Time-sharing electric quantity prediction method based on data driving

Abstract

The invention provides a time-sharing electric quantity prediction method based on data driving. The method comprises the following steps: acquiring first influence factor data of time-sharing electric quantity of a to-be-predicted day; according to the first influence factor data and a preset classification model, obtaining a time-sharing electric quantity category to which the to-be-predicted day belongs; determining time-sharing electric quantity characteristics and dominant influence factors of the to-be-predicted day according to the time-sharing electric quantity category to which the to-be-predicted day belongs, and determining first data corresponding to the dominant influence factors of the to-be-predicted day from the first influence factor data; determining a time-sharing electric quantity prediction model matched with the to-be-predicted day according to the time-sharing electric quantity characteristics of the to-be-predicted day, the first data and a preset association rule base; and determining time-sharing electric quantity data of the to-be-predicted day based on the first data and a time-sharing electric quantity prediction model. The method can improve the prediction precision of the time-sharing electric quantity.

Description

Technical Field

[0001] The present application relates to the technical field of power forecasting, and in particular to a data-driven time-sharing power forecasting method. Background Art

[0002] Time-of-use electricity forecasting is based on historical data. It summarizes the changing rules between time-of-use electricity and time, and parameterizes and models the rules. It uses model algorithms based on the "inertia principle" such as time series or trend extrapolation, or uses artificial neural networks, support vector machines, random forests, wavelet analysis and other intelligent algorithms that can perform real-time learning to infer future electricity. With the further development of electricity, the real-time results of time-of-use electricity forecasting have also become important reference data for the optimal allocation of power resources. Therefore, it is very necessary to conduct reliable and accurate time-of-use electricity forecasting.

[0003] In the related art, the prediction of time-sharing electricity is generally achieved through the influencing factors of time-sharing electricity. Since there are many influencing factors of time-sharing electricity, if all the influencing factors of time-sharing electricity are considered, the dimension of the input data of the time-sharing electricity prediction model is huge, which will lead to low calculation efficiency and low prediction accuracy. In order to improve the calculation efficiency, the main factors affecting the time-sharing electricity can be extracted as the input of the time-sharing electricity prediction model by reducing the dimension of the prediction model input data, but the accuracy of the obtained time-sharing electricity prediction result is still not high. Summary of the invention

[0004] The embodiment of the present application provides a data-driven time-sharing electricity prediction method to improve the prediction accuracy of time-sharing electricity.

[0005] In a first aspect, an embodiment of the present application provides a data-driven time-sharing power prediction method, comprising:

[0006] Obtain the first influencing factor data of the time-sharing electricity consumption on the day to be predicted;

[0007] According to the first influencing factor data and a preset classification model, obtaining the time-sharing electricity category to which the day to be predicted belongs;

[0008] Determine the time-sharing power characteristics and dominant influencing factors of the day to be predicted according to the time-sharing power category to which the day to be predicted belongs, and determine first data corresponding to the dominant influencing factors of the day to be predicted from the first influencing factor data;

[0009] Determine a time-sharing electricity prediction model matching the day to be predicted according to the time-sharing electricity characteristics of the day to be predicted, the first data and a preset association rule library;

[0010] Based on the first data and the time-sharing electricity quantity prediction model, the time-sharing electricity quantity data of the day to be predicted is determined.

[0011] In a possible implementation, determining a time-sharing power forecasting model matching the day to be predicted according to the time-sharing power characteristics of the day to be predicted, the first data, and a preset association rule library includes:

[0012] According to the time-sharing electricity quantity characteristics of the day to be predicted and the first data, matching is performed in the association rule library to determine a matching association rule;

[0013] Determine, according to the matching association rules, a time-sharing electricity quantity prediction model matched by the to-be-predicted day, and the first historical time-sharing electricity quantity data of the historical day matched by the to-be-predicted day;

[0014] The determining, based on the first data and the time-sharing power forecasting model, the time-sharing power data of the to-be-forecasted day includes:

[0015] The first data and the first historical time-sharing electricity data are input into the time-sharing electricity prediction model to obtain the time-sharing electricity data of the day to be predicted.

[0016] In a possible implementation, before obtaining the time-sharing electricity category to which the to-be-predicted day belongs according to the first influencing factor data and a preset classification model, the method further includes:

[0017] Obtain the second historical time-sharing electricity data and the second influencing factor data of different historical days;

[0018] Clustering the second historical time-sharing power data according to the time-sharing power extreme value and the extreme value time in the second historical time-sharing power data to obtain multiple time-sharing power categories and time-sharing power characteristics of each time-sharing power category;

[0019] Perform correlation analysis on the second historical time-of-use electricity data and the second influencing factor data of each time-of-use electricity category in the historical day to determine the dominant influencing factor of each time-of-use electricity category;

[0020] Determine the second data of the dominant influencing factor of each time-sharing power category from the second influencing factor data of each time-sharing power category;

[0021] A classification model is established according to the second data of each time-sharing power category.

[0022] In a possible implementation, before determining the time-sharing power forecasting model matching the day to be predicted according to the time-sharing power characteristics of the day to be predicted, the first data and a preset association rule library, the method further includes:

[0023] Obtain the applicability of each prediction model for time-of-use electricity forecasting in each time-of-use electricity category;

[0024] Establishing multiple project sets according to the time-sharing power characteristics corresponding to each time-sharing power category, the second data, and the applicability of each prediction model;

[0025] Determine the support of each item set based on the frequency of occurrence of each item set;

[0026] Construct a complex item pattern tree according to the item sets whose support is greater than or equal to the preset support threshold;

[0027] Using the frequent item pattern tree, mining frequent item sets;

[0028] Based on the frequent item sets, association rules between time-sharing power characteristics, dominant influencing factors and applicability of the prediction model are established to obtain the association rule base.

[0029] In a possible implementation, the multiple item sets are established according to the time-sharing power characteristics corresponding to each time-sharing power category, the second data, and the applicability of each prediction model, including:

[0030] The time-sharing power characteristics, the second data and the applicability of each prediction model corresponding to each time-sharing power category are clustered respectively to obtain multiple data categories under each time-sharing power category;

[0031] Generalizing the multiple data categories to obtain generalized results of the time-sharing power characteristics, the second data, and the applicability of each prediction model under each time-sharing power category;

[0032] A plurality of item sets are established according to the generalization results.

[0033] In a possible implementation, clustering the second historical time-sharing power data according to the time-sharing power extreme value and the extreme value time in the second historical time-sharing power data to obtain multiple time-sharing power categories and time-sharing power characteristics of each time-sharing power category includes:

[0034] Randomly select a preset number of historical days as cluster centers;

[0035] According to the time-sharing power extreme value and extreme value time in the second historical time-sharing power data of each historical day, and the time-sharing power extreme value and extreme value time in the second historical time-sharing power data corresponding to the cluster center, respectively calculate the distance from each historical day to each cluster center;

[0036] Divide each historical day into the cluster where its corresponding target cluster center is located; wherein the target cluster center corresponding to the historical day is the cluster center corresponding to the minimum distance of the historical day;

[0037] Calculate the time-sharing power average value of the second historical time-sharing power data of each historical day in each cluster, and determine a new cluster center based on the time-sharing power average value; complete an iterative calculation of the cluster center;

[0038] Iterate continuously until the cluster center remains unchanged or the maximum number of iterations is reached, and according to the final cluster center, multiple time-sharing power categories and the time-sharing power characteristics of each time-sharing power category are determined.

