. The specific technical solution adopted is as follows:

[0004] This application proposes a high-precision power grid load forecasting method based on big data analysis, which includes the following steps:

[0005] Collect meteorological data and power grid load data;

[0006] Obtain the spectrum diagrams of historically collected meteorological data and power grid load data, and determine the periodic coordination coefficient of each peak of each type of meteorological data based on the periodic differences of the peak values ​​in the power grid load spectrum diagram and the spectrum diagram of each type of meteorological data.

[0007] Periodic points are selected based on the magnitude of the periodic synergy coefficients of all types of meteorological data; the periodic forecast values ​​of future periodic points are obtained by predicting the changing trend of power grid load data based on the meteorological data and power grid load data of the periodic points.

[0008] A time window is constructed by pre-setting historical time points, and a contribution function is constructed based on power grid load data and meteorological data. The contribution of each type of meteorological data to the power grid load data is determined based on the maximum difference in data values ​​in each type of meteorological data and the partial derivative of each type of meteorological data in the contribution function. An optimization function is constructed based on the difference between the contribution of meteorological data at each time point and the average contribution at each time point to reduce the dimensionality of the meteorological data categories. The meteorological data and power grid load data of the remaining categories after dimensionality reduction in the time window are used to predict non-periodic predicted values ​​for a preset time period.

[0009] The predicted values ​​of the power grid load data for the preset time period are obtained based on the periodic and non-periodic predicted values ​​within the preset time period.

[0010] In the aforementioned scheme, this application performs frequency domain transformation on power grid load data and meteorological data. By extracting the frequency peak values ​​from the power grid load data and meteorological data, the periodic variation characteristics of the data are obtained. Combined with the meteorological data variation characteristics, the periodic pattern of the meteorological impact on power grid load is quantified. Furthermore, this application proposes a prediction mechanism that separates periodic and non-periodic data. Through periodic separation, dynamic weighting of meteorological contributions, and dual-model fusion, the prediction error in meteorologically sensitive scenarios is significantly reduced. This provides a solution that combines accuracy and efficiency for predicting highly meteorologically sensitive power grid loads, and is particularly suitable for power grid scenarios with complex load structures and variable climates.

[0011] In one embodiment, the meteorological data includes temperature, humidity, rainfall, and wind speed.

[0012] In one embodiment, the periodic synergy coefficient is negatively correlated with the periodic difference.

[0013] In one embodiment, the method for selecting periodic points based on the magnitude of all periodic synergy coefficients for each type of meteorological data is as follows:

[0014] Calculate the mean of all periodic coordination coefficients in each type of meteorological data, and use the mean as the power grid periodic load threshold; record the time corresponding to the peak value where the periodic coordination coefficient is greater than or equal to the power grid periodic load threshold as the periodic point.

[0015] In one embodiment, the dependent variable of the contribution function is the power grid load data, and the independent variable is meteorological data for all classes.

[0016] In one embodiment, the contribution is positively correlated with the partial derivative and negatively correlated with the maximum difference in the meteorological data.

[0017] In one embodiment, the expression for the optimization function is:

[0018] , This represents the contribution of the j-th type of meteorological data at time i to the power grid load data. This represents the average contribution at time i. Indicates the length of the time window. Indicates the final quantity. Represents the minimum optimization function. This represents the optimal function value.

[0019] In one embodiment, the average contribution at each moment is the average contribution of all types of meteorological data corresponding to each moment, and the final number is the number of categories after dimensionality reduction of the preset meteorological data categories.

[0020] In one embodiment, the method for obtaining the predicted value of the power grid load data for the preset time period based on the periodic and non-periodic predicted values ​​within the preset time period is as follows:

[0021] If there is a periodic point within the future time window, the predicted value for that moment is obtained by weighting the periodic and non-periodic predicted values; if there is no periodic point within the future time window, the non-periodic predicted value is the predicted value for that moment.

[0022] Secondly, this embodiment also provides a computer-readable storage medium, which includes a stored computer program; wherein, when the computer program is running, it controls the device where the computer-readable storage medium is located to execute the above-described method for high-precision prediction of power grid load based on big data analysis.

[0023] The beneficial effects of this application are as follows:

[0024] This application performs frequency domain transformations on power grid load data and meteorological data. By extracting the frequency peaks from these data, it obtains the periodic variation characteristics of the power grid load and meteorological data. Combined with these meteorological data variation characteristics, it quantifies the periodic patterns of meteorological influence on power grid load. Furthermore, this application proposes a prediction mechanism that separates periodic and non-periodic data. Through periodic separation, dynamic weighting of meteorological contributions, and dual-model fusion, it significantly reduces prediction errors in meteorologically sensitive scenarios. This provides a solution that combines accuracy and efficiency for predicting highly meteorologically sensitive power grid loads, and is particularly suitable for power grid scenarios with complex load structures and variable climates. Attached Figure Description

[0025] To more clearly illustrate the technical solutions and advantages in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0026] Figure 1 This is a flowchart of a high-precision power grid load prediction method based on big data analysis, provided as an embodiment of this application. Detailed Implementation

[0027] To further illustrate the technical means and effects adopted by this application to achieve the intended inventive objective, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of the high-precision power grid load prediction method and storage medium based on big data analysis proposed in this application. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0028] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0029] Example of a high-precision power grid load forecasting method and storage medium based on big data analysis:

[0030] The following description, in conjunction with the accompanying drawings, details the specific scheme of the high-precision power grid load prediction method and storage medium based on big data analysis provided in this application.

[0031] Please see Figure 1 The diagram illustrates a flowchart of a high-precision power grid load prediction method based on big data analysis according to an embodiment of this application. The method includes the following steps:

[0032] Step S001: Collect meteorological data and power grid load data.

[0033] Climate factors such as temperature, humidity, wind speed, sunshine duration, and rainfall can affect normal activities. The proportion of meteorologically sensitive loads in the peak load of the power grid is constantly increasing. Winter and summer are characterized by hot / cold weather with temperatures remaining at relatively high / low levels. However, spring and autumn are transitional seasons with the characteristics of two seasons blending together. At this time, the overall temperature fluctuates more strongly, and the fluctuation pattern of the power load curve under the influence of meteorological conditions becomes more complex. At the same time, the regional power grid load is significantly affected by the industrial structure. The high-energy-consuming industrial structure of large-scale industries and commerce will also affect the accuracy of power load forecasting.

[0034] Meteorological data is collected using data acquisition equipment; in this embodiment, the meteorological data includes temperature, humidity, rainfall, and wind speed.

[0035] During meteorological data acquisition, malfunctions in the acquisition or data recording equipment can lead to omissions in the final acquired meteorological data. These missing data can cause significant errors in subsequent analysis. Therefore, meteorological data is used as input to the Loophole-of-Flight (LOF) algorithm to extract outlier data items and mark their positions as 0.

[0036] For data items where the data record at a given location is 0, missing data items are filled in using a data interpolation algorithm. In this embodiment, B-spline interpolation is used to process the missing data items. Simultaneously, due to the diverse sources of meteorological data during power forecasting, the units of measurement for data from different dimensions are different. To eliminate the interference of different dimensions of data on the power load forecasting results, a normalization algorithm is used to process the meteorological data. In this embodiment, the normalization algorithm used is the maximum-minimum normalization method. In this embodiment, the data collection frequency for power grid load data and meteorological data is 15 minutes.

[0037] Simultaneously collect grid load data and perform the same operations described above on the grid load data.

[0038] At this point, meteorological data and power grid load data were obtained.

[0039] Step S002: Determine the periodic coordination coefficient based on the periodic differences of the peak values ​​in the spectrum diagrams of meteorological data and power grid load data.

[0040] Residential and commercial activities follow regular patterns, and electricity consumption should correspondingly exhibit regular variations. Consequently, the power grid load will also show corresponding periodic variations. Generally, peak grid load periods are concentrated in the morning and evening, while the load is lower at night. Furthermore, weekday grid load is higher than weekend load. Grid load changes exhibit a periodic characteristic.

[0041] A preset number of meteorological and power grid load data points are selected from historical data, and analysis is performed using these selected data. In this embodiment, data from the month preceding the current time are selected.