[0039] In a possible implementation, the distance from each historical day to each cluster center is calculated based on the time-sharing power extreme value and extreme value time in the second historical time-sharing power data of each historical day, and the time-sharing power extreme value and extreme value time in the second historical time-sharing power data corresponding to the cluster center, including:

[0040] According to the expression: Calculate the distance from each historical day to each cluster center respectively;

[0041] Where, d sCr represents the distance from the historical day s to the cluster center Cr, t maxks represents the peak time of the kth hourly electricity peak in the second historical hourly electricity data of historical day s, t maxkCr represents the peak time of the kth time-sharing power peak in the second historical time-sharing power data of the cluster center Cr, l maxks represents the kth hourly electricity peak value in the second historical hourly electricity data of historical day s, l maxkCr represents the kth time-sharing power peak value in the second historical time-sharing power data of the cluster center Cr, t minjs Indicates the valley time of the jth hourly electricity valley value in the second historical hourly electricity data of historical day s, t minjCr represents the valley time of the jth time-sharing electricity valley value in the second historical time-sharing electricity data of the cluster center Cr, l minjs represents the jth hourly electricity valley value in the second historical hourly electricity data of historical day s, l minjCr represents the jth time-sharing electricity valley value in the second historical time-sharing electricity data of the cluster center Cr, m represents the total number of time-sharing electricity peak values ​​in the second historical time-sharing electricity data of the historical day s, and n represents the total number of time-sharing electricity valley values ​​in the second historical time-sharing electricity data of the historical day s.

[0042] In a second aspect, an embodiment of the present application provides a data-driven time-sharing power prediction device, comprising:

[0043] An acquisition module, used to acquire the first influencing factor data of the time-sharing electricity quantity on the day to be predicted;

[0044] A classification module, used for obtaining the time-sharing electricity category to which the predicted day belongs according to the first influencing factor data and a preset classification model;

[0045] A determination module, configured to determine the time-sharing power characteristics and dominant influencing factors of the day to be predicted according to the time-sharing power category to which the day to be predicted belongs, and determine first data corresponding to the dominant influencing factors of the day to be predicted from the first influencing factor data;

[0046] A selection module, configured to determine a time-sharing electricity prediction model matching the day to be predicted based on the time-sharing electricity characteristics of the day to be predicted, the first data, and a preset association rule library;

[0047] A prediction module is used to determine the time-sharing electricity data of the day to be predicted based on the first data and the time-sharing electricity prediction model.

[0048] In a third aspect, an embodiment of the present application provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the steps of the method described in the first aspect or any possible implementation method of the first aspect are implemented.

[0049] In a fourth aspect, an embodiment of the present application provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the steps of the method described in the first aspect or any possible implementation method of the first aspect.

[0050] In a fifth aspect, an embodiment of the present application provides a computer program product. When the computer program product runs on an electronic device, the electronic device executes the steps of the method described in the first aspect or any possible implementation method of the first aspect.

[0051] Compared with the prior art, the embodiments of the present invention have the following beneficial effects:

[0052] The embodiment of the present application determines the time-sharing electricity category of the day to be predicted through the first influencing factor data and the classification model of the day to be predicted, and then determines the time-sharing electricity characteristics and dominant influencing factors of the day to be predicted through the determined time-sharing electricity category, so as to determine the change law of the time-sharing electricity on the day to be predicted, and select the influencing factors closely related to the change law of the time-sharing electricity on the day to be predicted, so as to improve the accuracy of time-sharing electricity prediction, reduce the interference of other influencing factors, and reduce the dimension of input data; through the first data corresponding to the time-sharing electricity characteristics and dominant influencing factors determined above, determine the time-sharing electricity prediction model matching the day to be predicted from the association rule library, and then obtain the time-sharing electricity data of the day to be predicted by inputting the first data into the time-sharing electricity prediction model, so as to use the association rule library to accurately match the prediction model with high applicability to the day to be predicted, and further improve the prediction accuracy of time-sharing electricity. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0054] Figure 1 is a flow chart of an implementation method of a time-sharing power prediction method based on data-driven provided in an embodiment of the present application;

[0055] Figure 2 This is a schematic diagram of the time-sharing electricity consumption of a historical day provided by an embodiment of the present application;

[0056] Figure 3 is a curve diagram of time-sharing power data as training data provided in an embodiment of the present application;

[0057] Figure 4 It is a structural schematic diagram of a time-sharing power prediction device based on data drive provided in an embodiment of the present application;

[0058] Figure 5 It is a schematic diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0059] In the following description, specific details such as specific system structures, technologies, etc. are provided for the purpose of illustration rather than limitation, so as to provide a thorough understanding of the embodiments of the present application. However, it should be clear to those skilled in the art that the present application may also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to prevent unnecessary details from obstructing the description of the present application.

[0060] In order to make the purpose, technical solutions and advantages of the present application clearer, specific embodiments will be described below in conjunction with the accompanying drawings.

[0061] Figure 1 The implementation flow chart of the data-driven time-sharing power prediction method provided in the embodiment of the present application is described in detail as follows:

[0062] Step 101, obtaining first influencing factor data of time-sharing electricity consumption on a day to be predicted.

[0063] In this embodiment, the first influencing factor data may include meteorological data and date type of the day to be predicted, etc. The meteorological data may be temperature, humidity, light intensity, etc., and the date type may be a week type, a holiday type, etc.

[0064] Step 102: Obtain the time-sharing electricity category to which the day to be predicted belongs based on the first influencing factor data and a preset classification model.

[0065] In this embodiment, the inventors consider that even in the same region, different types of time-sharing electricity have different time-sharing electricity variation patterns, and at the same time, the influencing factors may also be different. Therefore, the time-sharing electricity category to which the day to be predicted belongs is determined through the first influencing factor data and the classification model of the day to be predicted, so as to accurately determine the variation pattern of the time-sharing electricity of the day to be predicted, that is, the time-sharing electricity characteristics, and perform targeted prediction.

[0066] Here, the classification model can be established by using a Classification And Regression Tree (CART). Specifically, CART can be used to establish a relationship between different time-sharing power categories and corresponding dominant influencing factors, thereby obtaining a classification model.

[0067] Step 103, according to the time-sharing electricity category to which the day to be predicted belongs, determine the time-sharing electricity characteristics and dominant influencing factors of the day to be predicted, and determine the first data corresponding to the dominant influencing factor of the day to be predicted from the first influencing factor data.