[0042] Fast Fourier Transform was performed on the selected power grid load data and each type of meteorological data to obtain the power grid load spectrum and the spectrum of each type of meteorological data. The horizontal axis of the spectrum represents frequency, and its reciprocal represents the period.

[0043] Obtain all peak values ​​from the power grid load spectrum and the spectrum of each type of meteorological data. If the periodic patterns of the power grid load and the meteorological data are consistent, then the periods should also be consistent if the frequencies are consistent.

[0044] The periodic coordination coefficient is determined based on the periodic differences in the peak values ​​of the power grid load spectrum and the spectrum of each type of meteorological data.

[0045] The periodic synergy coefficient is negatively correlated with the periodic difference.

[0046] It should be noted that negative correlation means that when one variable increases, the other variable decreases accordingly, and the two variables change in opposite directions. When one variable changes from large to small or from small to large, the other variable also changes from small to large or from large to small. The specific relationship is determined by practical application, and this application does not impose any special restrictions.

[0047] Preferably, in this embodiment, the expression for the periodic coordination coefficient is:

[0048] , This represents the period corresponding to the a-th peak value in the power grid load spectrum diagram. This represents the period corresponding to the a-th peak in the spectrum of meteorological data. This represents the minimum number of peak values ​​in the power grid load spectrum diagram and the meteorological data spectrum diagram. This represents the periodic coordination coefficient of the a-th peak.

[0049] Generally, the changes in power grid load vary with the mode of electricity consumption. The daily power grid load variation should exhibit significant characteristics over 12 hours and 24 hours, and the daily power load variation should also show a certain periodicity within a specific time period. Simultaneously, meteorological factors such as temperature and humidity should also show a certain periodicity. When the periodicity of meteorological factors such as temperature and humidity is consistent with the periodicity of the daily power grid load variation, the corresponding frequencies in the spectral sequence should be consistent, and the periods should also be consistent. Therefore, the greater the difference between the two, the greater the difference in the changing trends of the power grid load data and the meteorological data, and the smaller the periodicity coefficient; conversely, the smaller the difference, the smaller the difference in the changing trends of the power grid load data and the meteorological data, and the larger the periodicity coefficient.

[0050] At this point, the periodic coordination coefficient for each peak value has been obtained.

[0051] It should be noted that positive correlation means that when one variable increases, the other variable also increases, and the two variables change in the same direction. When one variable changes from large to small or from small to large, the other variable also changes from large to small or from small to large. The specific relationship can be a multiplicative relationship, an additive relationship, or an exponential function, which is determined by the actual application. This application does not impose any special restrictions.

[0052] Step S003: Select periodic points based on the periodic coordination coefficient, and obtain periodic forecast values ​​based on meteorological data and power grid load data of the periodic points.

[0053] Because power grid load data exhibits periodic variations, the periodic variation sequence in the power grid load data is predicted to obtain periodic data of the power grid load data.

[0054] The above steps are used to obtain the periodic coordination coefficient for each peak value, calculate the average of all periodic coordination coefficients, and use this average as the grid periodic load threshold. The time corresponding to the peak value with a periodic coordination coefficient greater than or equal to the grid periodic load threshold is recorded as the periodic point. Since there is a periodic coordination coefficient between the grid load data and each type of meteorological data, a load average is calculated for each type of meteorological data when calculating the average. That is, the grid load data and each type of meteorological data will obtain different periodic points, and these periodic points are collectively referred to as periodic points.

[0055] When the coordination coefficient at a given time is greater than the threshold, it indicates that the power grid load data at that time is significantly affected by weather conditions; therefore, the power grid load can be inferred based on the weather data.

[0056] Therefore, meteorological data and power grid load data at the periodic points are used as inputs to the MUSIC algorithm to analyze the changing trend of power grid load data and obtain the predicted power grid load data at future periodic points, which are denoted as periodic predicted values.

[0057] Thus, the periodic prediction values ​​for future cycle points have been obtained.

[0058] Step S004: Construct a contribution function based on power grid load data and meteorological data, and determine the contribution degree based on the partial derivatives and maximum differences of the meteorological data; construct an optimization function based on the contribution degree to reduce the dimensionality of the meteorological data; determine the non-periodic predicted value based on the dimensionality-reduced meteorological data and power grid load data.

[0059] For the non-periodic variation portion of the power grid load data, the characteristics of the non-periodic variation of the power grid load are analyzed in conjunction with meteorological change data.

[0060] When meteorological data, i.e., external conditions, change, the electricity consumption on the user side will also change accordingly. At this time, different meteorological factors contribute differently to the change in grid load. Therefore, a preset amount of grid load data is used for analysis.

[0061] A time window is formed by taking the time of the previous month as the current moment; within the time window, the power grid load data is used as the dependent variable, and each type of meteorological data is used as an independent variable to form a multivariate contribution function.

[0062] For each type of meteorological data at each time point, the partial derivative is calculated. The partial derivative reflects the degree of influence of the independent variable on the dependent variable.

[0063] Therefore, the contribution of each type of meteorological data to the power grid load data is determined based on the partial derivative of each type of meteorological data and the maximum difference in each type of meteorological data.

[0064] The contribution rate is positively correlated with the partial derivative and negatively correlated with the maximum difference in meteorological data.

[0065] Preferably, in this embodiment, the expression for contribution is:

[0066] , This represents the partial derivative of the i-th time-th type of meteorological data with respect to the power grid load data. This represents the maximum value of the j-th type of meteorological data at time i. This represents the minimum value of the j-th type of meteorological data at time i. This represents an exponential function with the natural constant as its base. This represents the contribution of meteorological data of type j at time i to the power grid load data.

[0067] The above formula can be used to calculate the contribution of the j-th type of meteorological data to the power grid load data at time i. The smaller the extreme value difference of the j-th type of meteorological data in the power grid load data at time i, the less obvious the randomness of the j-th type of meteorological data. At the same time, the larger the partial derivative of the j-th type of meteorological data, the more significant the contribution of the j-th type of meteorological data to the power grid load forecast.

[0068] For each moment in the time window, the corresponding contribution can be calculated separately. Since the contribution of different meteorological data is different and there is a correlation between different categories of meteorological data, the contribution can be used to cluster meteorological data categories, thereby reducing the data dimensionality in the power grid load forecasting process and enabling rapid forecasting of power grid load data.

[0069] Specifically, within the time window, the number of categories of reduced meteorological data is set, and this number is recorded as the final number. In this embodiment, the final number is 2.

[0070] An optimization function is constructed based on the difference between the contribution of any two types of meteorological data at each time step and the mean contribution at each time step to determine the two remaining types of meteorological data after dimensionality reduction.

[0071] Preferably, in this embodiment, the expression for the optimization function is:

[0072] , This represents the contribution of the j-th type of meteorological data at time i to the power grid load data. This represents the average contribution at time i. Indicates the length of the time window. Indicates the final quantity. Represents the minimum optimization function. This represents the optimal function value.

[0073] The two types of meteorological data corresponding to the smallest optimal function value are taken as the two types of meteorological data after dimensionality reduction, and they are recorded as the meteorological selected data.

[0074] The above formula can be used to fuse and cluster the refined meteorological data categories that appear in the power grid load forecasting process. By using the contribution degree, multiple meteorological data categories can be aggregated into the two categories with the highest contribution degree, which significantly reduces the problem of excessive calculation complexity caused by too many forecast parameters of meteorological data types in the power grid load forecasting process.

[0075] Historical meteorological data and power grid load data are used as inputs to the LSTM model. Based on the changing trends of the meteorological and power grid load data, the power grid load data within the future time window is obtained, and its non-periodic prediction value is generated. The prediction accuracy is related to the length of the future time window. In this embodiment, the predicted values ​​of the power grid load data for the next 10 time points are generated.

[0076] Thus, the non-periodic forecast values ​​for the future time window have been obtained.

[0077] Step S005: Obtain the predicted value of the power grid load data based on the periodic and non-periodic predicted values.

[0078] Because residential and commercial activities follow regular patterns, electricity consumption should also exhibit regular variations. Consequently, the power grid load will also show corresponding periodic variations. For periodically changing grid load data, the MUSIC algorithm can be used to extend the periodicity and infer the changes in grid load data at the periodic points. Simultaneously, for aperiodic grid load data, the contribution of different types of meteorological data is used to categorize the data, and by considering the influence of different types of meteorological data, rapid prediction of aperiodic grid load data can be achieved.