[0068] In this embodiment, the dominant influencing factor is an influencing factor with a high degree of correlation with the time-sharing power of the day to be predicted. The first data corresponding to the dominant influencing factor is determined from the first influencing factor data of the day to be predicted, and the data of the dominant influencing factor corresponding to the time-sharing power category to which the day to be predicted belongs can be screened out, so as to reduce the input dimension of the time-sharing power prediction model and improve the prediction accuracy.

[0069] Step 104 , determining a time-sharing power forecasting model matching the day to be predicted based on the time-sharing power characteristics of the day to be predicted, the first data, and a preset association rule library.

[0070] In this embodiment, by utilizing the time-sharing electricity characteristics of the day to be predicted and the first data, matching is performed in the association rule library, and an association rule matching the day to be predicted can be found, thereby determining a time-sharing electricity prediction model applicable to the day to be predicted, and improving the prediction accuracy of time-sharing electricity from the model itself.

[0071] Here, considering the time-sharing electricity characteristics, the association rules that may match the day to be predicted can be quickly locked from the association rule library to improve the efficiency of association rule matching.

[0072] Step 105: Determine the time-sharing power data of the day to be predicted based on the first data and the time-sharing power prediction model.

[0073] In this embodiment, the time-sharing electricity consumption is predicted by using the first data determined in the above steps and the time-sharing electricity prediction model, which can reduce the dimensions of the influencing factors and the interference of the influencing factors with low correlation. At the same time, selecting a prediction model that matches the time-sharing electricity consumption characteristics of the day to be predicted can further improve the accuracy of the time-sharing electricity data prediction.

[0074] The embodiment of the present application determines the time-sharing electricity category of the day to be predicted through the first influencing factor data and the classification model of the day to be predicted, and then determines the time-sharing electricity characteristics and dominant influencing factors of the day to be predicted through the determined time-sharing electricity category, so as to determine the change law of the time-sharing electricity on the day to be predicted, and select the influencing factors closely related to the change law of the time-sharing electricity on the day to be predicted, so as to improve the accuracy of time-sharing electricity prediction, reduce the interference of other influencing factors, and reduce the dimension of input data; through the first data corresponding to the time-sharing electricity characteristics and dominant influencing factors determined above, determine the time-sharing electricity prediction model matching the day to be predicted from the association rule library, and then obtain the time-sharing electricity data of the day to be predicted by inputting the first data into the time-sharing electricity prediction model, so as to use the association rule library to accurately match the prediction model with high applicability to the day to be predicted, and further improve the prediction accuracy of time-sharing electricity.

[0075] In some embodiments, a time-sharing electricity prediction model that matches the day to be predicted is determined based on the time-sharing electricity characteristics of the day to be predicted, the first data, and a preset association rule library. First, a match is performed in the association rule library based on the time-sharing electricity characteristics of the day to be predicted and the first data to determine the matching association rules; then, based on the matching association rules, the time-sharing electricity prediction model that matches the day to be predicted and the first historical time-sharing electricity data of the historical day that matches the day to be predicted are determined.

[0076] In this embodiment, when determining the time-sharing electricity forecasting model that matches the day to be forecasted, the matching association rules can be determined from the association rule library, and the time-sharing electricity forecasting model that matches the day to be forecasted can be determined through the forecasting model in the matching association rules.

[0077] When there is a matching association rule, the prediction model corresponding to the association rule is the time-of-use electricity prediction model that matches the day to be predicted.

[0078] When there are multiple matching association rules, if the prediction models corresponding to the multiple matching association rules are the same, the prediction model can be directly determined as the time-sharing electricity prediction model that matches the day to be predicted. If the prediction models corresponding to the multiple matching association rules are different, the matched association rules can be sorted according to their support and confidence, and the time-sharing electricity prediction model that matches the day to be predicted can be determined according to the sorting results. For example, the prediction model corresponding to the association rule with the highest support or the highest confidence is used to determine the time-sharing electricity prediction model that matches the day to be predicted; or, the prediction models corresponding to the association rules are weighted according to the support and confidence, and the prediction model with the highest probability is selected as the time-sharing electricity prediction model that matches the day to be predicted.

[0079] In addition, when matching in the association rule library, the first historical time-sharing electricity data of the historical day that matches the day to be predicted can also be determined, and historical time-sharing electricity data with similar change rules to the time-sharing electricity data of the day to be predicted can be found.

[0080] Accordingly, based on the first data and the time-sharing electricity prediction model, the time-sharing electricity data of the day to be predicted is determined. The first data and the first historical time-sharing electricity data can be input into the time-sharing electricity prediction model to obtain the time-sharing electricity data of the day to be predicted.

[0081] In this embodiment, when predicting the time-sharing electricity data of the predicted day, the determined first historical time-sharing electricity data can also be used as input data of the time-sharing electricity prediction model, and the matched first historical time-sharing electricity data can be considered in the prediction process to further improve the accuracy of the time-sharing electricity prediction model.

[0082] In some embodiments, before obtaining the time-sharing electricity category to which the day to be predicted belongs according to the first influencing factor data and the preset classification model, it is also possible to:

[0083] Step 1: Obtain the second historical time-sharing electricity data and the second influencing factor data of different historical days.

[0084] In this embodiment, the influencing factors included in the second influencing factor data may be the same as the influencing factors included in the first influencing factor data, such as the meteorological data and date type of the day to be predicted.

[0085] Step 2: clustering the second historical time-sharing power data according to the time-sharing power extreme values ​​and extreme value moments in the second historical time-sharing power data to obtain multiple time-sharing power categories and time-sharing power characteristics of each time-sharing power category.

[0086] In this embodiment, considering that different time-sharing power types correspond to different time-sharing power characteristics, and thus correspond to different time-sharing power curves, but there are correlations and similarities between the same time-sharing power types, the time-sharing power characteristics of the collected time-sharing power data can be extracted and analyzed.

[0087] Here, a clustering method is used to extract the time-sharing power characteristics from the time-sharing power data, and then different time-sharing power prediction models are established based on the clustering results, which can improve the accuracy of time-sharing power prediction.

[0088] When clustering, the time-sharing power extreme values ​​and extreme value moments can be used for analysis, which can simultaneously consider the time-sharing power and time information to better reflect the user's power consumption patterns.

[0089] Step three, performing correlation analysis on the second historical time-sharing electricity data and the second influencing factor data of each time-sharing electricity category in the historical day to determine the dominant influencing factor of each time-sharing electricity category.

[0090] In this embodiment, there are many factors that affect the time-sharing power data, such as weather conditions, temperature, humidity, week type, holidays, etc. The factors affecting the time-sharing power of different types and regions will be different. For example, the influence of week type and holiday factors on municipal life electricity consumption will be much greater than the influence on other time-sharing power types. Even for the time-sharing power in the same area, its influencing factors will be different in different time periods. For example, in different seasons, the influence of temperature on the time-sharing power will be higher than other factors.

[0091] Based on this, the corresponding dominant influencing factors can be determined for each time-sharing power category. For example, the grey correlation analysis method can be used for research.

[0092] Step 4: Determine the second data of the dominant influencing factor of each time-sharing power category from the second influencing factor data of each time-sharing power category.

[0093] In this embodiment, each time-sharing electricity category corresponds to a group of dominant influencing factors. The data of the dominant influencing factors corresponding to each time-sharing electricity category can be found from the second influencing factor data to form the second data, so that accurate processing can be performed subsequently based on the second data corresponding to each time-sharing electricity category.

[0094] Step 5: Establish a classification model based on the second data of each time-sharing power category.

[0095] In this embodiment, a classifier is established to obtain the relationship between the time-sharing power category and the dominant influencing factors, thereby obtaining the time-sharing power category to which the day to be predicted belongs.

[0096] For example, a classification model can be established based on the degree of association between each time-sharing power category and the dominant influencing factor, as well as the data of the dominant influencing factor, i.e., the second data. The best feature can be selected for splitting through the information gain of the decision tree to form a classification model.

[0097] In some embodiments, according to the time-sharing power extreme value and the extreme value time in the second historical time-sharing power data, the second historical time-sharing power data is clustered to obtain multiple time-sharing power categories and time-sharing power characteristics of each time-sharing power category, which can be:

[0098] Step 1: Randomly select a preset number of historical days as cluster centers.

[0099] Step 2: Calculate the distance from each historical day to each cluster center based on the extreme value and extreme time of the time-sharing electricity in the second historical time-sharing electricity data of each historical day, and the extreme value and extreme time of the time-sharing electricity in the second historical time-sharing electricity data corresponding to the cluster center.

[0100] Step three, divide each historical day into the cluster where its corresponding target cluster center is located; wherein the target cluster center corresponding to the historical day is the cluster center corresponding to the minimum distance of the historical day.

[0101] Step 4: Calculate the average time-sharing electricity value of the second historical time-sharing electricity data of each historical day in each cluster, and determine a new cluster center based on the average time-sharing electricity value; complete an iterative calculation of the cluster center.

[0102] Step 5: Iterate continuously until the cluster center remains unchanged or the maximum number of iterations is reached, and according to the final cluster center, multiple time-sharing power categories and the time-sharing power characteristics of each time-sharing power category are determined.

[0103] In this embodiment, when calculating the distance based on the time-sharing power extreme value and extreme value time, the time-sharing power and time information can be considered simultaneously, reflecting the peaks, troughs and corresponding times of the time-sharing power curve, which can better reflect the user's power consumption pattern and achieve accurate clustering.

[0104] The extreme value includes the peak value of the time-sharing power and the valley value of the time-sharing power. The extreme value moment includes the peak moment corresponding to the peak value of the time-sharing power and the valley moment corresponding to the valley value of the time-sharing power. Figure 2 As shown, max1s Indicates the first hourly electricity peak value in the second historical hourly electricity data of historical day s, t max1s Indicates the peak time corresponding to the first time-sharing power peak; l max2s Indicates the second hourly electricity peak value in the second historical hourly electricity data of historical day s, t max2sIndicates the peak time corresponding to the second time-sharing power peak; l min1s Indicates the first hourly electricity valley value in the second historical hourly electricity data of historical day s, t min1s Indicates the valley time corresponding to the first time-sharing electricity valley.

[0105] Here, the number of clusters in cluster analysis should not be too many or too few. Too many clusters will result in too few samples in each class, which is not conducive to model training, while too few clusters will make the classification results too rough and cause inaccurate prediction results. For example, 10, 12 or 14 cluster centers can be selected.

[0106] Optionally, according to the extreme value and the time of the time-sharing electricity in the second historical time-sharing electricity data of each historical day, and the extreme value and the time of the time-sharing electricity in the second historical time-sharing electricity data corresponding to the cluster center, the distance from each historical day to each cluster center is calculated respectively, which can be based on the expression: Calculate the distance from each historical day to each cluster center separately.

[0107] Where, d sCr represents the distance from the historical day s to the cluster center Cr, t maxks represents the peak time of the kth hourly electricity peak in the second historical hourly electricity data of historical day s, t maxkCr represents the peak time of the kth time-sharing power peak in the second historical time-sharing power data of the cluster center Cr, l maxks represents the kth hourly electricity peak value in the second historical hourly electricity data of historical day s, l maxkCr represents the kth time-sharing power peak value in the second historical time-sharing power data of the cluster center Cr, t minjs Indicates the valley time of the jth hourly electricity valley value in the second historical hourly electricity data of historical day s, t minjCr represents the valley time of the jth time-sharing electricity valley value in the second historical time-sharing electricity data of the cluster center Cr, l minjs represents the jth hourly electricity valley value in the second historical hourly electricity data of historical day s, l minjCr represents the jth time-sharing electricity valley value in the second historical time-sharing electricity data of the cluster center Cr, m represents the total number of time-sharing electricity peak values ​​in the second historical time-sharing electricity data of the historical day s, and n represents the total number of time-sharing electricity valley values ​​in the second historical time-sharing electricity data of the historical day s.

[0108] In this embodiment, before calculating the distance, dimensionless processing or weighted processing may be performed on the second historical time-sharing power data, so that the time-sharing power and time can be calculated simultaneously.

[0109] In some embodiments, the second historical time-sharing power data and the second influencing factor data of each time-sharing power category are correlated and analyzed to determine the dominant influencing factor of each time-sharing power category, which may be:

[0110] For each time-sharing electricity category, according to the second historical time-sharing electricity data and the second influencing factor data of the historical day in the time-sharing electricity category, the correlation coefficient between each second influencing factor data in the time-sharing electricity category and the second historical time-sharing electricity data at each time point is calculated respectively.

[0111] Here, the second historical time-sharing electricity data corresponding to the time-sharing electricity category can be used as a reference series for association analysis, and the second influencing factor data corresponding to the time-sharing electricity category can be used as a comparison series for association analysis, wherein each second influencing factor data constitutes a comparison series, so as to perform analysis on each influencing factor.

[0112] Optionally, you can use an expression: Calculate the correlation coefficient between each second influencing factor data in the time-sharing power category and the second historical time-sharing power data at each time point. i (t) represents the correlation coefficient between the i-th second influencing factor data and the second historical time-sharing power data corresponding to the t-th time point, |x 0 (t)-x i (t)| represents the difference between the data at the tth time point in the i-th second influencing factor data and the data at the tth time point in the second historical time-sharing electricity data, ρ represents the resolution coefficient; Δmin represents the minimum value of the two poles, is the minimum difference between each comparison sequence and the reference sequence; Δmax represents the maximum difference between the two extremes, The maximum value of the difference between each compared sequence and the reference sequence.

[0113] Then, according to the correlation coefficients at various time points, the correlation between each second influencing factor data in the time-sharing power category and the second historical time-sharing power data is calculated.