[0079] If a periodic point exists within the future time window, the predicted value for that moment is obtained by weighting the periodic and non-periodic predicted values. If no periodic point exists within the future time window, the non-periodic predicted value is the predicted value for that moment. In this embodiment, the weights of both the periodic and non-periodic predicted values ​​are 0.5.

[0080] Thus, the predicted values ​​of the power grid load data for the future time window after the current moment have been obtained.

[0081] The present invention also provides a computer-readable storage medium, which includes a stored computer program; wherein, when the computer program is running, it controls the device where the computer-readable storage medium is located to execute the above-described method for high-precision prediction of power grid load based on big data analysis.

[0082] It should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

[0083] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A high-precision power grid load forecasting method based on big data analysis, characterized in that, The method includes the following steps: Collect meteorological data and power grid load data; Obtain the spectrum diagrams of historically collected meteorological data and power grid load data, and determine the periodic coordination coefficient of each peak of each type of meteorological data based on the periodic differences of the peak values ​​in the power grid load spectrum diagram and the spectrum diagram of each type of meteorological data. Periodic points are selected based on the magnitude of the periodic synergy coefficients of all types of meteorological data; the periodic forecast values ​​of future periodic points are obtained by predicting the changing trend of power grid load data based on the meteorological data and power grid load data of the periodic points. A time window is constructed by pre-setting historical time points, and a contribution function is constructed based on power grid load data and meteorological data. The contribution of each type of meteorological data to the power grid load data is determined based on the maximum difference in data values ​​in each type of meteorological data and the partial derivative of each type of meteorological data in the contribution function. An optimization function is constructed based on the difference between the contribution of meteorological data at each time point and the average contribution at each time point to reduce the dimensionality of the meteorological data categories. The meteorological data and power grid load data of the remaining categories after dimensionality reduction in the time window are used to predict non-periodic predicted values ​​for a preset time period. The predicted values ​​of the power grid load data for the preset time period are obtained based on the periodic and non-periodic predicted values ​​within the preset time period. The method for selecting periodic points based on the magnitude of all periodic synergy coefficients for each type of meteorological data is as follows: Calculate the mean of all periodic coordination coefficients in each type of meteorological data, and use the mean as the power grid periodic load threshold; record the time corresponding to the peak value where the periodic coordination coefficient is greater than or equal to the power grid periodic load threshold as the periodic point; The contribution rate is positively correlated with the partial derivative and negatively correlated with the maximum difference in meteorological data; The expression for the optimization function is: , This represents the contribution of the j-th type of meteorological data at time i to the power grid load data. This represents the average contribution at time i. Indicates the length of the time window. Indicates the final quantity. Represents the minimum optimization function. This represents the optimal function value; The method for obtaining the predicted value of the power grid load data for the preset time period based on the periodic and non-periodic predicted values ​​within the preset time period is as follows: If there is a periodic point within the future time window, the predicted value for that moment is obtained by weighting the periodic and non-periodic predicted values; if there is no periodic point within the future time window, the non-periodic predicted value is the predicted value for that moment.

2. The high-precision power grid load forecasting method based on big data analysis as described in claim 1, characterized in that, The meteorological data includes temperature, humidity, rainfall, and wind speed.

3. The high-precision power grid load forecasting method based on big data analysis as described in claim 1, characterized in that, The periodic synergy coefficient is negatively correlated with the periodic difference.

4. The high-precision power grid load forecasting method based on big data analysis as described in claim 1, characterized in that, The dependent variable of the contribution function is the power grid load data, and the independent variable is meteorological data for all classes.

5. The high-precision power grid load forecasting method based on big data analysis as described in claim 1, characterized in that, The average contribution at each time point is the average contribution of all types of meteorological data corresponding to that time point, and the final number is the number of categories after dimensionality reduction of the preset meteorological data categories.

6. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored computer program; wherein, when the computer program is executed, it controls the device on which the computer-readable storage medium is located to perform the high-precision power grid load forecasting method based on big data analysis as described in any one of claims 1-5.

Images