[0114] Here, for each type of second influencing factor data, the correlation coefficient at each corresponding time point is used to calculate the correlation degree, and the calculation formula can be: In the formula, r i represents the correlation between the i-th second influencing factor data and the second historical time-sharing power data, T represents the set of time points, ε i (t) represents the correlation coefficient of the i-th second influencing factor data at the t-th time point with respect to the second historical time-sharing electricity quantity data.

[0115] Finally, based on the correlation, the dominant influencing factors of the time-of-use electricity category are determined.

[0116] In this embodiment, the influencing factors may be sorted according to the magnitude of the correlation, and the influencing factors with a correlation greater than a preset correlation threshold may be determined as the dominant influencing factors of the time-sharing power category.

[0117] In addition, before calculating the correlation coefficient, the reference sequence and the comparison sequence can be dimensionally processed to accurately calculate the correlation coefficient and the degree of correlation and eliminate the influence of the dimension of the data.

[0118] In some embodiments, before determining the time-sharing power forecasting model matching the day to be predicted based on the time-sharing power characteristics of the day to be predicted, the first data, and a preset association rule library, it is also possible to:

[0119] Step 1: Obtain the applicability of each prediction model for time-of-use electricity prediction in each time-of-use electricity category.

[0120] In this embodiment, multiple prediction models can be trained using the second historical time-sharing electricity data corresponding to each time-sharing electricity category and the second data of the dominant influencing factors. The prediction model can be used to predict at least one of the mean square error, root mean square error and average relative error of the time-sharing electricity to determine the applicability of the prediction model.

[0121] Here, the prediction model can adopt models such as multiple correlation algorithm, time series method, cluster analysis and grey correlation analysis.

[0122] In addition, the applicability of the prediction model here can be determined for each time-sharing electricity category; or the dominant influencing factors and second data corresponding to each time-sharing electricity category can be classified, and the applicability of the prediction model can be determined for each dominant influencing factor and second data category.

[0123] Step 2: Establish multiple project sets according to the time-sharing power characteristics corresponding to each time-sharing power category, the second data, and the applicability of each prediction model.

[0124] In this embodiment, the item set is a set of items without a specific order, and the item set can be established by using the time-sharing power characteristics corresponding to each time-sharing power category and the second data, thereby forming multiple item sets. There is a corresponding relationship between the item set and the applicability of the prediction model.

[0125] Step three, determine the support of each item set based on the frequency of occurrence of each item set.

[0126] Here, support is a measure that indicates how often an itemset appears in all transactions. For example, the support of an itemset can be the ratio of the number of transactions containing the itemset to the total number of transactions.

[0127] Step 4: construct a complex item pattern tree based on the item set whose support is greater than or equal to the preset support threshold.

[0128] In this embodiment, if the number of occurrences of an itemset exceeds a preset support threshold, it means that the itemset is a frequent itemset, and such itemset can be used to establish a Frequent Pattern Tree (FP tree). The frequent itemset can be inserted into the tree structure, and if the itemset already exists in the tree structure, the corresponding count is updated; if it does not exist, a new node is added to the tree structure, thereby forming an FP tree.

[0129] Step 4: Use the frequent item pattern tree to mine frequent item sets.

[0130] In this embodiment, for each item in the FP tree, a conditional pattern base and a conditional FP tree are generated. Frequent item sets can be mined on the conditional FP tree in a recursive manner until all items are processed. When mining frequent item sets from the FP tree, the nodes in the tree can be recursively accessed and the support of the item set can be calculated. If the support of the item set is greater than or equal to the support threshold, it is regarded as a frequent item set.

[0131] Step 5: Based on frequent item sets, establish association rules between time-sharing power characteristics, dominant influencing factors and applicability of the prediction model to obtain an association rule base.

[0132] In this embodiment, after frequent item sets are mined, association rules can be generated based on these frequent item sets. The association rules can be generated by calculating the confidence between item sets.

[0133] For example, association rules can be generated from frequent item sets according to a preset confidence threshold. The form of association rules is "predecessor → consequent", for example, "if the weather is sunny and the time period is a weekday, then model A is applicable".

[0134] Optionally, multiple project sets are established based on the time-sharing power characteristics, the second data and the applicability of each prediction model corresponding to each time-sharing power category. The time-sharing power characteristics, the second data and the applicability of each prediction model corresponding to each time-sharing power category can be clustered separately to obtain multiple data categories under each time-sharing power category; the multiple data categories are generalized to obtain generalized results of the time-sharing power characteristics, the second data and the applicability of each prediction model under each time-sharing power category; finally, multiple project sets are established based on the generalized results.

[0135] In this embodiment, since the second data of each influencing factor is a specific numerical value, the second data corresponding to the same influencing factor on different historical days in the same time-sharing electricity category is likely to be different. In order to better mine frequent item sets, the time-sharing electricity characteristics, the second data and the applicability of each prediction model can be clustered and divided, and similar or similar data can be divided into one data category.

[0136] When performing clustering, the clustering characteristics of the second data corresponding to the same time-sharing power characteristics and the clustering characteristics of the second data corresponding to the same influencing factors can be considered simultaneously, so as to accurately and reasonably divide and classify the data of each influencing factor.

[0137] Generalize each data category, that is, abstract and summarize the characteristics and range of each data category obtained by clustering to form a concise description. For example, the generalization result of the data category corresponding to temperature can be that the temperature is 20℃~30℃, and an association rule of a certain time-sharing power category can be that the high power consumption period on weekdays is from 9 am to 5 pm, the temperature is 20℃~30℃, and model A is applicable.

[0138] In a specific embodiment, a city’s total time-sharing electricity data of 8760h in 2020 is used as training data, and the city’s time-sharing electricity data on March 11, 2021 is used to verify the effectiveness of time-sharing electricity prediction.

[0139] In addition to time-based electricity consumption data, the existing data also include data on influencing factors such as the city’s daily maximum temperature, average temperature, minimum temperature, light, relative humidity, week type and holiday type in 2020, and the data is pre-processed.

[0140] Preprocessing can include missing value imputation and outlier handling.

[0141] You can also quantify the attributes of week type and whether it is a holiday. The date of a legal holiday is quantified as 3, the day after a legal holiday is quantified as 2, and the dates other than these two are quantified as 1.

[0142] In addition, due to the different dimensions of data such as maximum temperature, average temperature, minimum temperature, light, relative humidity, holidays and week types, there are large differences in values, so the above data need to be normalized.

[0143] The time-sharing power data is extracted and analyzed for time-sharing power characteristics. The curve diagram of the time-sharing power data is as follows: Figure 3 As shown, cluster analysis is performed on the above time-sharing power data with cluster data 12.

[0144] The grey correlation analysis method is used to conduct correlation analysis on the influencing factor data and time-sharing power data to obtain the correlation degree of each influencing factor.

[0145] Taking one of the time-sharing electricity categories for analysis, the correlation distribution of each influencing factor is shown in Table 1:

[0146] Table 1 Correlation distribution of influencing factors

[0147]

[0148] As can be seen from Table 1, the top three influencing factors that are most closely related to the time-sharing power value are the maximum temperature, average temperature and relative humidity. At the same time, it can be seen that the correlation values ​​between each influencing factor and the time-sharing power are all greater than 0.4, so all of the above influencing factors can be considered without deletion; when there are many influencing factors and the correlation of some influencing factors is less than 0.4, the influencing factor can no longer be considered, thus obtaining the dominant influencing factor.

[0149] Then, through the dominant influencing factors of each time-sharing power category and its corresponding influencing factor data, a decision tree model can be used to establish a classification model. When predicting the day to be predicted, the values ​​of the influencing factors of the day to be predicted can be input into the above classification model to obtain the classification result of the day to be predicted, which can be shown in Table 2.

[0150] Table 2 Classification results of the days to be predicted

[0151]

[0152] In order to facilitate the use of FP growth algorithm to mine the correlation rules between influencing factors and model applicability, the discrete weather factors in the influencing factors can be binned and generalized. Clustering can be used to divide similar weather factors into the same range. For example, in a cluster, the highest temperatures include 22℃, 24.1℃, 23℃, 26℃, 25.2℃, 23.6℃, ​​22.7℃ and 24℃, among which the maximum value is 26℃ and the minimum value is 22℃, and the corresponding range can be divided into [22℃, 26℃].

[0153] The density estimation method can also be used to classify each influencing factor, and the results are shown in Table 3.

[0154] Table 3 Distribution of influencing factors

[0155] Meteorological factors 1 2 3 4 <![CDATA[F d_max (W / m 2 )]]> ≤732.5 732.5～867 867～986 ≥986 <![CDATA[F d_mean (W / m 2 )]]> ≤65 65～147.5 147.5～208 ≥208 <![CDATA[F d_difmax (W / m 2 )]]> ≤216 216～293.3 293.3～387 ≥387 <![CDATA[F d_difmean (W / m 2 )]]> ≤76.4 76.4～94.7 94.7～110.5 ≥110.5 <![CDATA[T d_max (℃)]]> ≤21.3 21.3～26 26～31.3 ≥31.3 <![CDATA[T d_mean (℃)]]> ≤10.5 10.5～19.6 19.6～25.5 ≥25.5 <![CDATA[T d_min (℃)]]> ≤8.4 8.4～15.6 15.6～24.8 ≥24.8 <![CDATA[S d_time (h)]]> ≤7 8～9 10 ≥11

[0156] In Table 3 above, F d_max Indicates the maximum light intensity, F d_mean Indicates the average light intensity, F d_difmaxIndicates the maximum light intensity difference, F d_difmean Represents the average light intensity difference, T d_max Indicates the maximum temperature, T d_mean represents the average temperature, T d_min Indicates the minimum temperature, S d_time Indicates the duration of lighting.

[0157] The FP growth algorithm is used to mine association rules for time-sharing power characteristics, influencing factors, and applicability of the prediction model. The support threshold can be set to 6, and the confidence threshold can be set to 0.7 to obtain the corresponding association rules and establish an association rule base for the prediction model, so that the prediction day can be analyzed and the time-sharing power prediction model corresponding to the prediction day can be determined.

[0158] Exemplarily, the association rule base may be as shown in Table 4 below:

[0159] Table 4 Schematic diagram of association rule base

[0160]

[0161] In Table 4 above, 1, 2, 3 and 4 in the conditions are the levels of the influencing factors in Table 3, and models A, B and C are the prediction models with the highest applicability of the corresponding association rules. When an association rule is matched, the time-sharing power prediction model can be obtained through the prediction model corresponding to the association rule. The power data 1, 2, 3...11 can be the time-sharing power data of a typical day, or the historical time-sharing power data of all historical days corresponding to the association rule.

[0162] A traditional prediction method was selected to perform time-sharing power prediction and compared with the method provided in this application. The prediction results are shown in Table 5.

[0163] Table 5 Comparison of prediction results

[0164]

[0165] As shown in Table 5 above, taking 12 o'clock as an example, the predicted value of the traditional algorithm is 27280GWh, the predicted value of the method provided by this application is 25630GWh, and the actual value is 26115GWh. The result error of the time-sharing power forecasting method provided by this application is smaller, and other time points are similar. It can be seen that the relative error of the method provided by this application is lower than that of the traditional algorithm, and can be controlled within 3%.

[0166] Using the average relative error to measure accuracy, the average relative error of the traditional algorithm is 4.69%, and the average relative error of the method provided in the present application is 1.56%. It can be seen that the prediction results of the method provided in the present application are closer to the actual values ​​and can improve the prediction accuracy.

[0167] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0168] The following is an embodiment of the device of the present application. For details not described in detail, please refer to the corresponding method embodiment described above.

[0169] Figure 4 The following is a schematic diagram of the structure of a data-driven time-sharing power prediction device provided in an embodiment of the present application. For ease of explanation, only the parts related to the embodiment of the present application are shown, which are described in detail as follows:

[0170] like Figure 4 As shown, the data-driven time-sharing power prediction device 40 includes:

[0171] An acquisition module 41 is used to acquire first influencing factor data of time-sharing electricity consumption on a day to be predicted;

[0172] The classification module 42 is used to obtain the time-sharing electricity category to which the day to be predicted belongs according to the first influencing factor data and a preset classification model;

[0173] The determination module 43 is used to determine the time-sharing power characteristics and dominant influencing factors of the day to be predicted according to the time-sharing power category to which the day to be predicted belongs, and determine the first data corresponding to the dominant influencing factor of the day to be predicted from the first influencing factor data;

[0174] A selection module 44 is used to determine a time-sharing power forecasting model matching the day to be predicted based on the time-sharing power characteristics of the day to be predicted, the first data and a preset association rule library;

[0175] The prediction module 45 is used to determine the time-sharing power data of the day to be predicted based on the first data and the time-sharing power prediction model.

[0176] In a possible implementation, the selection module 44 is specifically configured to:

[0177] According to the time-sharing electricity quantity characteristics of the day to be predicted and the first data, matching is performed in an association rule library to determine a matching association rule;

[0178] According to the matching association rules, determine the time-sharing electricity forecasting model that matches the day to be predicted, and the first historical time-sharing electricity data of the historical day that matches the day to be predicted;

[0179] The prediction module 45 is specifically used for:

[0180] The first data and the first historical time-sharing electricity data are input into a time-sharing electricity prediction model to obtain the time-sharing electricity data of the day to be predicted.

[0181] In a possible implementation, the classification module 42 is further configured to:

[0182] Obtain the second historical time-sharing electricity data and the second influencing factor data of different historical days;

[0183] Clustering the second historical time-sharing power data according to the time-sharing power extreme value and the extreme value time in the second historical time-sharing power data to obtain multiple time-sharing power categories and time-sharing power characteristics of each time-sharing power category;

[0184] Perform correlation analysis on the second historical time-of-use electricity data and the second influencing factor data of each time-of-use electricity category in the historical day to determine the dominant influencing factor of each time-of-use electricity category;

[0185] Determine the second data of the dominant influencing factor of each time-sharing power category from the second influencing factor data of each time-sharing power category;

[0186] A classification model is established according to the second data of each time-sharing power category.

[0187] In a possible implementation, the selection module 44 is further configured to:

[0188] Obtain the applicability of each prediction model for time-of-use electricity forecasting in each time-of-use electricity category;

[0189] Establishing multiple project sets according to the time-sharing power characteristics corresponding to each time-sharing power category, the second data, and the applicability of each prediction model;

[0190] Determine the support of each item set based on the frequency of occurrence of each item set;

[0191] Construct a complex item pattern tree according to the item sets whose support is greater than or equal to the preset support threshold;

[0192] Use the frequent item pattern tree to mine frequent item sets;

[0193] Based on frequent item sets, association rules among time-sharing electricity characteristics, dominant influencing factors and applicability of prediction models are established to obtain an association rule base.

[0194] In a possible implementation, the selection module 44 is specifically configured to:

[0195] The time-sharing power characteristics, the second data and the applicability of each prediction model corresponding to each time-sharing power category are clustered respectively to obtain multiple data categories under each time-sharing power category;

[0196] Generalizing multiple data categories to obtain generalized results of time-sharing power characteristics, second data, and applicability of each prediction model under each time-sharing power category;

[0197] Based on the generalization results, multiple item sets are established.

[0198] In a possible implementation, the classification module 42 is specifically used for:

[0199] Randomly select a preset number of historical days as cluster centers;

[0200] According to the extreme value and time of the time-sharing electricity quantity in the second historical time-sharing electricity quantity data of each historical day, and the extreme value and time of the time-sharing electricity quantity in the second historical time-sharing electricity quantity data corresponding to the cluster center, respectively calculate the distance from each historical day to each cluster center;

[0201] Divide each historical day into the cluster where its corresponding target cluster center is located; wherein the target cluster center corresponding to the historical day is the cluster center corresponding to the minimum distance of the historical day;

[0202] Calculate the average time-sharing electricity value of the second historical time-sharing electricity value data of each historical day in each cluster, and determine a new cluster center based on the average time-sharing electricity value; complete an iterative calculation of the cluster center;

[0203] Iterate continuously until the cluster center remains unchanged or the maximum number of iterations is reached, and according to the final cluster center, multiple time-sharing power categories and the time-sharing power characteristics of each time-sharing power category are determined.

[0204] In a possible implementation, the classification module 42 is specifically used for:

[0205] According to the expression: Calculate the distance from each historical day to each cluster center respectively;

[0206] Where, d sCr represents the distance from the historical day s to the cluster center Cr, t maxks represents the peak time of the kth hourly electricity peak in the second historical hourly electricity data of historical day s, t maxkCr represents the peak time of the kth time-sharing power peak in the second historical time-sharing power data of the cluster center Cr, l maxks represents the kth hourly electricity peak value in the second historical hourly electricity data of historical day s, l maxkCr represents the kth time-sharing power peak value in the second historical time-sharing power data of the cluster center Cr, t minjs Indicates the valley time of the jth hourly electricity valley value in the second historical hourly electricity data of historical day s, t minjCrrepresents the valley time of the jth time-sharing electricity valley value in the second historical time-sharing electricity data of the cluster center Cr, l minjs represents the jth hourly electricity valley value in the second historical hourly electricity data of historical day s, l minjCr represents the jth time-sharing electricity valley value in the second historical time-sharing electricity data of the cluster center Cr, m represents the total number of time-sharing electricity peak values ​​in the second historical time-sharing electricity data of the historical day s, and n represents the total number of time-sharing electricity valley values ​​in the second historical time-sharing electricity data of the historical day s.

[0207] Figure 5 Schematic diagram of an electronic device provided in an embodiment of the present application. Figure 5 As shown, the electronic device 50 of this embodiment includes: a processor 51, a memory 52, and a computer program 53 stored in the memory 52 and executable on the processor 51. When the processor 51 executes the computer program 53, the steps in the above-mentioned data-driven time-sharing power prediction method embodiments are implemented, for example Figure 1 Alternatively, when the processor 51 executes the computer program 53, the functions of each module in the above-mentioned device embodiments are realized, for example, Figure 4 The functions of the modules 41 to 45 are shown.

[0208] Exemplarily, the computer program 53 may be divided into one or more modules / units, one or more modules / units are stored in the memory 52 and executed by the processor 51 to complete the present application. One or more modules / units may be a series of computer program instruction segments that can complete specific functions, and the instruction segments are used to describe the execution process of the computer program 53 in the electronic device 50. For example, the computer program 53 may be divided into Figure 4 Modules 41 to 45 are shown.

[0209] The electronic device 50 may include, but is not limited to, a processor 51 and a memory 52. ​​Those skilled in the art will appreciate that Figure 5 It is only an example of the electronic device 50 and does not constitute a limitation of the electronic device 50. It may include more or fewer components than shown in the figure, or a combination of certain components, or different components. For example, the electronic device may also include input and output devices, network access devices, buses, etc.

[0210] The processor 51 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc.

[0211] The memory 52 may be an internal storage unit of the electronic device 50, such as a hard disk or memory of the electronic device 50. The memory 52 may also be an external storage device of the electronic device 50, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the electronic device 50. Further, the memory 52 may also include both an internal storage unit of the electronic device 50 and an external storage device. The memory 52 is used to store computer programs and other programs and data required by the electronic device. The memory 52 may also be used to temporarily store data that has been output or is to be output.

[0212] For the convenience and simplicity of description, only the division of the above functional modules / units is used as an example for illustration. In actual applications, the above functions can be assigned to different functional modules / units as needed. The above modules / units can be implemented in the form of hardware, software, or a combination of hardware and software.

[0213] The embodiment of the present invention further provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the methods in the above method embodiments are implemented.

[0214] The embodiment of the present invention further provides a computer program product, including a computer program. When the computer program is executed by a processor, the methods in the above method embodiments are implemented.

[0215] The computer program includes computer program code, which may be in source code form, object code form, executable file or some intermediate form, etc. Computer readable media may include: any entity or device capable of carrying computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal and software distribution medium, etc.

[0216] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments. If there is no special explanation and logical conflict, the terms and / or descriptions between different embodiments are consistent and can be referenced to each other. The technical features in different embodiments can be combined to form a new embodiment according to their internal logical relationship.

[0217] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A data-driven time-sharing electricity consumption prediction method, characterized in that: include: Obtain the first influencing factor data of the time-sharing electricity consumption on the day to be predicted; According to the first influencing factor data and a preset classification model, obtaining the time-sharing electricity category to which the day to be predicted belongs; Determine the time-sharing power characteristics and dominant influencing factors of the day to be predicted according to the time-sharing power category to which the day to be predicted belongs, and determine first data corresponding to the dominant influencing factors of the day to be predicted from the first influencing factor data; Determine a time-sharing electricity prediction model matching the day to be predicted based on the time-sharing electricity characteristics of the day to be predicted, the first data, and a preset association rule library; Based on the first data and the time-sharing electricity quantity prediction model, the time-sharing electricity quantity data of the day to be predicted is determined.

2. The data-driven time-sharing power forecasting method according to claim 1 is characterized in that: The determining, according to the time-sharing power characteristics of the day to be predicted, the first data and a preset association rule library, a time-sharing power prediction model matching the day to be predicted includes: According to the time-sharing electricity quantity characteristics of the day to be predicted and the first data, matching is performed in the association rule library to determine a matching association rule; Determine, according to the matching association rules, a time-sharing electricity quantity prediction model matched by the to-be-predicted day, and the first historical time-sharing electricity quantity data of the historical day matched by the to-be-predicted day; The determining, based on the first data and the time-sharing power forecasting model, the time-sharing power data of the to-be-forecasted day includes: The first data and the first historical time-sharing electricity data are input into the time-sharing electricity prediction model to obtain the time-sharing electricity data of the day to be predicted.

3. The data-driven time-sharing power forecasting method according to claim 1 is characterized in that: Before obtaining the time-sharing electricity category to which the predicted day belongs according to the first influencing factor data and the preset classification model, the method further includes: Obtain the second historical time-sharing electricity data and the second influencing factor data of different historical days; Clustering the second historical time-sharing power data according to the time-sharing power extreme value and the extreme value time in the second historical time-sharing power data to obtain multiple time-sharing power categories and time-sharing power characteristics of each time-sharing power category; Perform correlation analysis on the second historical time-of-use electricity data and the second influencing factor data of each time-of-use electricity category in the historical day to determine the dominant influencing factor of each time-of-use electricity category; Determine the second data of the dominant influencing factor of each time-sharing power category from the second influencing factor data of each time-sharing power category; A classification model is established according to the second data of each time-sharing power category.

4. The data-driven time-sharing electricity forecasting method according to claim 3 is characterized in that: Before determining the time-sharing power prediction model matching the day to be predicted according to the time-sharing power characteristics of the day to be predicted, the first data and a preset association rule library, the method further includes: Obtain the applicability of each prediction model for time-of-use electricity forecasting in each time-of-use electricity category; Establishing multiple project sets according to the time-sharing power characteristics corresponding to each time-sharing power category, the second data, and the applicability of each prediction model; Determine the support of each item set based on the frequency of occurrence of each item set; Construct a complex item pattern tree according to the item sets whose support is greater than or equal to the preset support threshold; Using the frequent item pattern tree, mining frequent item sets; Based on the frequent item sets, association rules between time-sharing power characteristics, dominant influencing factors and applicability of the prediction model are established to obtain the association rule base.

5. The data-driven time-sharing electricity forecasting method according to claim 4 is characterized in that: The method of establishing multiple project sets according to the time-sharing power characteristics corresponding to each time-sharing power category, the second data, and the applicability of each prediction model includes: The time-sharing power characteristics, the second data and the applicability of each prediction model corresponding to each time-sharing power category are clustered respectively to obtain multiple data categories under each time-sharing power category; Generalizing the multiple data categories to obtain generalized results of the time-sharing power characteristics, the second data, and the applicability of each prediction model under each time-sharing power category; A plurality of item sets are established according to the generalization results.

6. The data-driven time-sharing power forecasting method according to any one of claims 3 to 5, characterized in that: The clustering of the second historical time-sharing power data according to the time-sharing power extreme value and the extreme value time in the second historical time-sharing power data to obtain multiple time-sharing power categories and time-sharing power characteristics of each time-sharing power category includes: Randomly select a preset number of historical days as cluster centers; According to the time-sharing power extreme value and extreme value time in the second historical time-sharing power data of each historical day, and the time-sharing power extreme value and extreme value time in the second historical time-sharing power data corresponding to the cluster center, respectively calculate the distance from each historical day to each cluster center; Divide each historical day into the cluster where its corresponding target cluster center is located; wherein the target cluster center corresponding to the historical day is the cluster center corresponding to the minimum distance of the historical day; Calculate the time-sharing power average value of the second historical time-sharing power data of each historical day in each cluster, and determine a new cluster center based on the time-sharing power average value; complete an iterative calculation of the cluster center; Iterate continuously until the cluster center remains unchanged or the maximum number of iterations is reached, and according to the final cluster center, multiple time-sharing power categories and the time-sharing power characteristics of each time-sharing power category are determined.

7. The data-driven time-sharing electricity forecasting method according to claim 6 is characterized in that: The calculating the distance from each historical day to each cluster center respectively according to the time-sharing power extreme value and extreme value time in the second historical time-sharing power data of each historical day, and the time-sharing power extreme value and extreme value time in the second historical time-sharing power data corresponding to the cluster center, includes: According to the expression: Calculate the distance from each historical day to each cluster center respectively; Where, d sCr represents the distance from the historical day s to the cluster center Cr, t maxks represents the peak time of the kth hourly electricity peak in the second historical hourly electricity data of historical day s, t maxkCr represents the peak time of the kth time-sharing power peak in the second historical time-sharing power data of the cluster center Cr, l maxks represents the kth hourly electricity peak value in the second historical hourly electricity data of historical day s, l maxkCr represents the kth time-sharing power peak value in the second historical time-sharing power data of the cluster center Cr, t minjs Indicates the valley time of the jth hourly electricity valley value in the second historical hourly electricity data of historical day s, t minjCr represents the valley time of the jth time-sharing electricity valley value in the second historical time-sharing electricity data of the cluster center Cr, l minjs represents the jth hourly electricity valley value in the second historical hourly electricity data of historical day s, l minjCr represents the jth time-sharing electricity valley value in the second historical time-sharing electricity data of the cluster center Cr, m represents the total number of time-sharing electricity peak values ​​in the second historical time-sharing electricity data of the historical day s, and n represents the total number of time-sharing electricity valley values ​​in the second historical time-sharing electricity data of the historical day s.

8. A data-driven time-sharing power forecasting device, characterized in that: include: An acquisition module, used to acquire the first influencing factor data of the time-sharing electricity quantity on the day to be predicted; A classification module, used for obtaining the time-sharing electricity category to which the predicted day belongs according to the first influencing factor data and a preset classification model; A determination module, configured to determine the time-sharing power characteristics and dominant influencing factors of the day to be predicted according to the time-sharing power category to which the day to be predicted belongs, and determine first data corresponding to the dominant influencing factors of the day to be predicted from the first influencing factor data; A selection module, configured to determine a time-sharing electricity prediction model matching the day to be predicted based on the time-sharing electricity characteristics of the day to be predicted, the first data, and a preset association rule library; A prediction module is used to determine the time-sharing electricity data of the day to be predicted based on the first data and the time-sharing electricity prediction model.

9. An electronic device comprising a memory and a processor, wherein the memory is used to store a computer program, and the processor is used to call and run the computer program stored in the memory, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images